HBD854 - ON Semiconductor - Datasheet.Directory

http://onsemi.com

TVS/Zener Theory and Design Considerations

Handbook

HBD854/D

Rev. 0, Jun−2005

http://onsemi.com

Technical Information, Application Notes and Articles

Zener Diode Theory 3. . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Zener Diode Fabrication Techniques 8. . . . . . . . . . . . . . .

Reliability 12. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Zener Diode Characteristics 18. . . . . . . . . . . . . . . . . . . . .

Temperature Compensated Zeners 30. . . . . . . . . . . . . . .

Basic Voltage Regulation Using Zener Diodes 34. . . . .

Zener Protective Circuits and Techniques:

Basic Design Considerations 44. . . . . . . . . . . . . . . . .

Zener Voltage Sensing Circuits and Applications 54. . .

Miscellaneous Applications of

Zener Type Devices 61. . . . . . . . . . . . . . . . . . . . . . . . .

Transient Voltage Suppression 63. . . . . . . . . . . . . . . . . .

AN784 82. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

AN843 84. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Design Considerations and Performance of

Temperature Compensated Zener Diodes 97. . . . . .

MOSORBs 102. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

AR450 106. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Measurement of Zener Voltage to Thermal

Equilibrium with Pulsed Test Current 119. . . . . . . . . .

http://onsemi.com

ZENER DIODE THEORY

INTRODUCTION

The zener diode is a semiconductor device unique in its

mode of operation and completely unreplaceable by any

other electronic device. Because of its unusual properties it

fills a long-standing need in electronic circuitry. It provides,

among other useful functions, a constant voltage reference

or voltage control element available over a wide spectrum of

voltage and power levels.

The zener diode is unique among the semiconductor

family of devices because its electrical properties are

derived from a rectifying junction which operates in the

reverse breakdown region. In the sections that follow, the

reverse biased rectifying junction, some of the terms

associated with it, and properties derived from it will be

discussed fully.

The zener diode is fabricated from the element silicon.

Special techniques are applied in the fabrication of zener

diodes to create the required properties.

This manual was prepared to acquaint the engineer, the

equipment designer and manufacturer, and the experimenter

with the fundamental principles, design characteristics,

applications and advantages of this important

semiconductor device.

SEMICONDUCTOR THEORY

The active portion of a zener diode is a semiconductor PN

junction. PN junctions are formed in various kinds of

semiconductor devices by several techniques. Among these

are the widely used techniques known as alloying and

diffusion which are utilized in fabricating zener PN

junctions to provide excellent control over zener breakdown

voltage.

At the present time, zener diodes use silicon as the basic

material in the formation of their PN junction. Silicon is in

Group IV of the periodic table (tetravalent) and is classed as

a “semiconductor” due to the fact that it is a poor conductor

in a pure state. When controlled amounts of certain

“impurities” are added to a semiconductor it becomes a

better conductor of electricity. Depending on the type of

impurity added to the basic semiconductor, its conductivity

may take two different forms, called P- and N-type

respectively.

N-type conductivity in a semiconductor is much like the

conductivity due to the drift of free electrons in a metal. In

pure silicon at room temperature there are too few free

electrons to conduct current. However, there are ways of

introducing free electrons into the crystal lattice as we shall

now see. Silicon is a tetravalent element, one with four

valence electrons in the outer shell; all are virtually locked

into place by the covalent bonds of the crystal lattice

structure, as shown schematically in Figure 1a. When

controlled amounts of donor impurities (Group V elements)

such as phosphorus are added, the pentavalent phosphorus

atoms entering the lattice structure provide extra electrons

not required by the covalent bonds. These impurities are

called donor impurities since they “donate” a free electron

to the lattice. These donated electrons are free to drift from

negative to positive across the crystal when a field is applied,

as shown in Figure 1b. The “N” nomenclature for this kind

of conductivity implies “negative” charge carriers.

In P-type conductivity, the charges that carry electric

current across the crystal act as if they were positive charges.

We know that electricity is always carried by drifting

electrons in any material, and that there are no mobile

positively charged carriers in a solid. Positive charge

carriers can exist in gases and liquids in the form of positive

ions but not in solids. The positive character of the current

flow in the semiconductor crystal may be thought of as the

movement of vacancies (called holes) in the covalent lattice.

These holes drift from positive toward negative in an electric

field, behaving as if they were positive carriers.

P-type conductivity in semiconductors result from adding

acceptor impurities (Group III elements) such as boron to

silicon to the semiconductor crystal. In this case, boron

atoms, with three valence electrons, enter the tetravalent

silicon lattice. Since the covalent bonds cannot be satisfied

by only three electrons, each acceptor atom leaves a hole in

the lattice which is deficient by one electron. These holes

readily accept electrons introduced by external sources or

created by radiation or heat, as shown in Figure 1c. Hence

the name acceptor ion or acceptor impurity. When an

external circuit is connected, electrons from the current

source “fill up” these holes from the negative end and jump

from hole to hole across the crystal or one may think of this

process in a slightly different but equivalent way, that is as

the displacement of positive holes toward the negative

terminal. It is this drift of the positively charged holes which

accounts for the term P-type conductivity.

When semiconductor regions of N- and P-type

conductivities are formed in a semiconductor crystal

adjacent to each other, this structure is called a PN junction.

Such a junction is responsible for the action of both zener

diodes and rectifier devices, and will be discussed in the next

section.

http://onsemi.com

Figure 1. Semiconductor Structure

−

Si Si Si

Si Si

Si PSi

Si Si

Si BSi

Si Si

APPLIED FIELD

ELECTRONS ARE LOCKED

IN COVALENT BONDS

LOCKED COVALENT

BOND ELECTRONS

FREE ELECTRON

FROM PHOSPHOROUS

ATOM DRIFTS TOWARD

APPLIED POSITIVE POLE.

INCOMPLETED

COVALENT BOND

THIS ELECTRON JUMPS INTO

HOLE LEFT BY BORON ATOM.

HOLE POSITION IS DISPLACED

TO RIGHT. THIS RESULTS IN

A DRIFT OF HOLES TOWARD

THE NEGATIVE POLE, GIVING

THEM THE CHARACTER OF

MOBILE POSITIVE CHARGES.

(a) Lattice Structures of

Pure Silicon

(b) N-Type Silicon

−

THE SEMICONDUCTOR DIODE

In the forward-biased PN junction, Figure 2a, the P region

is made more positive than the N region by an external

circuit. Under these conditions there is a very low resistance

to current flow in the circuit. This is because the holes in the

positive P-type material are very readily attracted across the

junction interface toward the negative N-type side.

Conversely, electrons in the N-type are readily attracted by

the positive polarity in the other direction.

When a PN junction is reverse biased, the P-type side is

made more negative than the N-type side. (See Figure 2b.)

At voltages below the breakdown of the junction, there is

very little current flow across the junction interface. At first

thought one would expect no reverse current under reverse

bias conditions, but several effects are responsible for this

small current.

Under this condition the positive holes in the P-type

semiconductor are repelled from the junction interface by

the positive polarity applied to the N side, and conversely,

the electrons in the N material are repelled from the interface

by the negative polarity of the P side. This creates a region

extending from the junction interface into both P- and

N-type materials which is completely free of charge carriers,

that is, the region is depleted of its charge carriers. Hence,

this region is usually called the depletion region.

Although the region is free of charge carriers, the P-side

of the depletion region will have an excess negative charge

due to the presence of acceptor ions which are, of course,

fixed in the lattice; while the N-side of the depletion region

has an excess positive charge due to the presence of donor

ions. These opposing regions of charged ions create a strong

electric field across the PN junction responsible for the

creation of reverse current.

The semiconductor regions are never perfect; there are

always a few free electrons in P material and few holes in N

material. A more significant factor, however, is the fact that

great magnitudes of electron-hole pairs may be thermally

generated at room temperatures in the semiconductor. When

these electron-hole pairs are created within the depletion

region, then the intense electric field mentioned in the above

paragraph will cause a small current to flow. This small

current is called the reverse saturation current, and tends to

maintain a relatively constant value for a fixed temperature

at all voltages. The reverse saturation current is usually

negligible compared with the current flow when the junction

is forward biased. Hence, we see that the PN junction, when

not reverse biased beyond breakdown voltage, will conduct

heavily in only one direction. When this property is utilized

in a circuit we are employing the PN junction as a rectifier.

Let us see how we can employ its reverse breakdown

characteristics to an advantage.

As the reverse voltage is increased to a point called the

voltage breakdown point and beyond, current conduction

across the junction interface increases rapidly. The break

from a low value of the reverse saturation current to heavy

conductance is very sharp and well defined in most PN

junctions. It is called the zener knee. When reverse voltages

greater than the voltage breakdown point are applied to the

PN junction, the voltage drop across the PN junction

remains essentially constant at the value of the breakdown

voltage for a relatively wide range of currents. This region

http://onsemi.com

APPLIED FIELD

LARGE

CURRENT

SEVERAL VOLTS

APPLIED FIELD

VERY

SMALL

CURRENT

Figure 2. Effects of Junction Bias

CHARGES FROM BOTH P AND N REGIONS

DRIFT ACROSS JUNCTION AT VERY LOW

APPLIED VOLTAGES.

AT APPLIED VOLTAGES BELOW THE CRITICAL

BREAKDOWN LEVEL ONLY A FEW CHARGES

DRIFT ACROSS THE INTERFACE.

(a) Forward-Based PN

Junction

(b) Reverse-Biased PN

Junction

beyond the voltage breakdown point is called the zener

control region.

ZENER CONTROL REGION: VOLTAGE

BREAKDOWN MECHANISMS

Figure 3 depicts the extension of reverse biasing to the

point where voltage breakdown occurs. Although all PN

junctions exhibit a voltage breakdown, it is important to

know that there are two distinct voltage breakdown

mechanisms. One is called zener breakdown and the other is

called avalanche breakdown. In zener breakdown the value

of breakdown voltage decreases as the PN junction

temperature increases; while in avalanche breakdown the

value of the breakdown voltage increases as the PN junction

temperature increases. Typical diode breakdown

characteristics of each category are shown in Figure 4. The

factor determining which of the two breakdown

mechanisms occurs is the relative concentrations of the

impurities in the materials which comprise the junction. If

two different resistivity P-type materials are placed against

two separate but equally doped low-resistivity pieces of

N-type materials, the depletion region spread in the low

resistivity P-type material will be smaller than the depletion

region spread in the high resistivity P-type material.

Moreover, in both situations little of the resultant depletion

width lies in the N material if its resistivity is low compared

to the P-type material. In other words, the depletion region

always spreads principally into the material having the

highest resistivity. Also, the electric field (voltage per unit

Figure 3. Reverse Characteristic Extended

to Show Breakdown Effect

SLOPE

IREV

VBREAKDOWN

VREV

length) in the less resistive material is greater than the

electric field in the material of greater resistivity due to the

presence of more ions/unit volume in the less resistive

material. A junction that results in a narrow depletion region

will therefore develop a high field intensity and breakdown

by the zener mechanism. A junction that results in a wider

depletion region and, thus, a lower field intensity will break

down by the avalanche mechanism before a zener

breakdown condition can be reached.

The zener mechanism can be described qualitatively as

follows: because the depletion width is very small, the

application of low reverse bias (5 volts or less) will cause a

field across the depletion region on the order of 3 x 105V/cm.

A field of such high magnitude exerts a large force on the

http://onsemi.com

Figure 4. Typical Breakdown Diode Characteristics. Note Effects of Temperature for Each Mechanism

(A)

ZENER BREAKDOWN

OF A PN FUNCTION

VREV (VOLTS)

(B)

AVALANCHE BREAKDOWN OF A PN FUNCTION

IREV IREV

4321

VREV (VOLTS)

30 25 20 15 10 5

25°C65°C25°C65°C

valence electrons of a silicon atom, tending to separate them

from their respective nuclei. Actual rupture of the covalent

bonds occurs when the field approaches 3 x 105V/cm. Thus,

electron-hole pairs are generated in large numbers and a

sudden increase of current is observed. Although we speak

of a rupture of the atomic structure, it should be understood

that this generation of electron-hole pairs may be carried on

continuously as long as an external source supplies

additional electrons. If a limiting resistance in the circuit

external to the diode junction does not prevent the current

from increasing to high values, the device may be destroyed

due to overheating. The actual critical value of field causing

zener breakdown is believed to be approximately

3x10

5V/cm. On most commercially available silicon

diodes, the maximum value of voltage breakdown by the

zener mechanism is 8 volts. In order to fabricate devices

with higher voltage breakdown characteristics, materials

with higher resistivity, and consequently, wider depletion

regions are required. These wide depletion regions hold the

field strength down below the zener breakdown value

(3 x 105V/cm). Consequently, for devices with breakdown

voltage lower than 5 volts the zener mechanism

predominates, between 5 and 8 volts both zener and an

avalanche mechanism are involved, while above 8 volts the

avalanche mechanism alone takes over.

The decrease of zener breakdown voltage as junction

temperature increases can be explained in terms of the

energies of the valence electrons. An increase of temperature

increases the energies of the valence electrons. This weakens

the bonds holding the electrons and consequently, less applied

voltage is necessary to pull the valence electrons from their

position around the nuclei. Thus, the breakdown voltage

decreases as the temperature increases.

The dependence on temperature of the avalanche

breakdown mechanism is quite different. Here the depletion

region is of sufficient width that the carriers (electrons or

holes) can suffer collisions before traveling the region

completely i.e., the depletion region is wider than one

mean-free path (the average distance a carrier can travel

before combining with a carrier of opposite conductivity).

Therefore, when temperature is increased, the increased

lattice vibration shortens the distance a carrier travels before

colliding and thus requires a higher voltage to get it across

the depletion region.

As established earlier, the applied reverse bias causes a

small movement of intrinsic electrons from the P material to

the potentially positive N material and intrinsic holes from

the N material to the potentially negative P material (leakage

current). As the applied voltage becomes larger, these

electrons and holes increasingly accelerate. There are also

collisions between these intrinsic particles and bound

electrons as the intrinsic particles move through the

depletion region. If the applied voltage is such that the

intrinsic electrons do not have high velocity, then the

collisions take some energy from the intrinsic particles,

altering their velocity. If the applied voltage is increased,

collision with a valence electron will give considerable

energy to the electron and it will break free of its covalent

bond. Thus, one electron by collision, has created an

electron-hole pair. These secondary particles will also be

accelerated and participate in collisions which generate new

electron-hole pairs. This phenomenon is called carrier

multiplication. Electron-hole pairs are generated so quickly

and in such large numbers that there is an apparent avalanche

or self-sustained multiplication process (depicted

graphically in Figure 5). The junction is said to be in

breakdown and the current is limited only by resistance

external to the junction. Zener diodes above 7 to 8 volts

exhibit avalanche breakdown.

As junction temperature increases, the voltage breakdown

point for the avalanche mechanism increases. This effect can

be explained by considering the vibration displacement of

atoms in their lattice increases, and this increased

displacement corresponds to an increase in the probability

that intrinsic particles in the depletion region will collide

with the lattice atoms. If the probability of an intrinsic

particle-atom collision increases, then the probability that a

given intrinsic particle will obtain high momentum

http://onsemi.com

Figure 5. PN Junction in Avalanche Breakdown

WHEN THE APPLIED VOLTAGE IS

ABOVE THE BREAKDOWN POINT, A

FEW INJECTED ELECTRONS RECEIVE

ENOUGH ACCELERATION FROM THE

FIELD TO GENERATE NEW ELECTRONS

BY COLLISION. DURING THIS PROCESS

THE VOLTAGE DROP ACROSS THE

JUNCTION REMAINS CONSTANT.

RS ABSORBS EXCESS VOLTAGE.

LARGE CURRENT

CONSTANT VOLTAGE DROP

REVERSE-BIASED

PN JUNCTION IN AVALANCHE

decreases, and it follows that the low momentum intrinsic

particles are less likely to ionize the lattice atoms. Naturally,

increased voltage increases the acceleration of the intrinsic

particles, providing higher mean momentum and more

electron-hole pairs production. If the voltage is raised

sufficiently, the mean momentum becomes great enough to

create electron-hole pairs and carrier multiplication results.

Hence, for increasing temperature, the value of the

avalanche breakdown voltage increases.

VOLT-AMPERE CHARACTERISTICS

The zener volt-ampere characteristics for a typical 30 volt

zener diode is illustrated in Figure 6. It shows that the zener

diode conducts current in both directions; the forward

current IF being a function of forward voltage VF. Note that

IF is small until VF ≈ 0.65 V; then IF increases very rapidly.

For VF > 0.65 V IF is limited primarily by the circuit

resistance external to the diode.

ZZK

ZZT

30 20 10 0 0.5 1 1.5

0.5

1.5

REVERSE

CHARACTERISTIC

(VOLTS)

I (AMPS)

REVERSE CURRENT (AMPS)

Figure 6. Zener Diode Characteristics

IZT

IZM

1.40 A

IZK = 5 mA

FORWARD

CHARACTERISTIC TYPICAL

420 mA

The reverse current is a function of the reverse voltage VR

but for most prNO TAGactical purposes is zero until the

reverse voltage approaches VZ, the PN junction breakdown

voltage, at which time the reverse current increases very

rapidly. Since the reverse current is small for VR < VZ, but

great for VR > VZ each of the current regions is specified by

a different symbol. For the leakage current region, i.e.

non-conducting region, between 0 volts and VZ, the reverse

current is denoted by the symbol IR; but for the zener control

region, VR ≥ V

Z, the reverse current is denoted by the

symbol IZ. IR is usually specified at a reverse voltage

VR≈0.8 VZ.

The PN junction breakdown voltage, VZ, is usually called

the zener voltage, regardless whether the diode is of the

zener or avalanche breakdown type. Commercial zener

diodes are available with zener voltages from about

1.8 V −400 V. For most applications the zener diode is

operated well into the breakdown region (IZT to IZM). Most

manufacturers give an additional specification of IZK

(= 5 mA in Figure 6) to indicate a minimum operating

current to assure reasonable regulation.

This minimum current IZK varies in the various types of

zener diodes and, consequently, is given on the data sheets.

The maximum zener current IZM should be considered the

maximum reverse current recommended by the

manufacturer. Values of IZM are usually given in the data

sheets.

Between the limits of IZK and IZM, which are 5 mA and

1400 mA (1.4 Amps) in the example of Figure 6, the voltage

across the diode is essentially constant, and ≈ V

Z. This

plateau region has, however, a large positive slope such that

the precise value of reverse voltage will change slightly as

a function of IZ. For any point on this plateau region one may

calculate an impedance using the incremental magnitudes of

the voltage and current. This impedance is usually called the

zener impedance ZZ, and is specified for most zener diodes.

Most manufacturers measure the maximum zener

impedance at two test points on the plateau region. The first

is usually near the knee of the zener plateau, ZZK, and the

latter point near the midrange of the usable zener current

excursion. Two such points are illustrated in Figure 6.

This section was intended to introduce the reader to a few

of the major terms used with zener diodes. A complete

description of these terms may be found in chapter four. In

chapter four a full discussion of zener leakage, DC

breakdown, zener impedance, temperature coefficients and

many other topics may be found.

http://onsemi.com

ZENER DIODE FABRICATION TECHNIQUES

INTRODUCTION

A brief exposure to the techniques used in the fabrication

of zener diodes can provide the engineer with additional

insight using zeners in their applications. That is, an

understanding of zener fabrication makes the capabilities

and limitations of the zener diode more meaningful. This

chapter discusses the basic steps in the fabrication of the

zener from crystal growing through final testing.

ZENER DIODE WAFER FABRICATION

The major steps in the manufacture of zeners are provided

in the process flow in Figure 1. It is important to point out

that the manufacturing steps vary somewhat from

manufacturer to manufacturer, and also vary with the type of

zener diode produced. This is driven by the type of package

required as well as the electrical characteristics desired. For

example, alloy diffused devices provide excellent low

voltage reference with low leakage characteristics but do not

have the same surge carrying capability as diffused diodes.

The manufacturing process begins with the growing of high

quality silicon crystals.

Crystals for ON Semiconductor zener diodes are grown

using the Czochralski technique, a widely used process

which begins with ultra-pure polycrystalline silicon. The

polycrystalline silicon is first melted in a nonreactive

crucible held at a temperature just above the melting point.

A carefully controlled quantity of the desired dopant

impurity, such as phosphorus or boron is added. A high

quality seed crystal of the desired crystalline orientation is

then lowered into the melt while rotating. A portion of this

seed crystal is allowed to melt into the molten silicon. The

seed is then slowly pulled and continues to rotate as it is

raised from the melt. As the seed is raised, cooling takes

place and material from the melt adheres to it, thus forming

a single crystal ingot. With this technique, ingots with

diameters of several inches can be fabricated.

Figure 1. General Flow of the Zener Diode Process

SILICON CRYSTAL

GROWING

WAFER

PREPARATION

OXIDE

PASSIVATION

WAFER THINNING ANODE

METALLIZATION

JUNCTION

FORMATION

CATHODE

METALLIZATION

WAFER

TESTING WAFER DICING

TEST LEAD

FINISH ASSEMBLY

MARK TEST PACKAGE

SHIP

http://onsemi.com

Figure 2. Basic Fabrication Steps in the Silicon Planar Process: a) oxide formation, b) selective oxide removal,

c) deposition of dopant atoms, d) junction formation by diffusion of dopant atoms.

SILICON DIOXIDE GROWTH

SiO2

(a)

SILICON DIOXIDE SELECTIVELY REMOVED

(b)

(c)

DOPANT ATOMS DEPOSITED ONTO

THE EXPOSED SILICON

(d)

DOPANT ATOMS DIFFUSE INTO SILICON

BUT NOT APPRECIABLY INTO THE SILICON DIOXIDE

Once the single-crystal silicon ingot is grown, it is tested

for doping concentration (resistivity), undesired impurity

levels, and minority carrier lifetime. The ingot is then sliced

into thin, circular wafers. The wafers are then chemically

etched to remove saw damage and polished in a sequence of

successively finer polishing grits until a mirror-like defect

free surface is obtained. The wafers are then cleaned and

placed in vacuum sealed wafer carriers to prevent any

contamination from getting on them. At this point, the

wafers are ready to begin device fabrication.

Zener diodes can be manufactured using different

processing techniques such as planar processing or mesa

etched processing. The majority of ON Semiconductor

zener diodes are manufactured using the planar technique as

shown in Figure 2.

The planar process begins by growing an ultra-clean

protective silicon dioxide passivation layer. The oxide is

typically grown in the temperature range of 900 to 1200

degrees celcius. Once the protective layer of silicon dioxide

has been formed, it must be selectively removed from those

areas into which dopant atoms will be introduced. This is

done using photolithographic techniques.

First a light sensitive solution called photo resist is spun

onto the wafer. The resist is then dried and a photographic

negative or mask is placed over the wafer. The resist is then

exposed to ultraviolet light causing the molecules in it to

cross link or polymerize becoming very rigid. Those areas

of the wafer that are protected by opaque portions of the

mask are not exposed and are developed away. The oxide is

then etched forming the exposed regions in which the dopant

will be introduced. The remaining resist is then removed and

the wafers carefully cleaned for the doping steps.

Dopant is then introduced onto the wafer surface using

various techniques such as aluminum alloy for low voltage

devices, ion-implantation, spin-on dopants, or chemical

vapor deposition. Once the dopant is deposited, the

junctions are formed in a subsequent high temperature (1100

to 1250 degrees celcius are typical) drive-in. The resultant

junction profile is determined by the background

concentration of the starting substrate, the amount of dopant

placed at the surface, and amount of time and temperature

used during the dopant drive-in. This junction profile

determines the electrical characteristics of the device.

During the drive-in cycle, additional passivation oxide is

grown providing additional protection for the devices.

After junction formation, the wafers are then processed

through what is called a getter process. The getter step

utilizes high temperature and slight stress provided by a

highly doped phosphosilicate glass layer introduced into the

backside of the wafers. This causes any contaminants in the

area of the junction to diffuse away from the region. This

serves to improve the reverse leakage characteristic and the

stability of the device. Following the getter process, a second

photo resist step opens the contact area in which the anode

metallization is deposited.

Metal systems for ON Semiconductor’s zener diodes are

determined by the requirements of the package. The metal

systems are deposited in ultra-clean vacuum chambers

utilizing electron-beam evaporation techniques. Once the

metal is deposited, photo resist processing is utilized to form

the desired patterns. The wafers are then lapped to their final

thickness and the cathode metallization deposited using the

same e-beam process.

http://onsemi.com

The quality of the wafers is closely monitored throughout

the process by using statistical process control techniques

and careful microscopic inspections at critical steps. Special

wafer handling equipment is used throughout the

manufacturing process to minimize contamination and to

avoid damaging the wafers in any way. This further

enhances the quality and stability of the devices.

Upon completion of the fabrication steps, the wafers are

electrically probed, inspected, and packaged for shipment to

the assembly operations. All ON Semiconductor zener

diode product is sawn using 100% saw-through techniques

stringently developed to provide high quality silicon die.

ZENER DIODE ASSEMBLY

Surmetic 30, 40 and MOSORB

The plastic packages (Surmetic 30, 40 and MOSORBs)

are assembled using oxygen free high conductivity copper

leads for efficient heat transfer from the die and allowing

maximum power dissipation with a minimum of external

heatsinking. Figure 3 shows typical assembly. The leads are

of nail head construction, soldered directly to the die, which

further enhances the heat dissipating capabilities of the

package.

The Surmetic 30s, 40s and MOSORBs are basically

assembled in the same manner; the only difference being the

MOSORBs are soldered together using a solder disc

between the lead and die whereas the Surmetic 30s and

Surmetic 40s utilize pre-soldered leads.

Assembly is started on the Surmetic 30 and 40 by loading

the leads into assembly boats and pre-soldering the nail

heads. After pre-soldering, one die is then placed into each

cavity of one assembly boat and another assembly boat is

then mated to it. Since the MOSORBs do not use

pre-soldered leads, the leads are put into the assembly boat,

a solder disc is placed into each cavity and then a die is put

in on top. A solder disc is put in on top of the die. Another

assembly boat containing only leads is mated to the boat

containing the leads, die, and two solder discs. The boats are

passed through the assembly furnace; this operation requires

only one pass through the furnace.

After assembly, the leads on the Surmetic 30s, 40s and

MOSORBs are plated with a tin-lead alloy making them

readily solderable and corrosion resistant.

Double Slug (DO-35 and DO-41)

Double slugs receive their name from the dumet slugs, one

attached to one end of each lead. These slugs sandwich the

pre-tinned die between them and are hermetically sealed to

the glass envelope or body during assembly. Figure 4 shows

typical assembly.

The assembly begins with the copper clad steel leads

being loaded into assembly “boats.” Every other boat load

of leads has a glass body set over the slug. A pre-tinned die

is placed into each glass body and the other boat load of leads

is mated to the boat holding the leads, body and die. These

mated boats are then placed into the assembly furnace where

the total mass is heated. Each glass body melts; and as the

boat proceeds through the cooling portion of the furnace

chamber, the tin which has wetted to each slug solidifies

forming a bond between the die and both slugs. The glass

hardens, attaching itself to the sides of the two slugs forming

the hermetic seal. The above illustrates how the diodes are

completely assembled using a single furnace pass

minimizing assembly problems.

The encapsulated devices are then processed through lead

finish. This consists of dipping the leads in molten tin/lead

solder alloy. The solder dipped leads produce an external

finish which is tarnish-resistant and very solderable.

Figure 3. Double-Slug Plastic

Zener Construction

Figure 4. Double Slug Glass

Zener Construction

OFHC COPPER LEAD,

SOLDER PLATED

PLASTIC

(THERMO SET)

ENCAPSULATED

NAILHEAD LEAD

ZENER DIE Sn Pb

OFHC COPPER LEAD,

SOLDER PLATED

LEAD, STEEL, CU CLAD

SOLDER DIPPED

SLUG DUMET

GLASS SLEEVE

PASSIVATED

ZENER DIE

NAILHEAD LEAD

http://onsemi.com

ZENER DIODE TEST, MARK AND PACKAGING

Double Slug, Surmetic 30, 40 and MOSORB

After lead finish, all products are final tested, whether

they are double slug or of Surmetic construction, all are 100

percent final tested for zener voltage, leakage current,

impedance and forward voltage drop.

Process average testing is used which is based upon the

averages of the previous lots for a given voltage line and

package type. Histograms are generated for the various

parameters as the units are being tested to ensure that the lot

is testing well to the process average and compared against

other lots of the same voltage.

After testing, the units are marked as required by the

specification. The markers are equipped to polarity orient

the devices as well as perform 100% redundant test prior to

packaging.

After marking, the units are packaged either in “bulk”

form or taped and reeled or taped and ammo packed to

accommodate automatic insertion.

http://onsemi.com

RELIABILITY

INTRODUCTION

ON Semiconductor’s Quality System maintains

“continuous product improvement” goals in all phases of the

operation. Statistical process control (SPC), quality control

sampling, reliability audits and accelerated stress testing

techniques monitor the quality and reliability of its products.

Management and engineering skills are continuously

upgraded through training programs. This maintains a

unified focus on Six Sigma quality and reliability from the

inception of the product to final customer use.

STATISTICAL PROCESS CONTROL

ON Semiconductor’s Discrete Group is continually

pursuing new ways to improve product quality. Initial design

improvement is one method that can be used to produce a

superior product. Equally important to outgoing product

quality is the ability to produce product that consistently

conforms to specification. Process variability is the basic

enemy of semiconductor manufacturing since it leads to

product variability. Used in all phases of

ON Semiconductor’s product manufacturing,

STATISTICAL PROCESS CONTROL (SPC) replaces

variability with predictability. The traditional philosophy in

the semiconductor industry has been adherence to the data

sheet specification. Using SPC methods assures the product

will meet specific process requirements throughout the

manufacturing cycle. The emphasis is on defect prevention,

not detection. Predictability through SPC methods requires

the manufacturing culture to focus on constant and

permanent improvements. Usually these improvements

cannot be bought with state-of-the-art equipment or

automated factories. With quality in design, process and

material selection, coupled with manufacturing

predictability, ON Semiconductor can produce world class

products.

The immediate effect of SPC manufacturing is

predictability through process controls. Product centered and

distributed well within the product specification benefits ON

Semiconductor with fewer rejects, improved yields and lower

cost. The direct benefit to ON Semiconductor’s customers

includes better incoming quality levels, less inspection time

and ship-to-stock capability. Circuit performance is often

dependent on the cumulative effect of component variability.

Tightly controlled component distributions give the customer

greater circuit predictability. Many customers are also

converting to just-in-time (JIT) delivery programs. These

programs require improvements in cycle time and yield

predictability achievable only through SPC techniques. The

benefit derived from SPC helps the manufacturer meet the

customer’s expectations of higher quality and lower cost

product.

Ultimately, ON Semiconductor will have Six Sigma

capability on all products. This means parametric

distributions will be centered within the specification limits

with a product distribution of plus or minus Six Sigma about

mean. Six Sigma capability, shown graphically in Figure 1,

details the benefit in terms of yield and outgoing quality

levels. This compares a centered distribution versus a 1.5

sigma worst case distribution shift.

New product development at ON Semiconductor requires

more robust design features that make them less sensitive to

minor variations in processing. These features make the

implementation of SPC much easier.

A complete commitment to SPC is present throughout

ON Semiconductor. All managers, engineers, production

operators, supervisors and maintenance personnel have

received multiple training courses on SPC techniques.

Manufacturing has identified 22 wafer processing and 8

assembly steps considered critical to the processing of zener

products. Processes, controlled by SPC methods, that have

shown significant improvement are in the diffusion,

photolithography and metallization areas.

To better understand SPC principles, brief explanations

have been provided. These cover process capability,

implementation and use.

Figure 1. AOQL and Yield from a Normal Distribution of

Product With 6σ Capability

Standard Deviations From Mean

Distribution Centered Distribution Shifted ± 1.5

At ±3σ2700 ppm defective

99.73% yield

At ±4σ63 ppm defective

99.9937% yield

At ±5σ0.57 ppm defective

99.999943% yield

At ±6σ0.002 ppm defective

99.9999998% yield

66810 ppm defective

93.32% yield

6210 ppm defective

99.379% yield

233 ppm defective

99.9767% yield

3.4 ppm defective

99.99966% yield

-6σ-5s -4σ-3σ-2σ-1σ0 1σ2σ3σ4σ5σ6σ

PROCESS CAPABILITY

One goal of SPC is to ensure a process is CAPABLE.

Process capability is the measurement of a process to

produce products consistently to specification

requirements. The purpose of a process capability study is

to separate the inherent RANDOM VARIABILITY from

ASSIGNABLE CAUSES. Once completed, steps are taken

to identify and eliminate the most significant assignable

causes. Random variability is generally present in the

system and does not fluctuate. Sometimes, these are

considered basic limitations associated with the machinery,

materials, personnel skills or manufacturing methods.

Assignable cause inconsistencies relate to time variations in

yield, performance or reliability.

http://onsemi.com

Figure 2. Impact of Assignable Causes on Process Predictable

Figure 3. Difference Between Process Control and Process Capability

Process “under control” − all assignable causes are

removed and future distribution is predictable.

PREDICTION

TIME

SIZE

TIME

PREDICTION

SIZE

TIME

Out of control

(assignable causes present)

In control assignable

causes eliminated

SIZE

TIME

In control but not capable

(variation from random variability

excessive)

Lower

Specification Limit

Upper

Specification Limit

In control and capable

(variation from random

variability reduced)

Traditionally, assignable causes appear to be random due

to the lack of close examination or analysis. Figure 2 shows

the impact on predictability that assignable cause can have.

Figure 3 shows the difference between process control and

process capability.

A process capability study involves taking periodic

samples from the process under controlled conditions. The

performance characteristics of these samples are charted

against time. In time, assignable causes can be identified and

engineered out. Careful documentation of the process is key

to accurate diagnosis and successful removal of the

assignable causes. Sometimes, the assignable causes will

remain unclear requiring prolonged experimentation.

Elements which measure process variation control and

capability are Cp and Cpk respectively. Cp is the

specification width divided by the process width or Cp =

(specification width) / 6σ. Cpk is the absolute value of the

closest specification value to the mean, minus the mean,

divided by half the process width or Cpk = | closest

specification — X / 3σ.

At ON Semiconductor, for critical parameters, the process

capability is acceptable with a Cpk = 1.33. The desired

process capability is a Cpk = 2 and the ideal is a Cpk = 5.

Cpk, by definition, shows where the current production

process fits with relationship to the specification limits. Off

center distributions or excessive process variability will

result in less than optimum conditions.

SPC IMPLEMENTATION AND USE

The Discrete Group uses many parameters that show

conformance to specification. Some parameters are

sensitive to process variations while others remain constant

for a given product line. Often, specific parameters are

influenced when changes to other parameters occur. It is

both impractical and unnecessary to monitor all parameters

using SPC methods. Only critical parameters that are

sensitive to process variability are chosen for SPC

monitoring. The process steps affecting these critical

parameters must be identified also. It is equally important to

find a measurement in these process steps that correlates

with product performance. This is called a critical process

parameter.

Once the critical process parameters are selected, a sample

plan must be determined. The samples used for

measurement are organized into RATIONAL

SUBGROUPS of approximately 2 to 5 pieces. The

subgroup size should be such that variation among the

samples within the subgroup remain small. All samples must

come from the same source e.g., the same mold press

operator, etc.. Subgroup data should be collected at

appropriate time intervals to detect variations in the process.

As the process begins to show improved stability, the

interval may be increased. The data collected must be

carefully documented and maintained for later correlation.

Examples of common documentation entries would include

operator, machine, time, settings, product type, etc..

http://onsemi.com

Figure 4. Example of Process Control Chart Showing Oven Temperature Data

147

148

149

150

151

152

153

154

UCL = 152.8

= 150.4

LCL = 148.0

UCL = 7.3

= 3.2

LCL = 0

Once the plan is established, data collection may begin.

The data collected will generate X and R values that are

plotted with respect to time. X refers to the mean of the

values within a given subgroup, while R is the range or

greatest value minus least value. When approximately 20 or

more X and R values have been generated, the average of

these values is computed as follows:

X = (X + X2 + X3 + ...)/K

R = (R1 + R2 + R3 + ...)/K

where K = the number of subgroups measured.

The values of X and R are used to create the process

control chart. Control charts are the primary SPC tool used

to signal a problem. Shown in Figure 4, process control

charts show X and R values with respect to time and

concerning reference to upper and lower control limit

values. Control limits are computed as follows:

R upper control limit = UCLR = D4 R

R lower control limit LCLR = D3 R

X upper control limit = UCLX = X + A2 R

X lower control limit = LCLX = X − A

Where D4, D3 and A2 are constants varying by sample

size, with values for sample sizes from 2 to 10 shown in

the following partial table:

Control charts are used to monitor the variability of

critical process parameters. The R chart shows basic

problems with piece to piece variability related to the

process. The X chart can often identify changes in people,

machines, methods, etc. The source of the variability can be

difficult to find and may require experimental design

techniques to identify assignable causes.

Some general rules have been established to help

determine when a process is OUT-OF-CONTROL. Figure

5a shows a control chart subdivided into zones A, B, and C

corresponding to 3 sigma, 2 sigma, and 1 sigma limits

respectively. In Figure 5b through Figure 5e four of the tests

that can be used to identify excessive variability and the

presence of assignable causes are shown. As familiarity with

a given process increases, more subtle tests may be

employed successfully.

Once the variability is identified, the cause of the

variability must be determined. Normally, only a few factors

have a significant impact on the total variability of the

process. The importance of correctly identifying these

factors is stressed in the following example. Suppose a

process variability depends on the variance of five factors A,

B, C, D and E. Each has a variance of 5, 3, 2, 1 and 0.4

respectively.

Since:

σ tot = σA2 + σB2 + σC2 + σD2 + σE2

σ tot = 52 + 32 + 22 + 12 + (0.4)2 = 6.3

n2345678910

D43.27 2.57 2.28 2.11 2.00 1.92 1.86 1.82 1.78

D3* * * * * 0.08 0.14 0.18 0.22

A21.88 1.02 0.73 0.58 0.48 0.42 0.37 0.34 0.31

* For sample sizes below 7, the LCLR would technically be a negative number; in those cases there is no lower control limit;

this means that for a subgroup size 6, six “identical” measurements would not be unreasonable.

http://onsemi.com

Now if only D is identified and eliminated then;

s tot = 52 + 32 + 22 + (0.4)2= 6.2

This results in less than 2% total variability improvement.

If B, C and D were eliminated, then;

σ tot = 52 + (0.4)2= 5.02

This gives a considerably better improvement of 23%. If

only A is identified and reduced from 5 to 2, then;

σ tot = 22 + 32 + 22 + 12 + (0.4)2 = 4.3

Identifying and improving the variability from 5 to 2 gives

us a total variability improvement of nearly 40%.

Most techniques may be employed to identify the primary

assignable cause(s). Out-of-control conditions may be

correlated to documented process changes. The product may

be analyzed in detail using best versus worst part

comparisons or Product Analysis Lab equipment.

Multi-variance analysis can be used to determine the family

of variation (positional, critical or temporal). Lastly,

experiments may be run to test theoretical or factorial

analysis. Whatever method is used, assignable causes must

be identified and eliminated in the most expeditious manner

possible.

After assignable causes have been eliminated, new

control limits are calculated to provide a more challenging

variability criteria for the process. As yields and variability

improve, it may become more difficult to detect

improvements because they become much smaller. When all

assignable causes have been eliminated and the points

remain within control limits for 25 groups, the process is

said to be in a state of control.

SUMMARY

ON Semiconductor is committed to the use of

STATISTICAL PROCESS CONTROLS. These principles,

used throughout manufacturing, have already resulted in

many significant improvements to the processes. Continued

dedication to the SPC culture will allow ON Semiconductor

to reach the Six Sigma and zero defect capability goals. SPC

will further enhance the commitment to TOTAL

CUSTOMER SATISFACTION.

UCL

LCL

UCL

LCL

CENTERLINE

ZONE A (+ 3 SIGMA)

ZONE B (+ 2 SIGMA)

ZONE C (+ 1 SIGMA)

ZONE C (−1 SIGMA)

ZONE B (−2 SIGMA)

ZONE A (−3 SIGMA)

Figure 5a. Control Chart Zones Figure 5b. One Point Outside Control Limit

Indicating Excessive Variability

Figure 5c. Two Out of Three Points in Zone A or

Beyond Indicating Excessive Variability

Figure 5d. Four Out of Five Points in Zone B or

Beyond Indicating Excessive Variability

Figure 5e. Seven Out of Eight Points in Zone C or

Beyond Indicating Excessive Variability

http://onsemi.com

RELIABILITY STRESS TESTS

The following gives brief descriptions of the reliability

tests commonly used in the reliability monitoring program.

Not all of the tests listed are performed on each product.

Other tests may be performed when appropriate. In addition

some form of preconditioning may be used in conjunction

with the following tests.

AUTOCLAVE (aka, PRESSURE COOKER)

Autoclave is an environmental test which measures

device resistance to moisture penetration and the resultant

effects of galvanic corrosion. Autoclave is a highly

accelerated and destructive test.

Typical Test Conditions: TA = 121°C, rh = 100%, p =

1 atmosphere (15 psig), t = 24 to 96 hours

Common Failure Modes: Parametric shifts, high

leakage and/or catastrophic

Common Failure Mechanisms: Die corrosion or

contaminants such as foreign material on or within the

package materials. Poor package sealing

HIGH HUMIDITY HIGH TEMPERATURE BIAS

(H3TB or H3TRB)

This is an environmental test designed to measure the

moisture resistance of plastic encapsulated devices. A bias

is applied to create an electrolytic cell necessary to

accelerate corrosion of the die metallization. With time, this

is a catastrophically destructive test.

Typical Test Conditions: TA = 85°C to 95°C, rh = 85%

to 95%, Bias = 80% to 100% of Data Book max. rating,

t = 96 to 1750 hours

Common Failure Modes: Parametric shifts, high

leakage and/or catastrophic

Common Failure Mechanisms: Die corrosion or

contaminants such as foreign material on or within the

package materials. Poor package sealing

Military Reference: MIL-STD-750, Method 1042

HIGH TEMPERATURE REVERSE BIAS (HTRB)

The purpose of this test is to align mobile ions by means

of temperature and voltage stress to form a high-current

leakage path between two or more junctions.

Typical Test Conditions: TA = 85°C to 150°C, Bias =

80% to 100% of Data Book max. rating, t = 120 to 1000

hours

Common Failure Modes: Parametric shifts in leakage

Common Failure Mechanisms: Ionic contamination on

the surface or under the metallization of the die

Military Reference: MIL-STD-750, Method 1039

HIGH TEMPERATURE STORAGE LIFE (HTSL)

High temperature storage life testing is performed to

accelerate failure mechanisms which are thermally

activated through the application of extreme temperatures.

Typical Test Conditions: TA = 70°C to 200°C, no bias,

t = 24 to 2500 hours

Common Failure Modes: Parametric shifts in leakage

Common Failure Mechanisms: Bulk die and diffusion

defects

Military Reference: MIL-STD-750, Method 1032

INTERMITTENT OPERATING LIFE (IOL)

The purpose of this test is the same as SSOL in addition

to checking the integrity of both wire and die bonds by

means of thermal stressing.

Typical Test Conditions: T

A = 25°C, Pd = Data Book

maximum rating, Ton = Toff = Δof 50°C to 100°C, t =

42 to 30000 cycles

Common Failure Modes: Parametric shifts and

catastrophic

Common Failure Mechanisms: Foreign material, crack

and bulk die defects, metallization, wire and die bond

defects

Military Reference: MIL-STD-750, Method 1037

MECHANICAL SHOCK

This test is used to determine the ability of the device to

withstand a sudden change in mechanical stress due to

abrupt changes in motion as seen in handling, transportation,

or actual use.

Typical Test Conditions: Acceleration = 1500 g’s,

Orientation = X1, Y1, Y2 plane, t = 0.5 msec, Blows = 5

Common Failure Modes: Open, short, excessive

leakage, mechanical failure

Common Failure Mechanisms: Die and wire bonds,

cracked die, package defects

Military Reference: MIL-STD-750, Method 2015

MOISTURE RESISTANCE

The purpose of this test is to evaluate the moisture

resistance of components under temperature/humidity

conditions typical of tropical environments.

Typical Test Conditions: TA = −10°C to 65°C, rh = 80%

to 98%, t = 24 hours/cycle, cycle = 10

Common Failure Modes: Parametric shifts in leakage

and mechanical failure

Common Failure Mechanisms: Corrosion or

contaminants on or within the package materials. Poor

package sealing

Military Reference: MIL-STD-750, Method 1021

http://onsemi.com

SOLDERABILITY

The purpose of this test is to measure the ability of device

leads/terminals to be soldered after an extended period of

storage (shelf life).

Typical Test Conditions: Steam aging = 8 hours, Flux

= R, Solder = Sn60, Sn63

Common Failure Modes: Pin holes, dewetting,

non-wetting

Common Failure Mechanisms: Poor plating,

contaminated leads

Military Reference: MIL-STD-750, Method 2026

SOLDER HEAT

This test is used to measure the ability of a device to

withstand the temperatures as may be seen in wave soldering

operations. Electrical testing is the endpoint criterion for this

stress.

Typical Test Conditions: Solder Temperature = 260°C,

t = 10 seconds

Common Failure Modes: Parameter shifts, mechanical

failure

Common Failure Mechanisms: Poor package design

Military Reference: MIL-STD-750, Method 2031

STEADY STATE OPERATING LIFE (SSOL)

The purpose of this test is to evaluate the bulk stability of

the die and to generate defects resulting from manufacturing

aberrations that are manifested as time and stress-dependent

failures.

Typical Test Conditions: T

A = 25°C, PD = Data Book

maximum rating, t = 16 to 1000 hours

Common Failure Modes: Parametric shifts and

catastrophic

Common Failure Mechanisms: Foreign material, crack

die, bulk die, metallization, wire and die bond defects

Military Reference: MIL-STD-750, Method 1026

TEMPERATURE CYCLING (AIR TO AIR)

The purpose of this test is to evaluate the ability of the

device to withstand both exposure to extreme temperatures

and transitions between temperature extremes. This testing

will also expose excessive thermal mismatch between

materials.

Typical Test Conditions: TA = −65°C to 200°C, cycle

= 10 to 1000

Common Failure Modes: Parametric shifts and

catastrophic

Common Failure Mechanisms: Wire bond, cracked or

lifted die and package failure

Military Reference: MIL-STD-750, Method 1051

THERMAL SHOCK (LIQUID TO LIQUID)

The purpose of this test is to evaluate the ability of the

device to withstand both exposure to extreme temperatures

and sudden transitions between temperature extremes. This

testing will also expose excessive thermal mismatch

between materials.

Typical Test Conditions: T

A = 0°C to 100°C, cycles

= 10 to 1000

Common Failure Modes: Parametric shifts and

catastrophic

Common Failure Mechanisms: Wire bond, cracked or

lifted die and package failure

Military Reference: MIL-STD-750, Method 1056

VARIABLE FREQUENCY VIBRATION

This test is used to examine the ability of the device to

withstand deterioration due to mechanical resonance.

Typical Test Conditions: Peak acceleration = 20 g’s,

Frequency range = 20 Hz to 20 kHz, t = 48 minutes.

Common Failure Modes: Open, short, excessive

leakage, mechanical failure

Common Failure Mechanisms: Die and wire bonds,

cracked die, package defects

Military Reference: MIL-STD-750, Method 2056

http://onsemi.com

ZENER DIODE CHARACTERISTICS

INTRODUCTION

At first glance the zener diode is a simple device

consisting of one P-N junction with controlled breakdown

voltage properties. However, when considerations are given

to the variations of temperature coefficient, zener

impedance, thermal time response, and capacitance, all of

which are a function of the breakdown voltage (from 1.8 to

400 V), a much more complicated picture arises. In addition

to the voltage spectrum, a variety of power packages are on

the market with a variation of dice area inside the

encapsulation.

This chapter is devoted to sorting out the important

considerations in a “typical” fashion. For exact details, the

data sheets must be consulted. However, much of the

information contained herein is supplemental to the data

sheet curves and will broaden your understanding of zener

diode behavior.

Specifically, the following main subjects will be detailed:

Basic DC Volt-Ampere Characteristics

Impedance versus Voltage and Current

Temperature Coefficient versus Voltage and Current

Power Derating

Mounting

Thermal Time Response − Effective Thermal Impedance

Surge Capabilities

Frequency Response − Capacitance and Switching

Effects

BASIC ZENER DIODE DC VOLT-AMPERE

CHARACTERISTICS

Reverse and forward volt-ampere curves are represented

in Figure 1 for a typical zener diode. The three areas −

forward, leakage, and breakdown − will each be examined.

FORWARD DC CHARACTERISTICS

The forward characteristics of a zener diode are

essentially identical with an “ordinary” rectifier and is

shown in Figure 2. The volt-ampere curve follows the basic

diode equation of IF = IReqV/KT where KT/q equals about

0.026 volts at room temperature and IR (reverse leakage

current) is dependent upon the doping levels of the P-N

junction as well as the area. The actual plot of VF versus IF

deviates from the theoretical due to slightly “fixed” series

resistance of the lead wire, bonding contacts and some bulk

effects in the silicon.

Figure 1. Typical Zener Diode DC V-I Characteristics

(Not to Scale)

FORWARD VOLTAGE

REVERSE VOLTAGE

BREAKDOWN

REGION

LEAKAGE REGION

FORWARD CHARACTERISTIC

FORWARD CURRENT

REVERSE CURRENT

While the common form of the diode equation suggests

that IR is constant, in fact IR is itself strongly temperature

dependent. The rapid increase in IR with increasing

temperature dominates the decrease contributed by the

exponential term in the diode equation. As a result, the

forward current increases with increasing temperature.

Figure 2 shows a forward characteristic temperature

dependence for a typical zener. These curves indicate that

for a constant current, an increase in temperature causes a

decrease in forward voltage. The voltage temperature

coefficient values are typically in the range of −1.4 to

−2 mV/°C.

Figure 2. Typical Forward Characteristics of

Zener Diodes

0.90.80.70.60.50.40.3

1000

VF, FORWARD VOLTAGE (VOLTS)

500

200

100

I , FORWARD CURRENT (mA)

TJ = 150°C 100°C 25°C −55°C

http://onsemi.com

LEAKAGE DC CHARACTERISTICS

When reverse voltage less than the breakdown is applied

to a zener diode, the behavior of current is similar to any

back-biased silicon P-N junction. Ideally, the reverse current

would reach a level at about one volt reverse voltage and

remain constant until breakdown is reached. There are both

theoretical and practical reasons why the typical V-I curve

will have a definite slope to it as seen in Figure 3.

Multiplication effects and charge generation sites are

present in a zener diode which dictate that reverse current

(even at low voltages) will increase with voltage. In

addition, surface charges are ever present across P-N

junctions which appear to be resistive in nature.

The leakage currents are generally less than one

microampere at 150°C except with some large area devices.

Quite often a leakage specification at 80% or so of

breakdown voltage is used to assure low reverse currents.

Figure 3. Typical Leakage Current

versus Voltage

201612640

0.1

100

1000

10000

I , REVERSE LEAKAGE CURRENT (nA)

VR, REVERSE VOLTAGE (VOLTS)

TJ = 150°C

25°C

−55°C

VOLTAGE BREAKDOWN

At some definite reverse voltage, depending on the doping

levels (resistivity) of the P-N junction, the current will begin

to avalanche. This is the so-called “zener” or “breakdown”

area and is where the device is usually biased during use. A

typical family of breakdown curves showing the effect of

temperature is illustrated in Figure 4.

Between the minimum currents shown in Figure 4 and the

leakage currents, there is the “knee” region. The avalanche

mechanism may not occur simultaneously across the entire

area of the P-N junction, but first at one microscopic site,

then at an increasing number of sites as further voltage is

applied. This action can be accounted for by the

“microplasma discharge” theory and correlates with several

breakdown characteristics.

Figure 4. Typical Zener Characteristic

Variation with Temperature

3231302928272625

0.1

0.2

0.5

100

200

500

1000

VZ, ZENER VOLTAGE (VOLTS)

I , ZENER CURRENT (mA)

T = −55°C 100°C25°C 150°C

T = TJ

T = TA

An exaggerated view of the knee region is shown in

Figure 5. As can be seen, the breakdown or avalanche

current does not increase suddenly, but consists of a series

of smoothly rising current versus voltage increments each

with a sudden break point.

Figure 5. Exaggerated V-I Characteristics

of the Knee Region

EXAGGERATED V-I

OF KNEE REGION

ZENER VOLTAGE

VOLTAGE

ZENER CURRENT

CURRENT

At the lowest point, the zener resistance (slope of the

curve) would test high, but as current continues to climb, the

resistance decreases. It is as though each discharge site has

high resistance with each succeeding site being in parallel

until the total resistance is very small.

In addition to the resistive effects, the micro plasmas may

act as noise generators. The exact process of manufacturing

affects how high the noise will be, but in any event there will

be some noise at the knee, and it will diminish considerably

as current is allowed to increase.

Since the zener impedance and the temperature

coefficient are of prime importance when using the zener

diode as a reference device, the next two sections will

expand on these points.

http://onsemi.com

ZENER IMPEDANCE

The slope of the VZ − IZ curve (in breakdown) is defined

as zener impedance or resistance. The measurement is

generally done with a 60 Hz (on modern, computerized

equipment this test is being done at 1 kHz) current variation

whose value is 10% in rms of the dc value of the current.

(That is, ΔIZ peak to peak = 0.282 IZ.) This is really not a

small signal measurement but is convenient to use and gives

repeatable results.

The zener impedance always decreases as current

increases, although at very high currents (usually beyond IZ

max) the impedance will approach a constant. In contrast,

the zener impedance decreases very rapidly with increasing

current in the knee region. On Semiconductor specifies most

zener diode impedances at two points: IZT and IZK. The term

IZT usually is at the quarter power point, and IZK is an

arbitrary low value in the knee region. Between these two

points a plot of impedance versus current on a log-log scale

is close to a straight line. Figure 6 shows a typical plot of ZZ

versus IZ for a 20 volt−500 mW zener. The worst case

impedance between IZT and IZK could be approximated by

assuming a straight line function on a log-log plot; however,

at currents above IZT or below IZK a projection of this line

may give erroneous values.

Figure 6. Zener Impedance versus

Zener Current

ZENER CURRENT (mA)

10010

10.1

100

1000

ZENER IMPEDANCE (OHMS)

APPROXIMATE MAXIMUM LINE

ZZT(MAX)

ZZK(MAX)

The impedance variation with voltage is much more

complex. First of all, zeners below 6 volts or so exhibit “field

emission” breakdown converting to “avalanche” at higher

currents. The two breakdowns behave somewhat differently

with “field emission” associated with high impedance and

negative temperature coefficients and “avalanche” with

lower impedance and positive temperature coefficients.

A V-I plot of several low voltage 500 mW zener diodes is

shown in Figure 7. It is seen that at some given current

(higher for the lower voltage types) there is a fairly sudden

decrease in the slope of ΔV/ΔI. Apparently, this current is the

transition from one type of breakdown to the other. Above

6 volts the curves would show a gradual decrease of ΔV/ΔI

rather than an abrupt change, as current is increased.

Figure 7. Zener Current versus Zener Voltage

(Low Voltage Region)

100

0.1

0.01 234 5678

ZENER VOLTAGE (VOLTS)

ZENER CURRENT (mA)

Possibly the plots shown in Figure 8 of zener impedance

versus voltage at several constant IZ’s more clearly points

out this effect. It is obvious that zener diodes whose

breakdowns are about 7 volts will have remarkably low

impedance.

20010070503020107532

1 mA

10 mA

20 mA

Figure 8. Dynamic Zener Impedance (Typical)

versus Zener Voltage

100

200

Z , DYNAMIC IMPEDANCE (OHMS)

VZ, ZENER VOLTAGE (VOLTS)

TA = 25°C

IZ(ac) = 0.1 IZ(dc)

However, this is not the whole picture. A zener diode

figure of merit as a regulator could be ZZ/VZ. This would

give some idea of what percentage change of voltage could

be expected for some given change in current. Of course, a

low ZZ/VZ is desirable. Generally zener current must be

decreased as voltage is increased to prevent excessive power

dissipation; hence zener impedance will rise even higher and

the “figure of merit” will become higher as voltage

increases. This is the case with IZT taken as the test point.

However, if IZK is used as a comparison level in those

devices which keep a constant IZK over a large range of

voltage, the “figure of merit” will exhibit a bowl-shaped

curve − first decreasing and then increasing as voltage is

increased. Typical plots are shown in Figure 9. The

conclusion can be reached that for uses where wide swings

of current may occur and the quiescent bias current must be

high, the lower voltage zener will provide best regulation,

http://onsemi.com

but for low power applications, the best performance could

be obtained between 50 and 100 volts.

703010 50 90 110 130 150

0.85 mA

17 mA

250 mA

10 W, ZZT(MAX)

1.7 mA

SEE NOTE BELOW

1.3 mA

12.5 mA

3.8 mA

28 mA

75 mA

ZENER VOLTAGE (VOLTS)

ZENER IMPEDANCE (MAX)/ZENER VOLTAGE

100

0.1

(NOTE: CURVE IS APPROXIMATE, ACTUAL

ZZ(MAX) IS ROUNDED OFF TO NEAREST

WHOLE NUMBER ON A DATA SHEET)

Figure 9. Figure of Merit: ZZ(Max)/VZ versus VZ

(400 mW & 10 W Zeners)

400 mW, ZZK(MAX) AT 0.25 mA

10 W, ZZK(MAX) AT 1 mA

400 mW, ZZT(MAX)

TEMPERATURE COEFFICIENT

Below three volts and above eight volts the zener voltage

change due to temperature is nearly a straight line function

and is almost independent of current (disregarding

self-heating effects). However, between three and eight

volts the temperature coefficients are not a simple affair. A

typical plot of TC versus VZ is shown in Figure 10.

23 45 67 89 11

0.01 mA

0.1 mA

30 mA

1 mA

10 mA

Figure 10. Temperature Coefficient versus

Zener Voltage at 25°C Conditions Typical

−1

−2

−3

VZ, ZENER VOLTAGE (VOLTS)

10 12

T , TEMPERATURE COEFFICIENTS (mV/C)

C°

VZ REFERENCE AT IZ = IZT

& TA = 25°C

Any attempt to predict voltage changes as temperature

changes would be very difficult on a “typical” basis. (This,

of course, is true to a lesser degree below three volts and

above eight volts since the curve shown is a typical one and

slight deviations will exist with a particular zener diode.) For

example, a zener which is 5 volts at 25°C could be from 4.9

to 5.05 volts at 75°C depending on the current level.

Whereas, a zener which is 9 volts at 25°C would be close to

9.3 volts at 75°C for all useful current levels (disregarding

impedance effects).

As was mentioned, the situation is further complicated by

the normal deviation of TC at a given current. For example,

for 7.5 mA the normal spread of TC (expressed in %/°C) is

shown in Figure 11. This is based on limited samples and in

no manner implies that all On Semiconductor zeners

between 2 and 12 volts will exhibit this behavior. At other

current levels similar deviations would occur.

+0.08

+0.06

+0.04

+0.02

−0.02

−0.08

−0.10

−0.04

−0.06

02 4 68101214

ZENER VOLTAGE (VOLTS)

TYPICAL

MAX

Figure 11. Temperature Coefficient Spread

versus Zener Voltage

TEMPERATURE COEFFICIENT (%/C)°

MAX

MIN

TYPICAL

IZT = 7.5 mA

Obviously, all of these factors make it very difficult to

attempt any calculation of precise voltage shift due to

temperature. Except in devices with specified maximum

T.C., no “worse case” design is possible. Information

concerning the On Semiconductor temperature

compensated or reference diodes is given in Chapter 4.

Typical temperature characteristics for a broad range of

voltages is illustrated in Figure 12. This graphically shows

the significant change in voltage for high voltage devices

(about a 20 volt increase for a 100°C increase on a 200 volt

device).

NOTE: DV IS + ABOVE 5 VOLTS

− BELOW 4.3 VOLTS

BETWEEN 4.3 & 5 VOLTS

VARIES ABOUT + 0.08 VOLTS

ZENER VOLTAGE (VOLTS)

1 2 3 5 10 50 100 200 1,000

100

0.1

Figure 12. Typical Temperature

Characteristics

V (+25 C TO +125 C)

°°

ΔZ

0.01

http://onsemi.com

POWER DERATING AND MOUNTING

The zener diode like any other semiconductor has a

maximum junction temperature. This limit is somewhat

arbitrary and is set from a reliability viewpoint. Most

semiconductors exhibit an increasing failure rate as

temperature increases. At some temperature, the solder will

melt or soften and the failure rate soars. The 175°C to 200°C

junction temperature rating is quite safe from solder failures

and still has a very low failure rate.

In order that power dissipated in the device will never cause

the junction to rise beyond 175°C or 200°C (depending on the

device), the relation between temperature rise and power

must be known. Of course, the thermal resistance (RθJA or

RθJL) is the factor which relates power and temperature in the

well known “Thermal Ohm’s Law’’ relation:

ΔT = TJ − TA = RθJAPZ(1)

and ΔT = TJ − TL = RθJLPZ(2)

where

RθJA

RθJL

= Junction temperature

= Ambient temperature

= Lead temperature

= Thermal resistance junction to ambient

= Thermal resistance junction to lead

= Zener power dissipation

Obviously, if ambient or lead temperature is known and

the appropriate thermal resistance for a given device is

known, the junction temperature could be precisely

calculated by simply measuring the zener dc current and

voltage (PZ = IZVZ). This would be helpful to calculate

voltage change versus temperature. However, only

maximum and typical values of thermal resistance are given

for a family of zener diodes. So only “worst case” or typical

information could be obtained as to voltage changes.

The relations of equations 1 and 2 are usually expressed

as a graphical derating of power versus the appropriate

temperature. Maximum thermal resistance is used to

generate the slope of the curve. An example of a 400

milliwatt device derated to the ambient temperature and a 1

watt device derated to the lead temperature are shown in

Figures 13 and 14.

500

25 50 75 100 125 150 175 200

400

300

200

100

TA, AMBIENT TEMPERATURE (°C)

Figure 13. 400 mW Power Temperature

Derating Curve

P , POWER DISSIPATION (MILLIWATTS)

1.25

0 20 40 60 80 100 120 140 160 180 200

L = LEAD LENGTH

TO HEAT SINK

0.75

0.50

0.25

L = 1/8″

L = 3/8″

L = 1″

TL, LEAD TEMPERATURE (°C)

Figure 14. Power Temperature

Derating Curve

P , MAXIMUM POWER DISSIPATION (WATTS)

A lead mounted device can have its power rating

increased by shortening the lead length and “heatsinking”

the ends of the leads. This effect is shown in Figure 15, for

the 1N4728, 1 watt zener diode.

Each zener has a derating curve on its data sheet and

steady state power can be set properly. However,

temperature increases due to pulse use are not so easily

calculated. The use of “Transient Thermal Resistance”

would be required. The next section expounds upon

transient thermal behavior as a function of time and surge

power.

175

150

125

100

00.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

L, LEAD LENGTH TO HEAT SINK (INCH)

Figure 15. Typical 1N4728 Thermal

Resistance versus Lead Length

RqJL, JUNCTION−TO−LEAD

THERMAL RESISTANCE (°C/W)

THERMAL TIME RESPONSE

Early studies of zener diodes indicated that a “thermal

time constant” existed which allowed calculation of

temperature rise as a function of power pulse height, width,

and duty cycle. More precise measurements have shown that

temperature response as a function of time cannot be

represented as a simple time constant. Although as shown in

the preceding section, the steady state conditions are

analogous in every way to an electrical resistance; a simple

“thermal capacitance” placed across the resistor is not the

true equivalent circuit. Probably a series of parallel R-C

http://onsemi.com

networks or lumped constants representing a thermal

transmission line would be more accurate.

Fortunately a concept has developed in the industry

wherein the exact thermal equivalent circuit need not be

found. If one simply accepts the concept of a thermal

resistance which varies with time in a predictable manner,

the situation becomes very practical. For each family of

zener diodes, a “worst case” transient thermal resistance

curve may be generated.

The main use of this transient RθJL curve is when the zener

is used as a clipper or a protective device. First of all, the

power wave shape must be constructed. (Note, even though

the power-transient thermal resistance indicates reasonable

junction temperatures, the device still may fail if the peak

current exceeds certain values. Apparently a current

crowding effect occurs which causes the zener to short. This

is discussed further in this section.)

TRANSIENT POWER-TEMPERATURE EFFECTS

A typical transient thermal resistance curve is shown in

Figure 16. This is for a lead mounted device and shows the

effect of lead length to an essentially infinite heatsink.

To calculate the temperature rise, the power surge wave

shape must be approximated by its rectangular equivalent as

shown in Figure 17. In case of an essentially non-recurrent

pulse, there would be just one pulse, and ΔT = RθT1 Pp. In

the general case, it can be shown that

where

RθT1

RθT

RθT1 + T

RθJA(ss) or RθJL(ss) = Steady state value of thermal

RθJA(ss) or RθJL(ss) = resistance

= Duty cycle in percent

= Transient thermal resistance at the time

equal to the pulse width

= Transient thermal resistance at the time

equal to pulse interval

= Transient thermal resistance at the time

equal to the pulse interval

=plus one more pulse width.

ΔT = [DRθJA (ss) + (1 − D) RθT1 + T + RθT1 − RθT] PP

PW, PULSE WIDTH (ms)

100

3 5 10 20 50 100 200 500 1000 2000 5000 10k 30k

L = 1/32″

L = 1″

FOR θJL(t) VALUES AT PULSE WIDTHS

LESS THAN 3.0 ms, THE ABOVE

CURVE CAN BE EXTRAPOLATED

DOWN TO 10 μs AT A CONTINUING

SLOPE OF 1/2

Figure 16. Typical Transient Thermal

Resistance (For Axial Lead Zener)

L L

HEAT SINK

ÉÉ

THERMAL RESISTANCE ( C/W)°

RJL(t), JUNCTION-TO-LEAD TRANSIENT

Figure 17. Relation of Junction Temperature to

Power Pulses

T1T1

PEAK

TEMPERATURE RISE

AVERAGE

TEMPERATURE RISE

AMBIENT

TEMPERATURE

PEAK POWER (PP)

AVERAGE POWER = PP

This method will predict the temperature rise at the end of

the power pulse after the chain of pulses has reached

equilibrium. In other words, the average power will have

caused an average temperature rise which has stabilized, but

a temperature “ripple” is present.

Example: (Use curve in Figure 16)

PP = 5 watt (Lead length 1/32″)

D = 0.1

T1 = 10 ms

T = 100 ms

RθJA(ss) = 12°C/W (for 1/32″ lead length)

ThenRθT1 = 1.8°C/W

RθT = 5.8°C/W

RθT1 + T = 6°C/W

And ΔT = [0.1 x 12 + (1 − 0.1) 6 + 1.8 − 5.8] 5

ΔT = 13°C

Or at TA = 25°, TJ = 38°C peak

SURGE FAILURES

If no other considerations were present, it would be a

simple matter to specify a maximum junction temperature

no matter what pulses are present. However, as has been

noted, apparently other fault conditions prevail. The same

group of devices for which the transient thermal curves were

generated were tested by subjecting them to single shot

power pulses. A failure was defined as a significant shift of

leakage or zener voltage, or of course opens or shorts. Each

device was measured before and after the applied pulse.

Most failures were shifts in zener voltage. The results are

shown in Figure 18.

Attempts to correlate this to the transient thermal

resistance work quite well on a typical basis. For example,

assuming a value for 1 ms of 90 watts and 35 watts at 10 ms,

the predicted temperature rise would be 180°C and 190°C.

But on a worst case basis, the temperature rises would be

about one half these values or junction temperatures, on the

order of 85°C to 105°C, which are obviously low.

Apparently at very high power levels a current restriction

occurs causing hot spots. There was no apparent correlation

http://onsemi.com

of zener voltage or current on the failure points since each

group of failures contained a mixture of voltages.

1000

0.00001 0.0001 0.001 0.01 0.1 1

WORSE CASE

100

POWER (WATTS)

Figure 18. One Shot Power Failure Axial

Lead Zener Diode

TIME OF PULSE (SECONDS)

TYPICAL

VOLTAGE VERSUS TIME

Quite often the junction temperature is only of academic

interest, and the designer is more concerned with the voltage

behavior versus time. By using the transient thermal

resistance, the power, and the temperature coefficient, the

designer could generate VZ versus time curves. The

On Semiconductor zener diode test group has observed

device voltages versus time until the thermal equilibrium

was reached. A typical curve is shown for a lead mounted

low wattage device in Figure 19 where the ambient

temperature was maintained constant. It is seen that voltage

stabilizes in about 100 seconds for 1 inch leads.

Since information contained in this section may not be

found on data sheets it is necessary for the designer to

contact the factory when using a zener diode as a surge

suppressor. Additional information on transient suppression

application is presented elsewhere in this book.

166

0.01

165

164

163

162

161

160 0.1 1 10 100

TIME (SECONDS)

ZENER VOLTAGE (VOLTS)

Figure 19. Zener Voltage (Typical) versus

Time for Step Power Pulses

(500 mW Lead Mounted Devices)

FREQUENCY AND PULSE CHARACTERISTICS

The zener diode may be used in applications which

require a knowledge of the frequency response of the device.

Of main concern are the zener resistance (usually specified

as “impedance”) and the junction capacitance. The

capacitance curves shown in this section are typical.

ZENER CAPACITANCE

Since zener diodes are basically PN junctions operated in

the reverse direction, they display a capacitance that

decreases with increasing reverse voltage. This is so because

the effective width of the PN junction is increased by the

removal of charges (holes and electrons) as reverse voltage

is increased. This decrease in capacitance continues until the

zener breakdown region is entered; very little further

capacitance change takes place, owing to the now fixed

voltage across the junction. The value of this capacitance is

a function of the material resistivity, ρ, (amount of doping −

which determines VZ nominal), the diameter, D, of junction

or dice size (determines amount of power dissipation), the

voltage across the junction VC, and some constant, K. This

relationship can be expressed as:

KD4

pVC

CC =nKD4

ρVC

After the junction enters the zener region, capacitance

remains relatively fixed and the AC resistance then

decreases with increasing zener current.

TEST CIRCUIT CONSIDERATIONS: A capacitive

bridge was used to measure junction capacitance. In this

method the zener is used as one leg of a bridge that is

balanced for both DC at a given reverse voltage and for AC

(the test frequency 1 MHz). After balancing, the variable

capacitor used for balancing is removed and its value

measured on a test instrument. The value thus indicated is

the zener capacitance at reverse voltage for which bridge

balance was obtained. Figure 20 shows capacitance test

circuit.

Figure 21 is a plot of junction capacitance for diffused

zener diode units versus their nominal operating voltage.

Capacitance is the value obtained with reverse bias set at

one-half the nominal VZ. The plot of the voltage range from

6.8 V to 200 V, for three dice sizes, covers most

On Semiconductor diffused-junction zeners. Consult

specific data sheets for capacitance values.

Figures 22, 23, and 24 show plots of capacitance versus

reverse voltage for units of various voltage ratings in each

of the three dice sizes. Junction capacitance decreases as

reverse voltage increases to the zener region. This change in

capacitance can be expressed as a ratio which follows a

one-third law, and C1/C2 = (V2/V1)1/3. This law holds only

from the zener voltage down to about 1 volt, where the curve

begins to flatten out. Figure 25 shows this for a group of low

wattage units.

http://onsemi.com

Figure 20. Capacitance Test Circuit

VDC 1 k

1% 1 k

1 k

POWER

SUPPLY

NO. 712A

100 W

VAC

1 MHz

SIGNAL

GEN

TEK

NO. 190A

10/50

XZENER

UNDER

TEST

HI-GAIN

DIFF SCOPE

NULL IND

TEK

TYPE D

R = ZRR

CAP

DECADE

0−.09 mF

100 pF

STEPS

ΔC 1%

BAL READ

10/150

L/C

METER

TEK

140

0.1 μF

1,000

1 10 100

VR, REVERSE VOLTAGE (VOLTS)

100

C , CAPACITANCE (PICOFARADS)

100 VOLTS

50 VOLTS

20 VOLTS

10 VOLTS

10,000

1 10 100 1,000

1,000

100

C , CAPACITANCE (PICOFARADS) @ VZ/2

VZ, NOMINAL UNIT VOLTAGE (VOLTS)

HIGH WATTAGE

LOW WATTAGE

MEDIUM WATTAGE

Figure 21. Capacitance versus Voltage Figure 22. Capacitance versus Reverse Voltage

LOW WATTAGE UNITS

AVG. FOR 10 UNITS EACH

10,000

110

VR, REVERSE VOLTAGE (VOLTS)

100

C , CAPACITANCE (PICOFARADS)

100

1,000

10,000

1 10 1,000

VR, REVERSE VOLTAGE (VOLTS)

100

C , CAPACITANCE (PICOFARADS)

100

1,000

Figure 23. Capacitance versus

Reverse Voltage

Figure 24. Capacitance versus

Reverse Voltage

100 VOLTS

50 VOLTS

20 VOLTS

10 VOLTS

100 VOLTS

50 VOLTS

20 VOLTS

10 VOLTS

MEDIUM WATTAGE UNITS

AVG. FOR 10 UNITS EACH

HIGH WATTAGE UNITS

AVG. FOR 10 UNITS EACH

http://onsemi.com

C , CAPACITANCE (PICOFARADS)

100

1,000

10 110

VR, REVERSE VOLTAGE (VOLTS)

1000.1

LOW WATTAGE

UNITS

Figure 25. Flattening of Capacitance Curve at

Low Voltages

100 VOLTS

50 VOLTS

10 VOLTS

Figure 26. Impedance Test Circuit

1 k 1 k

1 Ω

RX = ZZ

SUPPLY

712A

Rx = E1 − E2

0.1 μF

600

READ S1 A

READ

SET

10M

100 pF

S1 B

SIGNAL

GEN

650A

VTVM

400H

VTVM

412A

ZENER IMPEDANCE

Zener impedance appears primarily as composed of a

current-dependent resistance shunted by a

voltage-dependent capacitor. Figure 26 shows the test

circuit used to gather impedance data. This is a

voltage-impedance ratio method of determining the

unknown zener impedance. The operation is as follows:

(1) Adjust for desired zener IZDC by observing IR drop

across the 1-ohm current-viewing resistor R2.

(2) Adjust IZAC to 100 μA by observing AC IR drop across

R2.

(3) Measure the voltage across the entire network by

switching S1. The ratio of these two AC voltages is

then a measure of the impedance ratio. This can be

expressed simply as RX = [(E1 − E2)/E2] R2.

Section A of S1 provides a dummy load consisting of a

10-M resistor and a 100 pF capacitor. This network is

required to simulate the input impedance of the AC VTVM

while it is being used to measure the AC IR drop across R2.

This method has been found accurate up to about three

megahertz; above this frequency, lead inductances and strap

capacitance become the dominant factors.

http://onsemi.com

Figure 27 shows typical impedance versus frequency

relationships of 6.8 volt 500 mW zener diodes at various DC

zener currents. Before the zener breakdown region is

entered, the impedance is almost all reactive, being provided

by a voltage-dependent capacitor shunted by a very high

resistance. When the zener breakdown region is entered, the

capacitance is fixed and now is shunted by

current-dependent resistance. For comparison, Figure 27

also shows the plot for a 680 pF capacitor XC, a 1K 1%

nonreactive resistor, R, and the parallel combination of these

two passive elements, ZT.

IZMA

1K & 680 pF

R, 1K 1% DC

10,000

10 100 1 kHz 10 kHz 100 kHz 1 MHz 10 MH

FREQUENCY (Hz)

1,000

100

ZENER

IMPEDANCE

(OHMS)

X , 680 pF

Figure 27. Zener Impedance

versus Frequency

1.00

2.50

0.25

0.50

HIGH FREQUENCY AND SWITCHING

CONSIDERATIONS

At frequencies about 100 kHz or so and switching speeds

above 10 microseconds, shunt capacitance of zener diodes

begins to seriously effect their usefulness. The upper photo

of Figure 28 shows the output waveform of a symmetrical

peak limiter using two zener diodes back-to-back. The

capacitive effects are obvious here. In any application where

the signal is recurrent, the shunt capacitance limitations can

be overcome, as lower photo of Figure 28 shows. This is

done by operating fast diodes in series with the zener. Upon

application of a signal, the fast diode conducts in the forward

direction charging the shunt zener capacitance to the level

where the zener conducts and limits the peak. When the

signal swings the opposite direction, the fast diode becomes

back-biased and prevents fast discharge of shunt

capacitance. The fast diode remains back-biased when the

signal reverses again to the forward direction and remains

off until the input signal rises and exceeds the charge level

of the capacitor. When the signal exceeds this level, the fast

diode conducts as does the zener. Thus, between successive

cycles or pulses the charge in the shunt capacitor holds off

the fast diode, preventing capacitive loading of the signal

until zener breakdown is reached. Figures 29 and 30 show

this method applied to fast-pulse peak limiting.

5 V/cm

0.5 μs/cm

5 V/cm

Figure 28. Symmetrical Peak Limiter

eieo

http://onsemi.com

2 V/cm

20 ns/cm

2 V/cm

20 ns/cm

Figure 29. Shunt Clipper

Figure 30. Shunt Clipper with Clamping Network

200 Ω

50 Ω

eieo

10 V

200 Ω

50 Ω

0.001

http://onsemi.com

Figure 31 is a photo of input-output pulse waveforms

using a zener alone as a series peak clipper. The smaller

output waveform shows the capacitive spike on the leading

edge. Figure 32 clearly points out the advantage of the

clamping network.

2 V/cm

20 ns/cm

2 V/cm

20 ns/cm

Figure 31. Basic Series Clipper

Figure 32. Series Clipper with Clamping Network

10 VZ

50 Ω

eieo

200 Ω

10 VZ

200 Ω

.001

http://onsemi.com

TEMPERATURE COMPENSATED ZENERS

INTRODUCTION

A device which provides reference voltages in a special

manner is a reference diode.

As was discussed in the preceding chapters, the zener

diode has the unique characteristic of exhibiting either a

positive or a negative temperature coefficient, or both. By

properly employing this phenomenon in conjunction with

other semiconductor devices, it is possible to manufacture a

zener reference element exhibiting a very low temperature

coefficient when properly used. This type of low

temperature coefficient device is referred to as a reference

diode.

INTRODUCTION TO REFERENCE DIODES

The temperature characteristics of the zener diode are

discussed in a previous chapter, where it was shown that

change in zener voltage with temperature can be significant

under severe ambient temperature changes (for example, a

100 V zener can change 12.5 volts from 0 to 125°C). The

reference diode (often called the temperature compensated

zener or the TC zener) is specially designed to minimize

these specific temperature effects.

Design of temperature compensated zeners make possible

devices with voltage changes as low as 5 mV from −55 to

+100°C, consequently, the advantages of the temperature

compensated zener are obvious. In critical applications, as

a voltage reference in precision dc power supplies, in high

stability oscillators, in digital voltmeters, in frequency

meters, in analog-to-digital converters, or in other precision

equipment, the temperature compensated zener is a

necessity.

Conceivably temperature compensated devices can be

designed for any voltage but present devices with optimum

voltage temperature characteristics are limited to specific

voltages. Each family of temperature compensated zeners is

attention to the use conditions (temperature range and

current). A distinct operating current is associated with each

device. Consequently, changes from the specified operating

current can only degrade the voltage-temperature

relationships. This will be discussed in more detail later.

The device “drift” or voltage-time stability is critical in

some reference applications. Typically zeners and TC

zeners offer stability of better than 500 parts per million per

1000 hours.

TEMPERATURE CHARACTERISTICS OF THE P-N

JUNCTION AND COMPENSATION

The voltage of a forward biased P-N junction, at a specific

current, will decrease with increasing temperature. Thus, a

device so biased displays a negative temperature coefficient

(Figure 1). A P-N junction in avalanche (above 5 volts

breakdown) will display a positive temperature coefficient;

that is, voltage will increase as temperature increases. Due

to energy levels of a junction which breaks down below

5 volts, the temperature coefficient is negative.

It follows that various combinations of forward biased

junctions and reverse biased junctions may be arranged to

achieve temperature compensation. From Figure 2 it can be

seen that if the absolute value of voltage change (ΔV) is the

same for both the forward biased diode and the zener diode

where the temperature has gone from 25°C to 100°C, then

the total voltage across the combination will be the same at

both temperatures since one ΔV is negative and the other

positive. Furthermore, if the rate of increase (or decrease) is

the same throughout the temperature change, voltage will

remain constant. The non-linearity associated with the

voltage temperature characteristics is a result of this rate of

change not being a perfect match.

VREF = VZ + ΔVZ + VD − ΔVD

THE METHODS OF TEMPERATURE

COMPENSATION

The effect of temperature is shown in Figure 1. The

forward characteristic does not vary significantly with

reverse voltage breakdown (zener voltage) rating. A change

in ambient temperature from 25° to 100°C produces a shift

in the forward curve in the direction of lower voltage (a

negative temperature coefficient — in this case about

150 mV change), while the same temperature change

produces approximately 1.9 V increase in the zener voltage

(a positive coefficient). By combining one or more silicon

diodes biased in the forward direction with the P-N biased

zener diode as shown in Figure 3, it is possible to

compensate almost completely for the zener temperature

coefficient. Obviously, with the example shown, 13

junctions would be needed. Usually reference diodes are low

voltage devices, using zeners with 6 to 8 volts breakdown

and one or two forward diodes.

http://onsemi.com

Figure 1. Effects of Temperature on Zener Diode Characteristics

FORWARD

CHARACTERISTIC

TYPICAL

(ALL TYPES)

VZ (VOLTS)

VF (VOLTS)

30 20 10

I (mA)

0.5 1 1.5

450

300

150

1.9 V

100°C

25°C

100°C

150 mV

25°C

I (mA)

Figure 2. Principle of Temperature Compensation

Figure 3. Zener Temperature Compensation with Silicon Forward Junctions

DIRECTION OF CURRENT FLOW

PACKAGE

OUTLINE

FORWARD-BIASED

PN JUNCTION

REVERSE-BIASED

ZENER JUNCTION

7.5 mA

100°C

25°C

−ΔV

+ΔV+ΔV

100°C25°C

−V

100°C25°C

100°C

25°C

7.5 mA

−

SILICON JUNCTION

DIODES

ZENER DIODES

−

http://onsemi.com

In ac regulator and clipper circuits where zener diodes are

normally connected cathode to cathode, the forward biased

diode during each half cycle can be chosen with the correct

forward temperature coefficient (by stacking, etc.) to

correctly compensate for the temperature coefficient of the

reverse-biased zener diode. It is possible to compensate for

voltage drift with temperature using this method to the

extent that zener voltage stabilities on the order of

0.001%/°C are quite feasible.

This technique is sometimes employed where higher

wattage devices are required or where the zener is

compensated by the emitter base junction of a transistor

stage. Consider the example of using discrete components,

1N4001 rectifier and ON Semiconductor 5 Watt zener, to

obtain compensated voltage-temperature characteristics.

Examination of the curve in Figure 4 indicates that a 10 volt

zener diode exhibits a temperature coefficient of

approximately +5.5 mV/°C. At a current level of 100 mA a

temperature coefficient of approximately −2.0 mV/°C is

characteristic of the 1N4001 rectifiers. A series connection

of three silicon 1N4001 rectifiers produces a total

temperature coefficient of approximately −6 mV/°C and a

total forward drop of approximately 2.17 volts at 25°C. The

combination of three silicon rectifiers and the 10 volt zener

diode produces a device with a coefficient of approximately

−0.5 mV/°C and a total breakdown voltage at 100 mA of

approximately 12.2 volts. Calculation shows this to be a

temperature stability of −0.004%/°C.

ǒ*0.5 mVńC

12.2 V Ǔ 100

The temperature compensated zener employs the

technique of specially selected dice. This provides optimum

voltage temperature characteristics by close control of dice

resistivities.

−1

−2

−30 1 2 3 4 5 6 7 8 9 10 11 12 13

ZENER VOLTAGE (10 mA AT 25°C)

ALLOY-DIFFUSED

JUNCTION

THREE

FORWARDS

ONE

FORWARD

TWO

FORWARDS

DIFFUSED

JUNCTION

VOLTS

Figure 4.

mV/ C

TEMPERATURE COEFFICIENT STABILITY

Figure 5 shows the voltage-temperature characteristics of

the TC diode. It can be seen that the voltage drops slightly

with increasing temperature.

VOLTAGE (VOLTS)

mV CHANGE FROM 25 C VOLTAGE

6.326

6.324

6.322

6.320

6.318

6.316

6.314

TEMPERATURE (°C)

−55 −10 25 62 100

Figure 5. Voltage versus Temperature,

Typical for ON Semiconductor 1N827

Temperature Compensated Zener Diode

−1

−2

−3

−4

−5

−6

This non-linearity of the voltage temperature

characteristic leads to a definition of a representative design

parameter ΔVZ. For each device type there is a specified

maximum change allowable. The voltage temperature

stability measurement consists of voltage measurement at

specified temperatures (for the 1N821 Series the

temperatures are −55, 0, +25, +75, and +100°C). The voltage

readings at each of the temperatures is compared with

readings at the other temperatures and the largest voltage

change between any of the specified temperatures

determines the exact device type. For devices registered

prior to complete definition of the voltage temperature

stability measurement, the allowable maximum voltage

change over the temperature range is derived from the

calculation converting %/°C to mV over the temperature

range. Under this standard definition, %/°C is merely a

nomenclature and the meaningful allowable voltage

deviation to be expected becomes the designed parameter.

http://onsemi.com

CURRENT

Thus far, temperature compensated zeners have been

discussed mainly with regard to temperature and voltage.

However, the underlying assumption throughout the

previous discussion was that current remained constant.

There is a significant change in the temperature

coefficient of a unit depending on how much above or below

the test current the device is operated.

A particular unit with a 0.01%/°C temperature coefficient

at 7.5 mA over a temperature range of −55°C to +100°C

could possibly have a 0.0005%/°C temperature coefficient

at 11 mA. In fact, there is a particular current which can be

determined for each individual unit that will give the lowest

TC.

Manufacturing processes are designed so that the yields of

low TC units are high at the test specification for current. A

unit with a high TC at the test current can have a low TC at

some other current. A look at the volt-ampere curves at

different temperatures illustrates this point clearly (see

Figure 6).

VOLTAGE (VOLTS)

−6.6 −6.5 −6.4 −6.3 −6.2 −6.1 −6 −5.9

CURRENT

25°C

−55 +100

ΔVB

ΔVC

Figure 6. Voltage-Ampere Curves Showing

Crossover at A

If the three curves intersect at A, then operation at IA

results in the least amount of voltage deviation due to

temperature from the +25°C voltage. At IB and IC there are

greater excursions (ΔVB and ΔVC) from the +25°C voltage

as temperature increases or decreases.

THE EFFECTS OF POOR CURRENT

REGULATION

If current shifts (randomly or as a function of

temperature), then an area of operation can be defined for the

temperature compensated zener.

Once again the curves are drawn, this time a shaded area

is shown on the graph. The upper and lower extremities

denote the maximum current values generated by the current

supply while the voltage extremes at each current are shown

by the left and right sides of the area, shown in Figure 7.

VOLTAGE (VOLTS)

−6.6 −6.5 −6.4 −6.3 −6.2 −6.1 −6 −5.9

CURRENT

25°C

−55 +100

ΔIMAX

ΔV

MAX

Figure 7. Effects of Poorly Regulated Current

The three volt-ampere curves do not usually cross over at

exactly the same point. However, this does not take away

from the argument that current regulation is probably the

most critical consideration when using

temperature-compensated units.

ZENER IMPEDANCE AND CURRENT

REGULATION

Zener impedance is defined as the slope of the V-I curve

at the test point corresponding to the test current. It is

measured by superimposing a small ac current on the dc test

current and then measuring the resulting ac voltage. This

procedure is identical with that used for regular zeners.

Impedance changes with temperature, but the variation is

usually small and it can be assumed that the amount of

current regulation needed at +25°C will be the same for other

temperatures.

As an example, one might want to determine the amount

of current regulation necessary for the device described

below when the maximum deviation in voltage due to

current variation is ±5 millivolts.

VZT = 6.32 V

IZT = 7.5 mA

ZZT = 15 Ω @ +25°C

ΔV = ΔI⋅VZZT

0.005 = ΔI⋅V15

ΔI = 0.005

15 = 0.33 mA

Therefore, the current cannot vary more than 0.33 mA.

The amount of current regulation necessary is:

0.33

7.5 x 100% = 4.5% regulation.

If the device of Figure 5 is considered to be the device used

in the preceding discussion, it becomes apparent that on the

average more voltage variation is due to current fluctuation

than is due to temperature variation. Therefore, to obtain a

truly stable reference source, the device must be driven from

a constant current source.

http://onsemi.com

BASIC VOLTAGE REGULATION USING ZENER DIODES

BASIC CONCEPTS OF REGULATION

The purpose of any regulator circuit is to minimize output

variations with respect to variations in input, temperature,

and load requirements. The most obvious use of a regulator

is in the design of a power supply, but any circuit that

incorporates regulatory technique to give a controlled

output or function can be considered as a regulator. In

general, to provide a regulated output voltage, electronic

circuitry will be used to pass an output voltage that is

significantly lower than the input voltage and block all

voltage in excess of the desired output. Allocations should

also be made in the regulation circuitry to maintain this

output voltage for variation in load current demand.

There are some basic rules of thumb for the electrical

requirements of the electronic circuitry in order for it to

provide regulation. Number one, the output impedance

should be kept as low as possible. Number two, a controlling

reference needs to be established that is relatively

insensitive to the prevailing variables. In order to illustrate

the importance of these rules, an analysis of some simple

regulator circuits will point out the validity of the

statements. The circuit of Figure 1 can be considered a

regulator. This circuit will serve to illustrate the importance

of a low output impedance.

The resistors RS and RR can be considered as the source

and regulator impedances, respectively.

The output of the circuit is:

ǒRS

RRRL

RR)RLǓ(1)

VO = VI x RRRL

RR + RLRS + RRRL

RR+R

=VI

++ 1

Figure 1. Shunt Resistance Regulator

−

RRRL

−

VIVO

For a given incremental change in VI, the changes in VO

will be:

(2)

ǒRRRRL

RR)Ǔ

ΔVO = ΔVI

++ 1

Assuming RL fixed at some constant value, it is obvious

from equation (2) that in order to minimize changes in VO

for variations in VI, the shunt resistor RR should be made as

small as possible with respect to the source resistor RS.

Obviously, the better this relation becomes, the larger VI is

going to have to be for the same VO, and not until the ratio

of RS to RR reaches infinity will the output be held entirely

constant for variation in VI. This, of course, is an

impossibility, but it does stress the fact that the regulation

improves as the output impedance becomes lower and lower.

Where the output impedance of Figure 1 is given by

(3)

RO = RSRR

RS+R

It is apparent from this relation that as regulation is

improving with RS increasing and RR decreasing the output

impedance RO is decreasing, and is approximately equal to

RR as the ratio is 10 times or greater. The regulation of this

circuit can be greatly improved by inserting a reference

source of voltage in series with RR such as Figure 2.

Figure 2. Regulator with Battery Reference Source

−

VIVO

The resistance RR represents the internal impedance of the

battery. For this circuit, the output is

(4)

VO = VR + VI

++ 1

Then for incremental changes in the input VI, the changes

in VO will be dependent on the second term of equation

(4), which again makes the regulation dependent on the

ratio of RS to RR. Where changes in the output voltage

or the regulation of the circuit in Figure 1 were directly

and solely dependent upon the input voltage and output

impedance, the regulation of circuit 2 will have an output

that varies about the reference source VR in accordance

with the magnitude of battery resistance RR and its

fluctuations for changes in VI. Theoretically, if a perfect

battery were used, that is, VR is constant and RR is zero,

the circuit would be a perfect regulator. In other words,

in line with the basic rules of thumb the circuit exhibits

optimum regulation with an output impedance of zero, and

a constant reference source.

For regulator application, a zener diode can be used

instead of a battery with a number of advantages. A battery’s

resistance and nominal voltage will change with age and

load demand; the ON Semiconductor zener diode

characteristics remain unchanged when operating within its

http://onsemi.com

specified limits. Any voltage value from a couple of volts to

hundreds of volts is available with zener diodes, where

conventional batteries are limited in the nominal values

available. Also, the zener presents a definite size advantage,

and is less expensive than a battery because it is permanent

and need not be regularly replaced. The basic zener diode

shunt regulator circuit is shown in Figure 3.

Figure 3. Basic Zener Diode Shunt Regulator

−

ZENER

DIODE

−

VIVO

Depending upon the operating conditions of the device, a

zener diode will exhibit some relatively low zener

impedance RZ and have a specified breakover voltage of VZ

that is essentially constant. These inherent characteristics

make the zener diode suited for voltage regulator

applications.

DESIGNING THE ZENER SHUNT REGULATOR

For any given application of a zener diode shunt regulator,

it will be required to know the input voltage variations and

output load requirements. The calculation of component

values will be directly dependent upon the circuit

requirements. The input may be constant or have maximum

and minimum values depending upon the natural regulation

or waveform of the supply source. The output voltage will

be determined by the designer’s choice of VZ and the circuit

requirements. The actual value of VZ will be dependent

upon the manufacturer’s tolerance and some small variation

for different zener currents and operating temperatures.

For all practical purposes, the value of VZ as specified on

the manufacturer’s data sheet can be used to approximate

VO in computing component values. The requirement for

load current will be known and will vary within some given

range of IL(min) to IL(max).

The design objective of Figure 3 is to determine the proper

values of the series resistance, RS, and zener power

dissipation, PZ. A general solution for these values can be

developed as follows, when the following conditions are

known:

VI (input voltage) from VI(min) to VI(max)

VO (output voltage) from VZ(min) to VZ(max)

IL (load current) from IL(min) to IL(max)

The value of RS must be of such a value so that the zener

current will not drop below a minimum value of IZ(min).

This minimum zener current is mandatory to keep the

device in the breakover region in order to maintain the

zener voltage reference. The minimum current can be either

chosen at some point beyond the knee or found on the

manufacturer’s data sheet (IZK). The basic voltage loop

equation for this circuit is:

(5)

VI = (IZ + IL)RS + VZ

The minimum zener current will occur when VI is

minimum, VZ is maximum, and IL is maximum, then

solving for RS, we have:

(6)

VI(min) − VZ(max)

IZ(min) + IL(max)

RS =

Having found RS, we can determine the maximum power

dissipation PZ for the zener diode.

PZ(max) =V

Z(max)

ƪVI(max) *VZ(min)

RS *IL(min)ƫ

(7)

(8)

(9)

PZ(max) = IZ(max) VZ(max)

Where:

IZ(max) =VI(max) − VZ(min)

− IL(min)

Therefore:

VI(max) − VZ(min)

− IL(min)

Once the basic regulator components values have been

determined, adequate considerations will have to be given to

the variation in VO. The changes in VO are a function of four

different factors; namely, changes in VI, IL, temperature, and

the value of zener impedance, RZ. These changes in VO can

be expressed as:

(10)

RSRZ

RS + RZ

ΔVI

1 +

ΔVO = −ΔIL + TCΔTVZ

The value of ΔVO as calculated with equation (10) will

quite probably be slightly different from the actual value

when measured empirically. For all practical purposes

though, this difference will be insignificant for regulator

designs utilizing the conventional commercial line of zener

diodes.

Obviously to precisely predict ΔVO with a given zener

diode, exact information would be needed about the zener

impedance and temperature coefficient throughout the

variation of zener current. The “worst case” change can only

be approximated by using maximum zener impedance and

with typical temperature coefficient.

The basic zener shunt regulator can be modified to

minimize the effects of each term in the regulation equation

(10). Taking one term at a time, it is apparent that the

regulation or changes in output ΔVO will be improved if the

magnitude of ΔVI is reduced. A practical and widely used

technique to reduce input variation is to cascade zener shunt

regulators such as shown in Figure 4.

http://onsemi.com

Figure 4. Cascaded Zener Shunt Regulators Reduce

ΔVO by Reducing ΔVI to the Succeeding Stages

−

RS2

Z1RL

RS1

−

RL′

VIVO

VO′= VI′= VZ1

This, in essence, is a regulator driven with a pre-regulator

so that the over all regulation is the product of both. The

regulation or changes in output voltage is determined by:

Where:

RL′ = RS2 + RLRZ2

RL + RZ2

and IL′ = IL + IZ2

(11)

RS2RZ2

RS2 + RZ2

ΔVZI

RS2

ΔVO = −

ΔIL + TC2 ΔTVZ2

RS2

RZ2

(12)

RS1RZ1

RS1 +R

ΔVI

RS1

RL′+1+

ΔVZ1 = ΔVO′ = −

ΔIL′ + TC1 ΔTVZ1

RS1

RZ1

The changes in output with respect to changes in input for

both stages assuming the temperature and load are constant

(13)

ΔVO

ΔVZ1

ΔVO

ΔVO′

= = Regulation of second stage

(14)

(15)

= Regulation of first stage

ΔVO′

ΔVI

= Combined regulationx

ΔVO

ΔVO′

ΔVO

ΔVI

ΔVO′

ΔVI

Obviously, this technique will vastly improve overall

regulation where the input fluctuates over a relatively wide

range. As an example, let’s say the input varies by ±20% and

the regulation of each individual stage reduces the variation

by a factor of 1/20. This then gives an overall output

variation of ±20% × (1/20)2 or ±0.05%.

The next two factors in equation (10) affecting regulation

are changes in load current and temperature excursions. In

order to minimize changes for load current variation, the

output impedance RZRS/(RZ + RS) will have to be reduced.

This can only be done by having a lower zener impedance

because the value of RS is fixed by circuit requirements.

There are basically two ways that a lower zener impedance

can be achieved. One, a higher wattage device can be used

which allows for an increase in zener current of which will

reduce the impedance. The other technique is to series lower

voltage devices to obtain the desired equivalent voltage, so

that the sum of the impedance is less than that for a single

high voltage device. So to speak, this technique will kill two

birds with one stone, as it can also be used to minimize

temperature induced variations of the regulator.

In most regulator applications, the single most detrimental

factor affecting regulation is that of variation in junction

temperature. The junction temperature is a function of both

the ambient temperature and that of self heating. In order to

illustrate how the overall temperature coefficient is

improved with series lower voltage zener, a mathematical

relationship can be developed. Consider the diagram of

Figure 5.

Figure 5. Series Zener Improve Dynamic Impedance

and Temperature Coefficient

−

VIVO

With the temperature coefficient TC defined as the % change

per °C, the change in output for a given temperature range

will equal some overall TC x ΔT x Total VZ. Such as

(16)

ΔVO(ΔT) = TC ΔT (VZ1 + VZ2 + . . . + VZN)

Obviously, the change in output will also be equal to the

sum of the changes as attributed from each zener.

(17)

ΔVO(ΔT) = ΔT(TC1VZ1 + TC2VZ2 + . . . + TCNVZN)

Setting the two equations equal to each other and solving

for the overall TC, we get

(18)

(19)

TCΔT(VZ1 + VZ2 + . . . + VZN) = ΔT(TC1VZ1

+ TC2VZ2 + . . . + TCNVZN)

TC = TC1 VZ1 + TC2VZ2 + . . . + TCNVZN

VZ1 + VZ2 + . . . + VZN

For equation (19) the overall temperature coefficient for

any combination of series zeners can be calculated. Say for

instance several identical zeners in series replace a single

higher voltage zener. The new overall temperature

coefficient will now be that of one of the low voltage

devices. This allows the designer to go to the manufacturer’s

http://onsemi.com

data sheet and select a combination of low TC zener diodes

in place of the single higher TC devices. Generally speaking,

the technique of using multiple devices will also yield a

lower dynamic impedance. Advantages of this technique are

best demonstrated by example. Consider a 5 watt diode with

a nominal zener voltage of 10 volts exhibits approximately

0.055% change in voltage per degree centrigrade, a 20 volt

unit approximately 0.075%/°C, and a 100 volt unit

approximately 0.1%/°C. In the case of the 100 volt diode,

five 20 volt diodes could be connected together to provide

the correct voltage reference, but the overall temperature

coefficient would remain that of the low voltage units, i.e.

0.075%/°C. It should also be noted that the same series

combination improves the overall zener impedance in

addition to the temperature coefficient. A 20 volt, 5 watt

ON Semiconductor zener diode has a maximum zener

impedance of 3 ohms, compared to the 90 ohms impedance

which is maximum for a 100 volt unit. Although these

impedances are measured at different current levels, the

series impedance of five 20 volt zener diodes is still much

lower than that of a single 100 volt zener diode at the test

current specified on the data sheet.

For the ultimate in zener shunt regulator performance, the

aforementioned techniques can be combined with the proper

selection of devices to yield an overall improvement in

regulation. For instance, a multiple string of low voltage

zener diodes can be used as a preregulator, with a series

combination of zero TC reference diodes in the final stage

such as Figure 6.

The first stage will reduce the large variation in VI to some

relatively low level, i.e. ΔVZ. This ΔVZ is optimized by

utilizing a series combination of zeners to reduce the overall

TC and ΔVZ. Because of this small fluctuation of input to the

second stage, and if RL is constant, the biasing current of the

TC units can be maintained at their specified level. This will

give an output that is very precise and not significantly

affected by changes in input voltage or junction temperature.

Figure 6. Series Zeners Cascaded With Series

Reference Diodes for Improved Zener

Shunt Regulation

RS1 RS2

TC1

TC2

−

VIVO

The basic zener shunt regulator exhibits some inherent

limitations to the designer. First of all, the zener is limited to

its particular power dissipating rating which may be less

than the required amount for a particular situation. The total

magnitude of dissipation can be increased to some degree by

utilizing series or parallel units. Zeners in series present few

problems because individual voltages are additive and the

devices all carry the same current and the extent that this

technique can be used is only restricted by the feasibility of

circuit parameters and cost. On the other hand, caution must

be taken when attempting to parallel zener diodes. If the

devices are not closely matched so that they all break over

at the same voltage, the low voltage device will go into

conduction first and ultimately carry all the current. In order

to avoid this situation, the diodes should be matched for

equal current sharing.

EXTENDING POWER AND CURRENT RANGE

The most common practice for extending the power

handling capabilities of a regulator is to incorporate

transistors in the design. This technique is discussed in detail

in the following sections of this chapter. The second

disadvantage to the basic zener shunt regulator is that

because the device does not have a gain function, a feedback

system is not possible with just the zener resistor

combination. For very precise regulators, the design will

normally be an electronic circuit consisting of transistor

devices for control, probably a closed loop feedback system

with a zener device as the basic referencing element.

The concept of regulation can be further extended and

improved with the addition of transistors as the power

absorbing elements to the zener diodes establishing a

reference. There are three basic techniques used that

combine zener diodes and transistors for voltage regulation.

The shunt transistor type shown in Figure 7 will extend the

power handling capabilities of the basic shunt regulator, and

exhibit marked improvement in regulation.

Figure 7. Basic Transistor Shunt Regulator

ZIZ

VBE

Q1IL

−

VIVO

In this configuration the source resistance must be large

enough to absorb the overvoltage in the same manner as in

the conventional zener shunt regulator. Most of the shunt

regulating current in this circuit will pass through the

transistor reducing the current requirements of the zener

diode by essentially the dc current gain of the transistor hFE.

Where the total regulating shunt current is:

(20)

IS = IZ + IC = IZ + IB hFE

where

therefore

IZ = IB + IRB and IB >> IRB

IS ≈ IZ + IZ hFE = IZ (1 + hFE)

http://onsemi.com

The output voltage is the reference voltage VZ plus the

forward junction drop from base to emitter VBE of the

transistor.

(21)VO = VZ + VBE

The values of components and their operating condition

is dictated by the specific input and output requirements and

the characteristics of the designer’s chosen devices, as

shown in the following relations:

(22)RS = VI(min) − VO(max)

IZ(min) [1+hFE(min)] + IL(max)

RB = VI(min) − VZ(max)

IZ(min)

PDZ = IZ(max) VZ(max)

when

IZ(max) = ƪVI(max) *VO(min)

RS*IL(min) ƫ

ǒ1

1)hFE(min)

hence

VI(max) − VO(min)

− IL(min)

1 + hFE(min)

PDZ = ƪVI(max) *VO(min)

RS*IL(min) ƫ

ǒ1

1)hFE(min)Ǔ

VI(max) − VO(min)

− IL(min)

VZ(max)

1 + hFE(min)

PDQ =

VI(max) *VO(min)

RS*IL(min)

ǒ1 888888 Ǔ

VI(max) − VO(min)

− IL(min) VO(max)

(23)

(24)

(25)

(26)

(27)

Regulation with this circuit is derived in essentially the

same manner as in the shunt zener circuit, where the output

impedance is low and the output voltage is a function of the

reference voltage. The regulation is improved with this

configuration because the small signal output impedance is

reduced by the gain of Q1 by 1/hFE.

One other highly desirable feature of this type of regulator

is that the output is somewhat self compensating for

temperature changes by the opposing changes in VZ and

VBE for VZ ≈ 10 volts. With the zener having a positive

2 mV/°C TC and the transistor base to emitter being a

negative 2 mV/°C TC, therefore, a change in one is cancelled

by the change in the other. Even though this circuit is a very

effective regulator it is somewhat undesirable from an

efficiency standpoint. Because the magnitude of RS is

required to be large, and it must carry the entire input

current, a large percentage of power is lost from input to

output.

EMITTER FOLLOWER REGULATOR

Another basic technique of transistor-zener regulation is

that of the emitter follower type shown in Figure 8.

Figure 8. Emitter Follower Regulator

IRB

−VBE

−

VIVO

This circuit has the desirable feature of using a series

transistor to absorb overvoltages instead of a large fixed

resistor, thereby giving a significant improvement in

efficiency over the shunt type regulator. The transistor must

be capable of carrying the entire load current and

withstanding voltages equal to the input voltage minus the

load voltage. This, of course, imposes a much more stringent

power handling requirement upon the transistor than was

required in the shunt regulator. The output voltage is a

function of the zener reference voltage and the base to

emitter drop of Q1 as expressed by the equation (28).

(28)VO = VZ − VBE

The load current is approximately equal to the transistor

collector current, such as shown in equation (29).

(29)

IL(max) ≈ IC(max)

The designer must select a transistor that will meet the

following basic requirements:

(30)

PD ≅ (VI(max) − VO)IL(max)

IC(max) ≈ IL(max)

BVCES ≥ (VI(max) − VO)

Depending upon the designer’s choice of a transistor and

the imposed circuit requirements, the operation conditions

of the circuit are expressed by the following equations:

(31)

VZ = VO + VBE

= VO + IL(max)/gFE(min) @ IL(max)

RS = VI(min) − VZ − VCE(min) @ IL(max)

IL(max)

Where VCE(min) is an arbitrary value of minimum

collector to emitter voltage and gFE is the transconductance.

This is sufficient to keep the transistor out of saturation,

which is usually about 2 volts.

http://onsemi.com

(32)

RB = VCE(min) @ IL(max)

IL(max)/hFE(min) @ IL(max) + IZ(min)

IZ(max) = VI(max) −VZ

RB + RZ

PDZ = IZ(max)VZ

Actual PDQ = (VI(max) − VO) IL(max)

(33)

(34)

(35)

There are two primary factors that effect the regulation

most in a circuit of this type. First of all, the zener current

may vary over a considerable range as the input changes

from minimum to maximum and this, of course, may have

a significant effect on the value of VZ and therefore VO.

Secondly, VZ and VBE will both be effected by temperature

changes which are additive on their effect of output voltage.

This can be seen by altering equation (28) to show changes

in VO as dependent on temperature, see equation (36).

(36)

VO(ΔT) = ΔT[(+TC) VZ − (−TC) VBE]

The effects of these detrimental factors can be minimized

by replacing the bleeder resistor RB with a constant current

source and the zener with a reference diode in series with a

forward biased diode (see Figure 9).

Figure 9. Improved Emitter Follower Regulator

−

FORWARD

BIAS DIODE

TC ZENER

IB = K

CONSTANT

CURRENT

SOURCE

VIVO

The constant current source can be either a current limiter

diode or a transistor source. The current limiter diode is

ideally suited for applications of this type, because it will

supply the same biasing current irregardless of collector to

base voltage swing as long as it is within the voltage limits

of the device. This technique will overcome changes in VZ

for changes in IZ and temperature, but changes in VBE due

to load current changes are still directly reflected upon the

output. This can be reduced somewhat by combining a

transistor with the zener for the shunt control element as

illustrated in Figure 10.

Figure 10. Series Pass Regulator

−

IC2

CONSTANT

CURRENT

SOURCE

VIVO

This is the third basic technique used for transistor-zener

regulators. This technique or at least a variation of it, finds

the widest use in practical applications. In this circuit the

transistor Q1 is still the series control device operating as an

emitter follower. The output voltage is now established by

the transistor Q2 base to emitter voltage and the zener

voltage. Because the zener is only supplying base drive to

Q2, and it derives its bias from the output, the zener current

remains essential constant, which minimizes changes in VZ

due to IZ excursions. Also, it may be possible (VZ ≈ 10 V)

to match the zener to the base-emitter junction of Q2 for an

output that is insensitive to temperature changes. The

constant current source looks like a very high load

impedance to the collector of Q2 thus assuming a very high

voltage gain. There are three primary advantages gained

with this configuration over the basic emitter follower:

1. The increased voltage gain of the circuit with the

addition of Q2 will improve regulation for changes in

both load and input.

2. The zener current excursions are reduced, thereby

improving regulation.

3. For certain voltages the configuration allows good

temperature compensation by matching the

temperature characteristics of the zener to the

base-emitter junction of Q2.

The series pass regulator is superior to the other transistor

regulators thus far discussed. It has good efficiency, better

stability and regulation, and is simple enough to be

economically practical for a large percentage of

applications.

http://onsemi.com

Figure 11. Block Diagram of Regulator with Feedback

REGULATING

POWER

ELEMENT

CONTROL

UNIT

AMPLIFIER

REFERENCE

AND

ERROR

DETECTION

LOAD

INPUT OUTPUT

EMPLOYING FEEDBACK FOR

OPTIMUM REGULATION

The regulators discussed thus far do not employ any

feedback techniques for precise control and compensation

and, therefore, find limited use where an ultra precise

regulator is required. In the more sophisticated regulators

some form of error detection is incorporated and amplified

through a feedback network to closely control the power

elements as illustrated in the block diagram of Figure 11.

Regulating circuits of this type will vary in complexity

and configuration from application to application. This

technique can best be illustrated with a couple of actual

circuits of this type. The feedback regulators will generally

be some form of series pass regulator, for optimum

performance and efficiency. A practical circuit of this type

that is extensively utilized is shown in Figure 12.

In this circuit, the zener establishes a reference level for

the differential amplifier composed of Q4 and Q5 which will

set the base drive for the control transistor Q3 to regulate the

series high gain transistor combination of Q1 and Q2. The

differential amplifier samples the output at the voltage

dividing network of R8, R9, and R10. This is compared to the

reference voltage provided by the zener Z1. The difference,

if any, is amplified and fed back to the control elements. By

adjusting the potentiometer, R9, the output level can be set

to any desired value within the range of the supply. (The

output voltage is set by the relation VO = VZ[(RX +

RY)/RX].) By matching the transistor Q4 and Q5 for

variations in VBE and gain with temperature changes and

incorporating a temperature compensated diode as the

reference, the circuit will be ultra stable to temperature

effects. The regulation and stability of this circuit is very

good, and for this reason is used in a large percentage of

commercial power supplies.

Figure 12. Series Pass Regulator with Error Detection and Feedback

Amplification Derived from a Differential Amplifier

Q2Q4

Q3C1

R5R6

R10

−−

VIVO

http://onsemi.com

Figure 13. Series Pass Regulator with Temperature Compensated Reference Amplifier

−

R2RL

REFERENCE

AMPLIFIER

VIVO

Another variation of the feedback series pass regulator is

shown in Figure 13. This circuit incorporates a stable

temperature compensated reference amplifier as the primary

control element.

This circuit also employs error detection and amplified

feedback compensation. It is an improved version over the

basic series pass regulator shown in Figure 10. The series

element is composed of a Darlington high gain

configuration formed by Q1 and Q2 for an improved

regulation factor. The combined gain of the reference

amplifier and Q3 is incorporated to control the series unit.

This reduced the required collector current change of the

reference amplifier to control the regulator so that the bias

current remains close to the specified current for low

temperature coefficient. Also the germanium diode D1 will

compensate for the base to emitter change in Q3 and keep the

reference amplifier collector biasing current fairly constant

with temperature changes. Proper biasing of the zener and

transistor in the reference amplifier must be adhered to if the

output voltage changes are to be minimized.

CONSTANT CURRENT SOURCES FOR

REGULATOR APPLICATIONS

Several places throughout this chapter emphasize the

need for maintaining a constant current level in the various

biasing circuits for optimum regulation. As was mentioned

previously in the discussion on the basic series pass

regulator, the current limiter diode can be effectively used

for the purpose.

Aside from the current limiter diode a transistorized

source can be used. A widely used technique is shown

incorporated in a basic series pass regulator in Figure 14.

The circuit is used as a preregulated current source to

supply the biasing current to the transistor Q2. The constant

current circuit is seldom used alone, but does find wide use

in conjunction with voltage regulators to supply biasing

current to transistors or reference diodes for stable

operation. The Zener Z2 establishes a fixed voltage across

RE and the base to emitter of Q3. This gives an emitter

current of IE = (VZ − VBE)/RE which will vary only slightly

for changes in input voltage and temperature.

Figure 14. Constant Current Source Incorporated in a Basic Regulator Circuit

RB1

RB2

CONSTANT

CURRENT

SOURCE

−−

VIVO

http://onsemi.com

IMPEDANCE CANCELLATION

One of the most common applications of zener diodes is

in the general category of reference voltage supplies. The

function of the zener diode in such applications is to provide

a stable reference voltage during input voltage variations.

This function is complicated by the zener diode impedance,

which effectively causes an incremental change in zener

breakdown voltage with changing zener current.

Figure 15. Impedance Cancellation with An

Uncompensated Zener

Z1TC1

−

It is possible, however, by employing a bridge type circuit

which includes the zener diode and current regulating

resistance in its branch legs, to effectively cancel the effect

of the zener impedance. Consider the circuit of Figure 15 as

an example. This is the common configuration for a zener

diode voltage regulating system. The zener impedance at 20

mA of a 1N4740 diode is typically 2 ohms. If the supply

voltage now changes from 30 V to 40 V, the diode current

determined by R1 changes from 20 to 30 mA; the average

zener impedance becomes 1.9 ohms; and the reference

voltage shifts by 19 mV. This represents a reference change

of .19%, an amount far too large for an input change of 30%

in most reference supplies.

The effect of zener impedance change with current is

relatively small for most input changes and will be neglected

for this analysis. Assuming constant zener impedance, the

zener voltage is approximated by

(37)

V′Z = VZ + Z(I′Z − IZ)

where V′Z is the new zener voltage

VZ is the former zener voltage

I′Z is the new zener current

IZ is the new zener current flowing at VZ

RZ is the zener impedance

Then

Let the input voltage VI in Figure 15 increase by an

amount ΔVI

Then ΔI = ΔVI − ΔVZ

ΔVZ

Also ΔI =

Solving ΔVIRZ − ΔVZRZ − ΔVZR1 = 0

ΔVZ

ΔVI

Or =RZ

R1 + RZ

(39)

(38)

(40)

ΔVZ = ZΔIZ

Equation 40 merely states that the change in reference

voltage with input tends to zero when the zener impedance

tends also to zero, as expected.

The figure of merit equation can be applied to the circuits

of Figure 16 and 17 to explain impedance cancellation. The

Change Factor equations for each leg and the reference

voltage VR are:

(41)

(43)

(42)

CFVZ = ΔVZ

ΔVI

R1 + RZ

== R

R2 + R3

ΔV2

ΔVI

CFV2 = = = RB

R1 + RZ

ΔVR

ΔVI

CFVR = = = RA − RB

R2 + R3

Figure 16. Standard Voltage

Regulation Circuit

VI′ ΔVI

Figure 17. Impedance Cancellation Bridge

VI′ ΔVI

R1R2

http://onsemi.com

Since the design is to minimize CFVR, RB can be set equal

to RA. The Input Regulation Factors are:

(44)

(46)

(45)

γVZ = ΔVZ

ΔVIǒVI

ViZǓ = 1

1 + VZ

VIǒVI

ViZǓ

γV2 = ΔV2

ΔVIǒVI

ViZǓ = 1

γVR = ΔVR

ΔVIǒVI

ViZǓ = 1

1 + ǒVI

ViZǓ

ǒVI

ViZǓ

VIǒVI

Vi ZǓ

1 −RB

It is seen that γVR can be minimized by setting RB = RA.

Note that it is not necessary to match R3 to RZ and R2 to

R1. Thus R3 and R2 can be large and hence dissipate low

power. This discussion is assuming very light load currents.

http://onsemi.com

ZENER PROTECTIVE CIRCUITS AND TECHNIQUES:

BASIC DESIGN CONSIDERATIONS

INTRODUCTION

The reliability of any system is a function of the ability of

the equipment to operate satisfactorily during moderate

changes of environment, and to protect itself during

otherwise damaging catastrophic changes. The silicon zener

diode offers a convenient, simple but effective means of

achieving this result. Its precise voltage sensitive

breakdown characteristic provides an accurate limiting

element in the protective circuit. The extremely high

switching speed possible with the zener phenomenon allows

the circuit to react faster by orders of magnitude that

comparable mechanical and magnetic systems.

By shunting a component, circuit, or system with a zener

diode, the applied voltage cannot exceed that of the

particular device’s breakdown voltage. (See Figure 1.)

A device should be chosen so that its zener voltage is

somewhat higher than the nominal operating voltage but

lower than the value of voltage that would be damaging if

allowed to pass. In order to adequately incorporate the zener

diode for circuit protection, the designer must consider

several factors in addition to the required zener voltage. The

first thing the designer should know is just what transient

characteristics can be anticipated, such as magnitude,

duration, and the rate of reoccurrence. For short duration

transients, it is usually possible to suppress the voltage spike

and allow the zener to shunt the transient current away from

the load without a circuit shutdown. On the other hand, if the

over-voltage condition is for a long duration, the protective

circuit may need to be complimented with a disconnect

element to protect the zener from damage created by

excessive heating. In all cases, the end circuit will have to be

designed around the junction temperature limits of the

device.

The following sections illustrate the most common zener

protective circuits, and will demonstrate the criteria to be

followed for an adequate design.

BASIC PROTECTIVE CIRCUITS

FOR SUPPLY TRANSIENTS

The simple zener shunt protection circuit shown in

Figure 1 is widely used for supply voltage transient

protection where the duration is relatively short. The circuit

applies whether the load is an individual component or a

complete circuit requiring protection. Whenever the input

exceeds the zener voltage, the device avalanches into

conduction clamping the load voltage to VZ. The total

current the diode must carry is determined by the magnitude

of the input voltage transient and the total circuit impedance

minus the load current. The worst case occurs when load

current is zero and may be expressed as follows:

IZ(max) = VI(max) − VZ

(1)

Figure 1. Basic Shunt Zener Transient

Protection Circuit

LOAD

POWER

SUPPLY

−

The maximum power dissipated by the zener is

PZ(max) =I

Z(max) VZ(max)=VI(max) − VZ

(2)VZ(max)

Also, more than one device can be used, i.e., a series

string, which will reduce the percentage of total power to be

dissipated per device by a factor equal to the number of

devices in series. The number of diodes required can be

found from the following expression:

Number = PZ(max)

PZ (allowable per device)

(3)

Any fraction of a zener must be taken as the next highest

whole number. This design discussion has been based upon

the assumption that the transient is of a single shot,

non-recurrent type. For all practical purposes it can be

considered non-recurrent if the “off period” between

transients is at least four times the thermal time constant of

the device. If the “off period” is shorter than this, then the

design calculations must include a factor for the duty cycle.

This is discussed in detail in Chapter 4. In Chapter 4 there are

also some typical curves relating peak power, pulse duration

and duty cycle that may be appropriate for some designs.

Obviously, the factor that limits the feasibility of the basic

zener shunt protective circuit is the pulse durations “t”. As

the duration increases, the allowable peak power for a given

configuration decreases and will approach a steady state

condition.

http://onsemi.com

Figure 2. Overvoltage Protection with Zener Diodes and Fuses

LOAD

RSFUSE

POWER

SUPPLY

−

When the anticipated transients expected to prevail for a

specific situation are of long duration, a basic zener shunt

becomes impractical, in such a case the circuit can be

improved by using a complementary disconnect element.

The most common overload protective element is without a

doubt the standard fuse. The common fuse adequately

protects circuit components from over-voltage surges, but at

the same time must be chosen to eliminate “nuisance fusing”

which results when the maximum current rating of the fuse

is too close to the normal operational current of the circuit.

AN EXAMPLE PROBLEM: SELECTING A

FUSE-ZENER COMBINATION

Consider the case illustrated in Figure 2. Here the load

components are represented by a parallel combination of R

and C, equivalent to many loads found in practice. The

maximum capacitor voltage rating is usually the

circuit-voltage limiting factor due to the cost of high voltage

capacitors. Consequently, a protective circuit must be

designed to prevent voltage surges greater than 1.5 times

normal working voltage of the capacitor. It is common,

however, for the supply voltage to increase to 135% normal

for long periods. Examination of fuse manufacturers’

melting time-current curves shows the difficulty of trying to

select a fuse which will melt rapidly at overload (within one

or two cycles of the supply frequency to prevent capacitor

damage), and will not melt when subjected to voltages close

to overload for prolonged periods.

By connecting a zener diode of correct voltage ratings

across the load as shown, a fuse large enough to withstand

normal current increases for long periods may be chosen.

The sudden current increase when zener breakdown occurs

melts the fuse rapidly and protects the load from large

surges. In Figure 3, fuse current was plotted against supply

voltage to illustrate the improvement in load protection

obtained with zener-fuse combinations. Fuse current “A”

would be selected to limit current resulting from voltage

surges above 112 V to 90 mA, which would melt the fuse in

100 ms. It is a simple matter, however, to select a fuse which

melts in 30 ms at 200 mA but is unaffected by 100 mA

currents. The zener connection allows fuse current “B” to be

selected, eliminating this design problem and providing a

faster, more reliable protective circuit. If the same fuse was

used without the zener diode, a supply voltage of 210 volts

would be reached before the fuse would begin to protect the

load.

VS, SUPPLY VOLTAGE (VOLTS)

60 70 80 90 100 110 120 130 140

100

120

140

160

180

200

220

A"

B"

FUSE CURRENT (mA)

Figure 3. Fuse Current versus

Supply Voltage

RESISTIVE

LOAD ONLY

ZENER DIODE WITH,

RESISTIVE LOAD

ZENER BREAKDOWN,

VOLTAGE

NORMAL LOAD VOLTAGE

Selection of the correct power rating of zener diodes to be

used for surge protection depends upon the magnitude and

duration of anticipated surges. Often in circuits employing

both fuses and zener diodes, the limiting surge duration will

be the melting time of the fuse. This, in turn, depends on the

nature of the load protected and the length of time it will

tolerate an overload.

As a first solution to the example problem, consider a

zener diode with a nominal breakdown voltage of 110 volts

measured at a test current (IZT) of 110 mA. Since the fuse

requires about 200 mA to melt and 100 mA are drawn

through the load at this voltage, the load voltage will never

exceed the zener breakdown voltage on slowly rising inputs.

Transients producing currents of approximately 200 mA but

of shorter duration than 30 ms will simply be clipped by

zener action and diverted from the load. Transients of very

high voltage will produce larger currents and, hence, will

melt the fuse more rapidly. In the limiting case where

transient power might eventually destroy the zener diode,

the fuse always melts first because of the slower thermal

time constant inherent in the zener diode’s larger geometry.

The curves in Figure 4 illustrate the change in zener

voltage as a function of changing current for a typical device

type.

http://onsemi.com

LOG IZ/IZK

CHANGE, IN VZ

VZ = V1

Figure 4. Change in VZ for Changes in IZ

If an actual curve for the device being used is not

available, the zener voltage at a specific current above or

below the test current may be approximated by equation 4.

Where: V = VZ + ZZT (I−IZT)

VZ = zener voltage at test current IZT

ZZT = zener impedance at test current IZT

IZT = test current

V = zener voltage at current I

(4)

For a given design, the maximum zener voltage to expect

for the higher zener current should be determined to make

sure the limits of the circuit are met. If the maximum limit

is excessive for the original device selection, the next lower

voltage rating should be used.

The previous discussion on design consideration for

protective circuits incorporating fuses is applicable to any

protective element that permanently disconnects the supply

when actuated. Rather than a fuse, a non-resetting magnetic

circuit breaker could have been used, and the same

reasoning would have applied.

LOAD CURRENT SURGES

In many actual problems the designer must choose a

protective circuit to perform still another task. Not only must

the equipment be protected from the voltage surges in the

supply, but the supply itself often requires protection from

shorts or partial shorts in the load. A direct short in the load

is fairly easy to handle, as the drastic current change permits

the use of fuses with ratings high enough to avoid problems

with supply surges. More common is the partial short, as

illustrated in Figure 5. If a short circuit occurs in the

capacitive section of the load (represented by C) the

resulting fault current is limited by the resistive section

(represented by R) to a value which may not be great enough

to melt the fuse. The fault current could be sufficient,

however, to damage the supply and other components in the

load.

The problem is resolved by employing a zener diode to

protect against supply surges as described in the previous

section, and by selecting a separate fuse to protect from load

faults. The load fuse in Figure 5 is chosen close to the normal

operating current. Abnormal supply surges do not affect it

and equipment operates reliably but with ample protection

for the supply against load changes.

ZENER DIODES AND RECLOSING

DISCONNECT ELEMENTS

An interesting application of zener diodes as overvoltage

protectors, which offers the possibility of designing for both

long and short duration surges, is shown in Figure 6.

Figure 6. Zener Diode Reclosing Circuit Breaker

Protective Circuit

LOAD

POWER

SUPPLY

RECLOSING

CIRCUIT

BREAKER

In the event of a voltage overload exceeding a chosen

zener voltage, a large current will be drawn through the

diode. The reclosing disconnect element opens after an

interval determined by its time constant, and the supply is

disconnected. After another interval, again depending on the

switch characteristics, the supply is reconnected and the

voltage “sampled” by the zener diode. This leads to an

“on-off” action which continues until the supply voltage

drops below the predetermined limit. At no time can the load

voltage or current exceed that set by the zener. The chief

advantage in this type of circuit is the elimination of fuse

replacement in similar fusing circuits, while providing

essentially the same load protection.

Figure 5. Supply and Load with Zener Diode; Fuse Circuitry

POWER

SUPPLY

RSUPPLY

FUSE

LOAD

FUSE R

LOAD

http://onsemi.com

Figure 7. (Typical) Voltage, Current and Temperature Waveforms

for a Thermal Breaker

TIME

VOLTS

SUPPLY

VOLTAGE

°C

THERMAL

BREAKER

TEMPERATURE

AMPS

ZENER

DIODE

CURRENT

ZENER

DIODE

JUNCTION

TEMPERATURE

SURGE VOLTAGE

OVER VOLTAGE

NORMAL OPERATING VOLTAGE

BREAK TEMPERATURE

MAKE

TEMPERATURE

MAXIMUM TJ

°C

It is difficult to define a set design procedure in this case,

because of the wide variety of reclosing, magnetic and

thermal circuit breakers available. Care should be taken to

ensure that the power dissipated in the zener diode during the

conduction time of the disconnect element does not exceed

its rating. As an example, assume the disconnect element

was a thermal breaker switch. The waveforms for a typical

over-voltage situation are shown in Figure 7.

It is apparent that the highest zener diode junction

temperature is reached during the first conduction period. At

this time the thermal breaker is cold and requires the greatest

time to reach its break temperature. The breaker then cycles

thermally between the make and break temperatures as long

as the supply voltage is greater than the zener voltage, as

shown in Figure 7.

The zener diode current and junction temperature

variation are shown in the last two waveforms of Figure 7.

Overvoltage durations longer than the trip time of the

thermal breaker do not affect the diode as the supply is

disconnected. An overvoltage of much higher level simply

causes the thermal breaker to open sooner. In effect, the

zener diode rating must be high enough to ensure that

maximum junction temperature is not reached during the

longest interval that the thermal switch will be closed.

Manufacturers of thermally operated circuit breakers

publish current-time curves for their devices similar to that

shown in Figure 8. By estimating the peak supply

overvoltage and determining the maximum overvoltage

tolerated by the load, an estimation of peak zener current can

be made. The maximum breaker trip time may then be read

from Figure 8. (After the initial current surge, the duration

of “of” time is determined entirely by the breaker

characteristics and will vary widely with manufacture.) The

zener diode junction temperature rise during conduction

may be calculated now from the thermal time constant of the

device and the heatsink used.

Because the reclosing circuit breaker is continually

cycling on and off, the zener current takes on the

characteristics of a repetitive surge, as can be seen in

Figure 7.

http://onsemi.com

001 23

CURRENT (AMPS)

TRIP TIME (SECONDS)

Figure 8. Trip Time versus Current for

Thermal Breaker

TRANSISTOR OVERVOLTAGE PROTECTION

In many electronic circuits employing transistors, high

internal voltages can be developed and, if applied to the

transistors, will destroy them. This situation is quite

common in transistor circuits that are switching highly

inductive loads. A prime example of this would be in

transistorized electronic ignition systems such as shown in

Figures 9a and 9b.

The zener diode is an important component to assure solid

state ignition system reliability. There are two basic methods

of using a zener diode to protect an ignition transistor. These

are shown in Figures 9a and 9b. In Figure 9b the transistor

is protected by a zener diode connected between base and

collector and in Figure 9a, the zener is connected between

emitter and collector. In both cases the voltage level of the

zener must be selected carefully so that the voltage stress on

the transistor is in a region where the safe operating area is

adequate for reliable circuit operation.

Figure 10 illustrates “safe” and “unsafe” selection of a

zener diode for collector-base protection of a transistor in the

ignition coil circuit. It can be seen that the safe operating area

of a transistor must be known if an adequate protective zener

is to be selected.

The zener diode must be able to take the stress of peak

pulse current necessary to clamp the voltage rise across the

transistor to a safe value. In a typical case, a 5 watt, 100 volt

zener transient suppressor diode is required to operate with

an 80 μs peak pulse current of 8 amperes when connected

between the collector-emitter of the transistor. The

waveform of this pulse current approaches a sine wave in

shape (Figure 11). The voltage rise across a typical transient

suppressor diode due to this current pulse is shown in Figure

12. This voltage rise of approximately 8 volts indicates an

effective zener impedance of approximately 1 ohm.

However, a good share of this voltage rise is due to the

temperature coefficient and thermal time constant of the

zener. The temperature rise of the zener diode junction is

indicated by the voltage difference between the rise and fall

of the current pulse.

Figure 9. Transistor Ignition Systems with Zener

Overvoltage Surge Protection

10 Ω

1/2 W

1 Ω

10 Ω

10 W

1 Ω

100 W

560 pF

2 μF

5 Ω

1N5374B

1N6295

10 Ω

2N5879

+12 V

H.V. TO

DIST.

PRESTO-LITE

201

MALLORY

COIL

28100

H.V. TO

DIST.

200 V

PAPER

2N6031

+12 V

(A) (B)

http://onsemi.com

COLLECTOR-BASE

ZENER CLAMP

SAFE

UNSAFE

10 30 40 50 60 70 80 90 100 110 120

VCE

COLLECTOR-

EMITTER

ZENER

CLAMP

TYPICAL TRANSISTOR SAFE AREA LIMIT

Figure 10. Safe Zener Protection

LOAD

LINE

SAFE

TIME 10 μs/div

2 A/div ZENER CURRENT

Figure 11. Zener Diode Current Pulse

100 V

ZENER VOLTAGE 1 V/div

1 A/div ZENER CURRENT

Figure 12. Voltage-Current Representation on

100 V Zener

In order to assure safe operation, the change in zener

junction temperature for the peak pulse conditions must be

analyzed. In making the calculation, the method described

in Chapter 3 should be used, taking into account duty cycle,

pulse duration, and pulse magnitude.

When the zener diode is connected between the collector

and emitter of the transistor, additional power dissipation

will result from the clipping of the ringing voltage of the

ignition coil by the forward conduction of the zener diode.

This power dissipation by the forward diode current will

result in additional zener voltage rise. It is not uncommon to

observe a 15-volt rise above the zener device voltage rating

due to temperature coefficient and impedance under these

pulse current conditions.

The zener diode should be connected as close as possible

to the terminals of the transistor the zener is intended to

protect. This insures that induced voltage transients, caused

by current changes in long lead lengths, are clamped by the

zener and do not appear across the transistor.

http://onsemi.com

Figure 13. DC-DC Converter with Surge Protecting Diodes

Another example of overvoltage protection of transistor

operating in an inductive load switch capacity is illustrated

in Figure 13. The DC-DC converter circuit shows a

connection from collector to emitter of two zener diodes as

collector overvoltage protectors. Without some type of

limiting device, large voltage spikes may appear at the

collectors, due to the switching transients produced with

normal circuit operation. When this spike exceeds the

collector breakdown rating of the transistor, transistor life is

considerably shortened. The zener diodes shown are chosen

with zener breakdowns slightly below transistor breakdown

voltage to provide the necessary clipping action. Since the

spikes are normally of short duration (0.5 to 5 μs) and duty

cycle is low, normal chassis mounting provides adequate

heatsinking.

METER PROTECTION

The silicon zener diode can be employed to prevent

overloading sensitive meter movements used in low range

DC and AC voltmeters, without adversely affecting the

meter linearity. The zener diode has the advantage over

thermal protective devices of instantaneous action and, of

course, will function repeatedly for an indefinite time (as

compared to the reset time necessary with thermal devices).

While zener protection is presently available for voltages as

low as 2.4 volts, forward diode operation can be used for

meter protection where the voltage drop is much smaller. A

typical protective circuit is illustrated in Figure 14. Here the

meter movement requires 100 μAmps for full scale

deflection and has 940 ohms resistance. For use in a

voltmeter to measure 25 V, approximately 249 thousand

ohms are required in series.

Figure 14. Meter Protection with Zener Diode

70K

179

1N4746

(18 VOLT

ZENER DIODE)

−

25 V

μA

The protection provided by the addition of an 18 volt zener

is illustrated in Figure 15. With an applied voltage of

25 volts, the 100 μAmps current in the circuit produces a

drop of 17.9 volts across the series resistance of 179

thousand ohms. A further increase in voltage causes the

zener diode to conduct, and the overload current is shunted

away from the meter. Since ON Semiconductor zener diodes

have zener voltages specified within 5 and 10%, a safe

design may always be made with little sacrifice in meter

linearity by assuming the lowest breakdown voltage within

the tolerance. The shunting effect on the meter of the reverse

biased diode is generally negligible below breakdown

voltage (on the order of 0.5° full scale). For very precise

work, the zener diode breakdown voltage must be accurately

known and the design equations solved for the correct

resistance values.

http://onsemi.com

200

0 50 100 150 200

TOTAL CURRENT (μA)

150

100

METER CURRENT ( A)

WITH PROTECTIVE CIRCUIT

Figure 15. Meter Protection with

Zener Diodes

ZENER DIODES USED WITH SCRS FOR

CIRCUIT PROTECTION

An interesting aspect of circuit protection incorporating

the reliable zener diode is the protective circuits shown in

Figures 16 and 17.

In a system that is handling large amounts of power, it may

become impractical to employ standard zener shunt

protection because of the large current it would be required

to carry. The SCR crowbar technique shown in Figure 16 can

be effectively used in these situations. The zener diode is still

the transient detection component, but it is only required to

carry the gate current for SCR turn on, and the SCR will

carry the bulk of the shunt current. Whenever the incoming

voltage exceeds the zener voltage, it avalanches, supplying

gate drive to the SCR which, when fired, causes a current

demand that will trip the circuit breaker. The resistors shown

are for current limiting so that the SCR and zener ratings are

not exceeded.

The circuit of Figure 17 is designed to disconnect the

supply in the event a specified load current is exceeded. This

is done by means of a series sense resistor and a compatible

zener to turn the shunt SCR on. When the voltage across the

series resistor, which is a function of the load current,

becomes sufficient to break over the zener, the SCR is fired,

causing the circuit breaker to trip.

Figure 16. SCR Crowbar Over-Voltage Protection Circuit for AC Circuit Operation

Figure 17. SCR Longterm Current Overload Protection

ZENER

SCR SCR

CIRCUIT

BREAKER

CIRCUIT

BREAKER

http://onsemi.com

ZENER TRANSIENT SUPPRESSORS

The transient suppressor is used as a shunt element in

exactly the same manner as a conventional zener. It offers

the same advantages such as low insertion loss, immediate

recovery after operation, a clamping factor approaching

unity, protection against fast rising transients, and simple

circuitry. The primary difference is that the transient

suppressor extends these advantages to higher power levels.

Even in the event of transients with power contents far in

excess of the capacity of the zeners, protection is still

provided the load. When overloaded to failure, the zener will

approximate a short. The resulting heavy drain will aid in

opening the fuse or circuit breaker protecting the load

against excess current. Thus, even if the suppressor is

destroyed, it still protects the load.

The design of the suppressor-fuse combination for the

required level of protection follows the techniques for

conventional zeners discussed earlier in this chapter.

TRANSIENT SUPPRESSION CHARACTERISTICS

Zener diodes, being nearly ideal clippers (that is, they

exhibit close to an infinite impedance below the clipping

level and close to a short circuit above the clipping level), are

often used to suppress transients. In this type of application,

it is important to know the power capability of the zener for

short pulse durations, since they are intolerant of excessive

stress.

Some ON Semiconductor data sheets such as the ones for

devices shown in Table 1 contain short pulse surge

capability. However, there are many data sheets that do not

contain this data and Figure 18 is presented here to

supplement this information.

Table 1. Transient Suppressor Diodes

Series

Numbers

Steady State

Power Package Description

1N4728A 1 W DO-41 Double Slug

Glass

1N6267A 5 W Case 41A Axial Lead

Plastic

1N5333B 5 W Case 102 Surmetic 40

1N746A/957B

/4370A

500 mW DO-35 Double Slug

Glass

1N5221B 500 mW DO-35 Double Slug

Glass

Some data sheets have surge information which differs

slightly from the data shown in Figure 18. A variety of

reasons exist for this:

1. The surge data may be presented in terms of actual

surge power instead of nominal power.

2. Product improvements have occurred since the data

sheet was published.

3. Large dice are used, or special tests are imposed on the

product to guarantee higher ratings than those shown

in Figure 18.

4. The specifications may be based on a JEDEC

registration or part number of another manufacturer.

The data of Figure 18 applies for non-repetitive conditions

and at a lead temperature of 25°C. If the duty cycle increases,

the peak power must be reduced as indicated by the curves

of Figure 19. Average power must be derated as the lead or

ambient temperature rises above 25°C. The average power

derating curve normally given on data sheets may be

normalized and used for this purpose.

100

0.01 0.02 0.05 0.1 0.2 0.5 1 2 5 10

PULSE WIDTH (ms)

0.5

0.2

0.1

0.05

0.02

0.01

PK(nom), NOMINAL PEAK POWER (kW)

Figure 18. Peak Power Ratings of

Zener Diodes

1N6267 SERIES

5 WATT TYPES

250 mW TO 1 W TYPES

GLASS DO-35 & GLASS DO-41

1 TO 3 W TYPES

PLASTIC DO-41

0.1 0.2 0.5 1 25

10 5020 100

D, DUTY CYCLE (%)

0.7

0.5

0.3

0.2

0.1

DERATING FACTOR

0.07

0.05

0.03

0.02

0.01

Figure 19. Typical Derating Factor for

Duty Cycle

PULSE WIDTH

10 ms

1 ms

100 μs

10 μs

When it is necessary to use a zener close to surge ratings,

and a standard part having guaranteed surge limits is not

suitable, a special part number may be created having a surge

limit as part of the specification. Contact your nearest

ON Semiconductor OEM sales office for capability, price,

delivery, and minimum order quantities.

http://onsemi.com

MATHEMATICAL MODEL

Since the power shown on the curves is not the actual

transient power measured, but is the product of the peak

current measured and the nominal zener voltage measured

at the current used for voltage classification, the peak current

can be calculated from:

IZ(PK) = P(PK)

VZ(nom)

(5)

The peak voltage at peak current can be calculated from:

(6)

VZ(PK) = FC x VZ(nom)

where FC is the clamping factor. The clamping factor is

approximately 1.20 for all zener diodes when operated at

their pulse power limits. For example, a 5 watt, 20 volt

zener can be expected to show a peak voltage of 24 volts

regardless of whether it is handling 450 watts for 0.1 ms

or 50 watts for 10 ms. This occurs because the voltage is

a function of junction temperature and IR drop. Heating

of the junction is more severe at the longer pulse width,

causing a higher voltage component due to temperature

which is roughly offset by the smaller IR voltage

component.

For modeling purposes, an approximation of the zener

resistance is needed. It is obtained from:

(7)

RZ(nom) = VZ(nom)(FC−1)

PPK(nom) /V

Z(nom)

The value is approximate because both the clamping

factor and the actual resistance are a function of temperature.

CIRCUIT CONSIDERATIONS

It is important that as much impedance as circuit

constraints allow be placed in series with the zener diode and

the components to be protected. The result will be a lower

clipping voltage and less zener stress. A capacitor in parallel

with the zener is also effective in reducing the stress imposed

by very short duration transients.

To illustrate use of the data, a common application will be

analyzed. The transistor in Figure 20 drives a 50 mH

solenoid which requires 5 amperes of current. Without some

means of clamping the voltage from the inductor when the

transistor turns off, it could be destroyed.

The means most often used to solve the problem is to

connect an ordinary rectifier diode across the coil; however,

this technique may keep the current circulating through the

coil for too long a time. Faster switching is achieved by

allowing the voltage to rise to a level above the supply before

being clamped. The voltage rating of the transistor is 60 V,

indicating that approximately a 50 volt zener will be

required.

The peak current will equal the on-state transistor current

(5 amperes) and will decay exponentially as determined by

the coil L/R time constant (neglecting the zener impedance).

A rectangular pulse of width L/R (0.01 s) and amplitude of

IPK (5 A) contains the same energy and may be used to select

a zener diode. The nominal zener power rating therefore

must exceed (5 A × 50) = 250 watts at 10 ms and a duty cycle

of 0.01/2 = 0.5%. From Figure 19, the duty cycle factor is

0.62 making the single pulse power rating required equal to

250/0.62 = 403 watts. From Figure 18, one of the 1N6267

series zeners has the required capability. The 1N6287 is

specified nominally at 47 volts and should prove

satisfactory.

Although this series has specified maximum voltage

limits, equation 7 will be used to determine the maximum

zener voltage in order to demonstrate its use.

RZ = 47(1.20 − 1)

500/47

9.4

10.64

= = 0.9 Ω

At 5 amperes, the peak voltage will be 4.5 volts above

nominal or 51.5 volts total which is safely below the 60

volt transistor rating.

Figure 20. Circuit Example

10 ms

2 s

50 mH, 5 Ω

26 Vdc

USED TO SELECT A ZENER DIODE HAVING THE PROPER

VOLTAGE AND POWER CAPABILITY TO PROTECT THE TRANSISTOR

http://onsemi.com

ZENER VOLTAGE SENSING CIRCUITS

AND APPLICATIONS

BASIC CONCEPTS OF VOLTAGE SENSING

Numerous electronic circuits require a signal or voltage

level to be sensed for circuit actuation, function control, or

circuit protection. The circuit may alter its mode of

operation whenever an interdependent signal reaches a

particular magnitude (either higher or lower than a specified

value). These sensing functions may be accomplished by

incorporating a voltage dependent device in the system

creating a switching action that controls the overall

operation of the circuit.

The zener diode is ideally suited for most sensing

applications because of its voltage dependent

characteristics. The following sections are some of the more

common applications and techniques that utilize the zener in

a voltage sensing capacity.

Figure 1. Basic Transistor-Zener Diode

Sensing Circuits

VIN

VIN Z1Q1

VO1

VO2

(a)

(b)

TRANSISTOR-ZENER SENSING CIRCUITS

The zener diode probably finds its greatest use in sensing

applications in conjunction with other semiconductor

devices. Two basic widely used techniques are illustrated in

Figures 1a and 1b.

In both of these circuits the output is a function of the input

voltage level. As the input goes from low to high, the output

will switch from either high to low (base sense circuit) or

low to high (emitter sense circuit), (see Figure 2).

The base sense circuit of Figure 1a operates as follows:

When the input voltage is low, the voltage dropped across R2

is not sufficient to bias the zener diode and base emitter

junction into conduction, therefore, the transistor will not

conduct. This causes a high voltage from collector to

emitter. When the input becomes high, the zener is biased

into conduction, the transistor turns on, and the collector to

emitter voltage, which is the output, drops to a low value.

Figure 2. Outputs of Transistor-Zener Voltage

Sensing Circuits

R2 + R1

VZ + VBE(SAT)

VIN =

SENSING LEVEL

TIME

VIN BOTH CIRCUITS

VOUT BASE SENSE

VOUT EMITTER SENSE

VO2

VO1

VIN

(OUTPUT OF FIGURE 1b)

(OUTPUT OF FIGURE 1a)

http://onsemi.com

The emitter sense circuit of Figure 1b operates as follows:

When the input is low the voltage drop across R3 (the output)

is negligible. As the input voltage increases the voltage drop

across R2 biases the zener into conduction and forward

biases the base-emitter junction. A large voltage drop across

R3 (the output voltage) is equal to the product of the collector

current times the resistance, R3. The following relationships

indicate the basic operating conditions for the circuits in

Figure 1.

Circuit Output

High

VOUT = VIN − ICOR3 ≅ VIN

Low

VOUT = VIN − ICR3 = VCE(sat)

Low

VOUT = VIN − VZ − VCE(off) = ICOR3

High

VOUT = VIN − VCE(sat) = ICR3

In addition, the basic circuits of Figure 1 can be rearranged

to provide inverse output.

AUTOMOTIVE ALTERNATOR VOLTAGE

REGULATOR

Electromechanical devices have been employed for many

years as voltage regulators, however, the regulation setting

of these devices tend to change and have mechanical contact

problems. A solid state regulator that controls the charge rate

by sensing the battery voltage is inherently more accurate

and reliable. A schematic of a simplified solid state voltage

regulator is shown in Figure 3.

The purpose of an alternator regulator is to control the

battery charging current from the alternator. The charge

level of the battery is proportional to the battery voltage

level. Consequently, the regulator must monitor the battery

voltage level allowing charging current to pass when the

battery voltage is low. When the battery has attained the

proper charge the charging current is switched off. In the

case of the solid state regulator of Figure 3, the charging

current is controlled by switching the alternator field current

on and off with a series transistor switch, Q2. The switching

action of Q2 is controlled by a voltage sensing circuit that is

identical to the base sense circuit of Figure 1a. When

under-charged, the zener Z1 does not conduct keeping Q1

off. The collector-emitter voltage of Q1 supplies a forward

bias to the base-emitter of Q2, turning it on. With Q2 turned

on, the alternator field is energized allowing a charging

current to be delivered to the battery. When the battery

attains a proper charge level, the zener conducts causing Q1

to turn on, and effectively shorting out the base-emitter

junction of Q2. This short circuit cuts off Q2, turns off the

current flowing in the field coil which consequently, reduces

the output of the alternator. Diode D1 acts as a field

suppressor preventing the build up of a high induced voltage

across the coil when the coil current is interrupted.

Figure 3. Simplified Solid State Voltage Regulator

B+

Z1Q1

ALTERNATOR

OUTPUT

D1ALTERNATOR

FIELD

http://onsemi.com

In actual operation, this switching action occurs many

times each second, depending upon the current drain from

the battery. The battery charge, therefore, remains

essentially constant and at the maximum value for optimum

operation.

A schematic of a complete alternator voltage regulator is

shown in Figure 4.

It is also possible to perform the alternator regulation

function with the sensing element in the emitter of the

control transistor as shown in Figure 5.

In this configuration, the sensing circuit is composed of Z1

and Q1 with biasing components. It is similar to the sensing

circuit shown in Figure 1b. The potentiometer R1 adjusts the

conduction point of Q1 establishing the proper charge level.

When the battery has reached the desired level, Q1 begins to

conduct. This draws Q2 into conduction, and therefore

shorts off Q3 which is supplying power to the alternator

field. This type of regulator offers greater sensitivity with an

increase in cost.

Figure 4. Complete Solid State Alternator Voltage Regulator

Figure 5. Alternator Regulator with Emitter Sensor

B+

100 Ω

15 Ω

30 Ω

70 Ω

1N961B SENSING

ZENER

DIODE

RT*

THER-

MISTOR

0.05 μF

FEEDBACK

CAPACITOR

1N4001

FIELD

SUPPRESSION

DIODE

TO ALTERNATOR

FIELD COIL

2N5879

1N3493

BIAS

DIODE

ALTERNATOR

OUTPUT

0.05 Ω

2N4234

*THE VALUE OF RT DEPENDS ON THE SLOPE OF THE VOLTAGE REGULATION

VERSUS TEMPERATURE CURVE.

B+ ALTERNATOR

OUTPUT

R2ALTERNATOR

FIELD

http://onsemi.com

UNIJUNCTION-ZENER SENSE CIRCUITS

Unijunction transistor oscillator circuits can be made

GO-NO GO voltage sensitive by incorporating a zener diode

clamp. The UJT operates on the criterion: under proper

biasing conditions the emitter-base one junction will

breakover when the emitter voltage reaches a specific value

given by the equation:

Vp = ηVBB + VD(1)

where:

VBB

= peak point emitter voltage

= intrinsic stand-off ratio for the device

= interbase voltage, from base two to base one

= emitter to base one diode forward junction dro

Obviously, if we provide a voltage clamp in the circuit

such that the conditions of equation 1 are met only with

restriction on the input, the circuit becomes voltage

sensitive. There are two basic techniques used in clamping

UJT relaxation oscillators. They are shown in Figure 6 and

Figure 7.

The circuit in Figure 6 is that of a clamped emitter type.

As long as the input voltage VIN is low enough so that Vp

does not exceed the Zener voltage VZ, the circuit will

generate output pulses. At some given point, the required Vp

for triggering will exceed VZ. Since Vp is clamped at VZ, the

circuit will not oscillate. This, in essence, means the circuit

is GO as long as VIN is below a certain level, and NO GO

above the critical clamp point.

The circuit of Figure 7, is a clamped base UJT oscillator.

In this circuit VBB is clamped at a voltage VZ and the emitter

tied to a voltage dividing network by a diode D1. When the

input voltage is low, the voltage drop across R2 is less than

Vp. The forward biased diode holds the emitter below the

trigger level. As the input increases, the R2 voltage drop

approaches Vp. The diode D1 becomes reversed biased and,

the UJT triggers. This phenomenon establishes the

operating criterion that the circuit is NO GO at a low input

and GO at an input higher than the clamp voltage. Therefore,

the circuits in Figures 6 and 7 are both input voltage

sensitive, but have opposite input requirements for a GO

condition. To illustrate the usefulness of the clamped UJT

relaxation oscillators, the following two sections show them

being used in practical applications.

Figure 6. UJT Oscillator, GO — NO GO Output,

GO for Low VIN — NO GO for High VIN

−

CZUJT

RB2

RB1

VP = ηVBB + VD

VOUT

VIN

VBB

Figure 7. UJT — NO GO Output, NO GO for Low VIN — GO for High VIN

−

RTRB2

RB1

VIN

VOUT

UJT

http://onsemi.com

BATTERY VOLTAGE SENSITIVE SCR CHARGER

A clamped emitter unijunction sensing circuit of the type

shown in Figure 6 makes a very good battery charger

(illustrated in Figure 8). This circuit will not operate until the

battery to be charged is properly connected to the charger.

The battery voltage controls the charger and will dictate its

operation. When the battery is properly charged, the charger

will cease operation.

The battery charging current is obtained through the

controlled rectifier. Triggering pulses for the controlled

rectifier are generated by unijunction transistor relaxation

oscillator (Figure 9). This oscillator is activated when the

battery voltage is low.

While operating, the oscillator will produce pulses in the

pulse transformer connected across the resistance, RGC

(RGC represents the gate-to-cathode resistance of the

controlled rectifier), at a frequency determined by the

resistance, capacitance, R.C. time delay circuit.

Since the base-to-base voltage on the unijunction

transistor is derived from the charging battery, it will

increase as the battery charges. The increase in base-to-base

voltage of the unijunction transistor causes its peak point

voltage (switching voltage) to increase. These waveforms

are sketched in Figure 9 (this voltage increase will tend to

change the pulse repetition rate, but this is not important).

Figure 8. 12 Volt Battery Charger Control

Figure 9. UJT Relaxation Oscillator Operation

RECTIFIED A.C.

VOLTAGE FROM

CHARGER

UJT

2R2

SCR

12 V

−

R1  3.9K, 1/2 W

R2  1K, POT.

R3  5.1K, 1/2 W

C1  .25 μf

Z1  1N753, 6.2 V

SCR  MCR3813

VB1 B2

BATTERY

CHARGING

BATTERY

CHARGED

TIME

VC1

VRGC

ZENER

VOLTAGE

UJT PEAK

POINT VOLTAGE

SCR

CONDUCTS

SCR

NONCONDUCTING

UJT B2

RGC

VBATT.

UJT  2N2646

T1  PR1, 30T, no. 22

SEC, 45T, NO. 22

CORE: FERROX CUBE

203F181-303

A+

http://onsemi.com

When the peak point voltage (switching voltage) of the

unijunction transistor exceeds the breakdown voltage of the

Zener diode, Z1, connected across the delay circuit

capacitor, C1, the unijunction transistor ceases to oscillate.

If the relaxation oscillator does not operate, the controlled

rectifier will not receive trigger pulses and will not conduct.

This indicates that the battery has attained its desired charge

as set by R2.

The unijunction cannot oscillate unless a voltage

somewhere between 3 volts and the cutoff setting is present

at the output terminals with polarity as indicated. Therefore,

the SCR cannot conduct under conditions of a short circuit,

an open circuit, or a reverse polarity connection to the

battery.

ALTERNATOR REGULATOR FOR

PERMANENT MAGNET FIELD

In alternator circuits such as those of an outboard engine,

the field may be composed of a permanent magnet. This

increases the problem of regulating the output by limiting

the control function to opening or shorting the output.

Because of the high reactance source of most alternators,

opening the output circuit will generally stress the bridge

rectifiers to a very high voltage level. It is, therefore,

apparent that the best control function would be shorting the

output of the alternator for regulation of the charge to the

battery.

Figure 10 shows a permanent magnet alternator regulator

designed to regulate a 15 ampere output. The two SCRs are

connected on the ac side of the bridge, and short out the

alternator when triggered by the unijunction voltage

sensitive triggering circuit. The sensing circuit is of the type

shown in Figure 7. The shorted output does not appreciably

increase the maximum output current level.

A single SCR could be designed into the dc side of the

bridge. However, the rapid turn-off requirement of this type

of circuit at high alternator speeds makes this circuit

impractical.

The unijunction circuit in Figure 10 will not oscillate until

the input voltage level reaches the voltage determined by the

intrinsic standoff ratio. The adjustable voltage divider will

calibrate the circuit. The series diode in the voltage divider

circuit will compensate for the emitter-base-one diode

temperature change, consequently, temperature

compensation is necessary only for the zener diode

temperature changes.

Due to the delay in charging the unijunction capacitor,

when the battery is disconnected the alternator voltage will

produce high stress voltage on all components before the

SCRs will be fired. The 1N971B Zener was included in the

circuit to provide a trigger pulse to the SCRs as soon as the

alternator output voltage level approaches 30 volts.

Figure 10. Permanent Magnet Field Alternator Regulator

SEC. 1

SEC. 2

PRI.

MCR

2304-2

ALT.

OUT MCR

2304-2

MDA2500

1N960B

200 Ω5K Ω27 Ω

0.1 μF

2N2646

1N971B

1N4001

200 Ω

27 Ω

BATTERY

CORE: ARNOLD no. 4T5340 D1 DD1

PRIMARY 125 TURNSAWG 36

SEC no. 1 125 TURNSAWG 36

SEC no. 2 125 TURNSAWG 36

TRIFILAR WOUND

http://onsemi.com

Figure 11. Zener-Resistor Voltage Sensitive Circuit

Figure 12. Improving Meter Resolution

−

BASE CIRCUIT

VIN VOUT VIN VOUT

(LEVEL DETECTION) (MAGNITUDE REDUCTION)

−

VZ = 20 V

VOLTMETER

10 V  FULL SCALE

TYPICAL OUTPUTS

VIN VOUT

VIN = 24 V − 28 V

ZENER-RESISTOR VOLTAGE SENSING

A simple but useful sense circuit can be made from just a

Zener diode and resistor such as shown in Figure 11.

Whenever the applied signal exceeds the specific Zener

voltage VZ, the difference appears across the dropping

resistor R. This level dependent differential voltage can be

used for level detection, magnitude reduction, wave

shaping, etc. An illustrative application of the simple series

Zener sensor is shown in Figure 12, where the resistor drop

is monitored with a voltmeter.

If, for example, the input is variable from 24 to 28 volts,

a 30 voltmeter would normally be required. Unfortunately,

a 4 volt range of values on a 30 volt scale utilizes only 13.3%

of the meter movement — greatly limiting the accuracy with

which the meter can be read. By employing a 20 volt zener,

one can use a 10 voltmeter instead of the 30 volt unit, thereby

utilizing 40% of the meter movement instead of 13.3% with

a corresponding increase in accuracy and readability. For

ultimate accuracy a 24 volt zener could be combined with a

5 voltmeter. This combination would have the disadvantage

of providing little room for voltage fluctuations, however.

In Figure 13, a number of sequentially higher-voltage

Zener sense circuits are cascaded to actuate transistor

switches. As each goes into avalanche its respective

switching transistor is turned on, actuating the indicator

light for that particular voltage level. This technique can be

expanded and modified to use the zener sensors to actuate

some form of logic system.

Figure 13. Sequential Voltage Level Indicator

Q1Q2Q3

R1R2R3

RE2

RE1 RE3

Z2Z3

LIGHT

(1)

LIGHT

(2)

LIGHT

(3)

INPUT OUTPUT

http://onsemi.com

MISCELLANEOUS APPLICATIONS

OF ZENER TYPE DEVICES

INTRODUCTION

Many of the commonly used applications of zener diodes

have been illustrated in some depth in the preceding

chapters. This chapter shows how a zener diode may be used

in some rarer applications such as voltage translators, to

provide constant current, wave shaping, frequency control

and synchronized SCR triggers.

The circuits used in this chapter are not intended as

finished designs since only a few component values are

given. The intent is to show some general broad ideas and

not specific designs aimed at a narrow use.

FREQUENCY REGULATION OF A

DC TO AC INVERTER

Zener diodes are often used in control circuits, usually to

control the magnitude of the output voltage or current. In this

unusual application, however, the zener is used to control the

output frequency of a current feedback inverter. The circuit

is shown in Figure 1.

Figure 1. Frequency Controlled Current

Feedback Inverter

−

LOAD

The transformer T1 functions as a current transformer

providing base current IB = (NC/NB)IC. Without the zener

diode, the voltage across NB windings of the timing

transformer T1 is clamped to VBE of the ON device, giving

an inverter frequency of

f = VBE x 108

4BS1A1NB

where BS1A1 is the flux capacity of T1 transformer core.

The effect on output frequency of VBE variations due to

changing load or temperature can be reduced by using a

zener diode in series with VBE as shown in Figure 1. For

this circuit, the output frequency is given by

f = (VBE + VZ) x 108

4BS1A1NB

If VBE is small compared to the zener voltage VZ, good

frequency accuracy is possible. For example, with VZ =

9.1 volts, a 40 Watt inverter using 2N3791 transistors

(operating from a 12 volt supply), exhibited frequency

regulation of ±2% with ±25% load variation.

Care should be taken not to exceed V(BR)EBO of the

non-conducting transistor, since the reverse emitter-base

voltage will be twice the introduced series voltage, plus VBE

of the conducting device.

Transformer T2 should not saturate at the lowest inverter

frequency.

Inverter starting is facilitated by placing a resistor from

point A to B1 or a capacitor from A to B2.

SIMPLE SQUARE WAVE GENERATOR

The zener diode is widely used in wave shaping circuits;

one of its best known applications is a simple square wave

generator. In this application, the zener clips sinusoidal

waves producing a square wave such as shown in Figure 2a.

In order to generate a wave with reasonably vertical sides,

the ac voltage must be considerably higher than the zener

voltage.

Clipper diodes with opposing junctions built into the

device are ideal for applications of the type shown in

Figure 2b.

http://onsemi.com

(a) Single Zener Diode Square Wave Generator

(b) Opposed Zener Diodes Square Wave Generator

Figure 2. Zener Diode Square Wave Generator

ZENER

VOLTAGE

FORWARD

DROP

VOLTAGE

ZENER Z1

VOLTAGE

ZENER Z2

VOLTAGE

A.C. INPUT OUTPUT

http://onsemi.com

TRANSIENT VOLTAGE SUPPRESSION

INTRODUCTION

Electrical transients in the form of voltage surges have

always existed in electrical distribution systems, and prior to

the implementation of semiconductor devices, they were of

minor concern. The vulnerability of semiconductors to

lightning strikes was first studied by Bell Laboratories in

1961.1 A later report tried to quantify the amount of energy

certain semiconductors could absorb before they suffered

latent or catastrophic damage from electrostatic discharge.2

Despite these early warnings, industry did not begin to

address the issue satisfactorily until the late 1970s. Listed

below are the seven major sources of overvoltage.

•Lightning

•Sunspots

•Switching of Loads in Power Circuits

•Electrostatic Discharge

•Nuclear Electromagnetic Pulses

•Microwave Radiation

•Power Cross

Most electrical and electronic devices can be damaged by

voltage transients. The difference between them is the

amount of energy they can absorb before damage occurs.

Because many modern semiconductor devices, such as

small signal transistors and integrated circuits can be

damaged by disturbances that exceed the voltage ratings at

only 20 volts or so, their survivability is poor in unprotected

environments.

In many cases, as semiconductors have evolved their

ruggedness has diminished. The trend to produce smaller

and faster devices, and the advent of MOSFET and gallium

arsenide FET technologies has led to an increased

vulnerability. High impedance inputs and small junction

sizes limit the ability of these devices to absorb energy and

to conduct large currents. It is necessary, therefore, to

supplement vulnerable electronic components with devices

specially designed to cope with these hazards. Listed below

are the four primary philosophies for protecting against

transients.

•Clamping, or “clipping” is a method of limiting the

amplitude of the transient.

•Shunting provides a harmless path for the transient,

usually to ground by way of an avalanche or a crowbar

mechanism.

•Interrupting opens the circuit for the duration of the

transient.

•Isolating provides a transient barrier between hostile

environments and vulnerable circuits through the use of

transformers or optoisolators.

Selection of the proper protective method should be made

based upon a thorough investigation of the potential sources

of the overvoltage hazard. Different applications and

environments present different sources of overvoltage.

LIGHTNING

At any given time there are about 1800 thunderstorms in

progress around the world, with lightning striking about 100

times each second. In the U.S., lightning kills about 150

people each year and injures another 250. In flat terrain with

an average lightning frequency, each 300 foot structure will

be hit, on average, once per year. Each 1200 foot structure,

such as a radio or TV tower, will be hit 20 times each year,

with strikes typically generating 600 million volts.

Each cloud-to-ground lightning flash really contains from

three to five distinct strokes occurring at 60 ms intervals,

with a peak current of some 20,000 amps for the first stroke

and about half that for subsequent strokes. The final stroke

may be followed by a continuing current of around 150 amps

lasting for 100 ms.

The rise time of these strokes has been measured at around

200 nanoseconds or faster. It is easy to see that the

combination of 20,000 amps and 200 ns calculates to a value

of dI/dt of 1011 amps per second! This large value means that

transient protection circuits must use RF design techniques,

particularly considerate of parasitic inductance and

capacitance of conductors.

While this peak energy is certainly impressive, it is really

the longer-term continuing current which carries the bulk of

the charge transferred between the cloud and ground. From

various field measurements, a typical lightning model has

been constructed, as shown in Figure 1.

Figure 1. Typical Lightning Model, with and without

Continuing Current

Flash with Continuing Current

Flash with No Continuing Current

40 μs

6060

TIME (ms)

60 0.3 100

150 A

http://onsemi.com

Depending on various conditions, continuing current may

or may not be present in a lightning strike. A severe lightning

model has also been created, which gives an indication of the

strength which can be expected during worst case conditions

at a point very near the strike location. Figure 2 shows this

model. Note that continuing current is present at more than

one interval, greatly exacerbating the damage which can be

expected. A severe strike can be expected to ignite

combustible materials.

Figure 2. Severe Lightning Model

460 8607006405805201601106010

400 A

200 A

TIME (ms)

A direct hit by lightning is, of course, a dramatic event. In

fact, the electric field strength of a lightning strike nearby

may be enough to cause catastrophic or latent damage to

semiconductor equipment. It is a more realistic venture to try

to protect equipment from these nearby strikes than to

expect survival from a direct hit.

With this in mind, it is important to be able to quantify the

induced voltage as a function of distance from the strike.

Figure 3 shows that these induced voltages can be quite high,

explaining the destruction of equipment from relatively

distant lightning flashes.

Figure 3. Voltage Induced by Nearby

Lightning Strike

10000

1000

100

11010.1

DISTANCE FROM STRIKE (km)

INDUCED VOLTAGE IN 1 m OF WIRE (V)

Burying cables does not provide appreciable protection as

the earth is almost transparent to lightning radiated fields. In

fact, underground wiring has a higher incidence of strikes

than aerial cables.3

SUNSPOTS

The sun generates electromagnetic waves which can

disrupt radio signals and increase disturbances on residential

and business power lines. Solar flares, which run in cycles

of 11 years (1989 was a peak year) send out electromagnetic

waves which disrupt sensitive equipment.

Although not quantified, the effects of sunspot activity

should be considered. Sunspots may be the cause of

sporadic, and otherwise unexplainable problems in such

sensitive areas.

SWITCHING OF LOADS IN POWER CIRCUITS

Inductive switching transients occur when a reactive load,

such as a motor or a solenoid coil, is switched off. The

rapidly collapsing magnetic field induces a voltage across

the load’s winding which can be expressed by the formula:

V = −L (dI/dt)

where L is inductance in henrys and dI/dt is the rate of

change of current in amps per second.

Such transients can occur from a power failure or the

normal opening of a switch. The energy associated with the

transient is stored within the inductance at power

interruption and is equal to:

W = 1/2 Li2

where W is energy in joules and i is instantaneous current in

amps at the time of interruption.

As an example, a 1.4 to 2.5 kV peak transient can be

injected into a 120 vac power line when the ignition system

of an oil furnace is fired. It has also been shown that there are

transients present on these lines which can reach as high as

6 kV. In locations without transient protection devices, the

maximum transient voltage is limited to about 6 kV by the

insulation breakdown of the wiring.

Inductive switching transients are the silent killers of

semiconductors as they often occur with no outward

indication. A graphic example is the report of a large

elevator company indicating the failure of 1000 volt

rectifiers during a power interruption. In another area, power

interruption to a 20 HP pump motor in a remote area was

directly related to failure of sensitive monitoring equipment

at that same site.4

http://onsemi.com

Figure 4. Switching Transient Definition for Aircraft

and Military Buses, per Boeing Document D6-16050

28 Vdc BA

WAVEFORM AT POINT A

TIME (μs)

t1t2

V = 600 V pk-pk

No. of Repetitions = 5 to 100

t1 = 0.2 to 10 μs

t2 = 50 to 1000 μs

After characterizing electrical overstress on aircraft

power buses, Boeing published Document D6-16050 as

shown in Figure 4.

The military has developed switching transient

definitions within several specifications including:

DOD-STD-1399 for shipboard

MIL-STD-704 for aircraft

MIL-STD-1275 for ground vehicles

The International Electrotechnical Commission (IEC) is

now promoting their specification IEC 801-4 throughout

the European community. This describes an inductive

switching transient voltage threat having 50 ns wide spikes

with amplitudes from 2 kV to 4 kV occurring in 300 ms wide

bursts.5

Besides these particular military specifications, many are

application specific and functional tests exist. A supplier of

transient voltage suppressor components will be expected to

perform to a wide variety of them.

ELECTROSTATIC DISCHARGE (ESD)

ESD is a widely recognized hazard during shipping and

handling of many semiconductor devices, especially those

that contain unprotected MOSFETs, semiconductors for use

at microwave frequencies and very high speed logic with

switching times of 2 ns or less. In response to this threat,

most semiconductors are routinely shipped in containers

made of conductive material.

In addition to various shipping precautions, electronic

assembly line workers should be grounded, use

grounded-tip soldering irons, ionized air blowers and other

techniques to prevent large voltage potentials to be

generated and possibly discharged into the semiconductors

they are handling.

Once the assembled device is in normal operation, ESD

damage can still occur. Any person shuffling his feet on a

carpet and then touching a computer keyboard can possibly

cause a software crash or, even worse, damage the keyboard

electronics.

The electrical waveform involved in ESD is a brief pulse,

with a rise time of about 1 ns, and a duration of 100−300 μs.

The peak voltage can be as large as 30 kV in dry weather, but

is more commonly 0.5−5.0 kV.6 The fastest rise times occur

from discharges originating at the tip of a hand-held tool,

while discharges from the finger tip and the side of the hand

are slightly slower.7 A typical human with a body

capacitance of 150 pF, charged to 3 microcoulombs, will

develop a voltage potential of 20 kV, according to the

formula:

V = Q / C

where V is voltage, Q is charge and C is capacitance. The

energy delivered upon discharge is:

W = 1/2 CV2

where W is energy in joules, C is capacitance and V is

voltage.

It is interesting to note that most microcircuits can be

destroyed by a 2500 volt pulse, but a person cannot feel a

static spark of less than 3500 volts!

NUCLEAR ELECTROMAGNETIC PULSES (NEMP)

When a nuclear weapon is detonated, a very large flux of

photons (gamma rays) is produced. These rays act to

produce an electromagnetic field known as a nuclear

electromagnetic pulse or NEMP. When a nuclear detonation

occurs above the atmosphere, a particulary intense pulse

illuminates all objects on the surface of the earth, and all

objects in the lower atmosphere within line of sight of the

burst. A burst 300−500 km above Kansas would illuminate

the entire continental U.S.

A typical NEMP waveform is a pulse with a rise time of

about 5 ns and a duration of about 1 μs. Its peak electric field

is 50−100 kV/m at ground level. After such a pulse is

coupled into spacecraft, aircraft and ground support

equipment, it produces a waveform as described in

MIL-STD-461C. The insidious effect of NEMP is its broad

coverage and its potential for disabling military defense

systems.

MICROWAVE RADIATION

Microwaves can be generated with such high power that

they can disable electronic hardware upon which many

military systems depend. A single pulse flux of 10−8 MJ/cm2

burns out receiver diodes, and a flux of 10−4 MJ/cm2 causes

bit errors in unshielded computers.8 With automobiles

utilizing MPU controls in more applications, it is important

to protect against the effects of driving by a microwave

transmitter. Likewise, a nearby lightning strike could also

have detrimental effects to these systems.

POWER CROSS

Yet another source of electrical overstress is the accidental

connection of signal lines, such as telephone or cable

television, to an ac or dc power line. Strictly speaking, this

phenomenon, known as a power cross, is a continuous state,

not a transient. However, the techniques for ensuring the

http://onsemi.com

survival of signal electronics after a power cross are similar

to techniques used for protection against transient

overvoltages.

STANDARDIZED WAVEFORMS

Fortunately, measurements of these hazards have been

studied and documented in several industry specifications.

For example, Bellcore Technical Advisory

TA-TSY-000974 defines the generic measurement

waveform for any double exponential waveform, which is

the basis for most of the specific applications norms.

The predominant waveform for induced lightning

transients, set down by Rural Electrification Administration

Document PE-60, is shown in Figure 5. This pulse test,

performed at the conditions of 100 V/μs rise, 10/1000 μs, Ip

= 1 kV, is one of the two most commonly specified in the

industry. The other is the 8/20 μs waveform, shown in

Figure 6.

Figure 5. Pulse Waveform (10/1000 μs)

Figure 6. Pulse Waveform (8/20 μs)

3020100

.9 lp

.5 lp

.1 lp

TIME, μs

123

.1 Ip

.5 Ip

.9 Ip

TIME, ms

TRANSIENT VOLTAGE SUPPRESSION

AND TELECOM

TRANSIENT VOLTAGE SURGE SUPPRESSION

COMPONENTS ON DATA AND TELEPHONE LINES

Lines carrying data and telephone signals are subject to a

number of unwanted and potentially damaging transients

primarily from two sources: lightning and “power crosses.”

A power cross is an accidental connection of a signal line to

a powerline. Transients from lightning can impress voltages

well above a kilovolt on the line but are of short duration,

usually under a millisecond. Lightning transients are

suppressed by using Transient Voltage Surge Suppressor

(TVS) devices. TVS devices handle high peak currents

while holding peak voltage below damaging levels, but have

relatively low energy capability and cannot protect against

a power cross fault. The first TVS used by telephone

companies is the carbon block, but its peak let-through

voltage was too high for modern equipment using

unprotected solid state circuitry. A number of other

components fill today’s needs.

The power cross condition causes a problem with

telephone lines. Fast acting fuses, high speed circuit

breakers and positive temperature coefficient thermistors

have been successfully used to limit or interrupt current

surges exceeding a millisecond.

Over the years, telecommunications switching equipment

has been transitioning from electromechanical relays to

integrated circuits and MOSFET technology. The newer

equipment operates at minimal electrical currents and

voltages, which make it very efficient. It is therefore quite

sensitive to electrical overloads caused by lightning strikes

and other transient voltage sources, and by power crosses.

Because of the deployment of new technology, both in

new installations and in the refurbishment of older systems,

the need for transient protection has grown rapidly. It is

widely recognized that any new equipment must include

protection devices for reasons of safety, reliability and long

term economy.

The major telecom companies, in their never ending quest

for the elimination of electromechanical technology have

been looking at a number of novel methods and

implementations of protection. These methods provide for

solutions to both the primary and secondary protection

categories.

A number of studies have been conducted to determine the

transient environment on telephone lines. Very little has

been done with data lines because a typical situation does not

exist. However, information gathered from telephone line

studies can serve as a guide for data lines.

Past studies on telephone lines coupled with the high

current capability of arc type arrestors and the conservative

nature of engineers seem to have produced specifications

which far exceed the real need. A recent study by Bell South

Services9 reported that the highest level of transient energy

encountered was well below standards and specifications in

common use. Now, solid state devices perform adequately

for many applications. However the stringent specifications

of some regulatory agencies promote arc-type arrestors,

though solid state devices would be a better choice.

PRIMARY PROTECTION

Primary protection is necessary to protect against high

voltage transients which occur in the outdoor environment.

These transients include induced lightning surges and ac

cross conditions.

As such, primary protection is located at the point where

wiring enters the building or terminal box. It is the first line

defense against outside hazards. TVS devices located where

lines enter a building are called primary protectors.

http://onsemi.com

Protectors connected to indoor lines are referred to as

secondary protectors. Both primary and secondary

protectors are required to provide complete equipment

protection.

Today, primary protection is most generally accomplished

through the use of surge protector modules. For telecom,

these are designed specifically for the environment and the

standards dictated by the telecom applications. They

typically contain a two or three element gas arrestor tube and

a mechanically-triggered heat coil. Some also include air

gap carbon block arrestors which break over at voltages

above about 1500 volts.

Some modules contain high speed diodes for clamp

response in the low nanoseconds. This provides protection

until the gas tube fires, generally in about one microsecond.

The diodes may be connected between the tip, ring and

ground conductors in various combinations. The 5ESS

electronic switching system norms dictate design and

performance requirements of TVS modules in use today.

Test methods are spelled out in REA PE-80, a publication

of the Rural Electrification Administration.

In the U.S. alone, 58 million primary protection modules

are sold annually, about 40 million for central offices and 18

million for station locations, such as building entrances.

Eighty percent of these use gas tubes, 16% use air-gap

carbon blocks, and only 2% (so far) are solid state.

SECONDARY PROTECTION

Secondary protection is necessary for the equipment

inputs, and as such, is located between the primary protector

and the equipment. Secondary protection is generally

accomplished with one or more TVS components, as

opposed to the modules used for primary. It is often placed

on a circuit board along with other components handling

other duties, such as switching. Secondary protection is

applied to lines associated with long branch circuits which

have primary protection a significant distance away, to

internal data lines, and to other locations requiring

additional local hazard-proofing.

While not as open to external transients as the primary,

secondary can still see peak open circuit voltages in excess

of 1000 volts and short circuit currents of hundreds of amps.

These transients may be locally generated, or they may be

residuals from the primary protectors upstream.

STANDARDS

Transient voltage waveforms are commonly described in

terms of a dual exponential wave as defined in Figure 7. The

standard chosen for power lines is a 1.2/50 μs voltage wave

which causes an 8/20 μs current wave. Although the source

of the most severe transients on telecom lines is the same as

for power lines and lightning, the higher impedance per unit

length of the telephone line stretches the waves as they

propagate through the lines.

Figure 7. Definition of Double Exponential

Impulse Waveform

0.9 P

0.5 P

0.1 P

abT1T2

TIME

WAVEFORM IS DEFINED AS tr/td

WHERE

tr : FRONT TIME = 1.25 (b - a)

= (T1-T0)

td : DURATION = (T2-T0)

The 10/1000 μs wave approximates the worst case

waveform observed on data and telecom lines. TVS devices

intended for this service are usually rated and characterized

using a 10/1000 waveform. The Bell South study revealed

that the worst transient energy handled by primary

protectors on lines entering a central office was equivalent

to only 27 A peak of a 10/1000 wave. This level is

considerably less than that required by secondary protectors

in most of the standards in use today. This finding is

particularly significant because the Bell South service area

includes Central Florida, the region experiencing the highest

lightning activity in the U.S.

The United States Federal Communications Commission

(FCC) has defined mandatory requirements for equipment

which is to be connected to the U.S. telephone network. In

some cases, U.S. equipment must meet standards developed

by the Rural Electrification Agency (REA). Many nations

demand compliance to standards imposed by the

Consultative Committee, International Telegraph and

Telephone (CCITT). In addition, most equipment users

demand safety certification from U.L., which has its own

standards.

The FCC Standards are based on a worst case residue from

a carbon block primary protector installed where the phone

line enters the building. The CCITT standard is applicable

for situations lacking primary protection, other than wiring

flashover. Companies entering the telephone equipment or

protector market will need to obtain and become familiar

with the appropriate governing standards.

TRANSIENT VOLTAGE PROTECTION

COMPONENTS

GENERAL TVS CHARACTERISTICS

A number of transient voltage suppressor (TVS) devices

are available. Each finds use in various applications based

upon performance and cost. All types are essentially

transparent to the line until a transient occurs; however,

some devices have significant capacitance which loads the

line for ac signals. A few of the these are described in

Table 1.

Based upon their response to an overvoltage, TVS devices

fit into two main categories, clamps and crowbars. A clamp

conducts when its breakdown voltage is exceeded and

reverts back to an open circuit when the applied voltage

http://onsemi.com

drops below breakdown. A crowbar switches into a low

voltage, low impedance state when its breakover voltage is

exceeded and restores only when the current flowing

through it drops below a “holding” level.

CLAMP DEVICES

All clamp devices exhibit the general V-I characteristic of

Figure 8. There are variations; however, some clamps are

asymmetric. In the non clamping direction, some devices

such as the zener TVS exhibit the forward characteristic of

a diode while others exhibit a very high breakdown voltage

and are not intended to handle energy of “reverse” polarity.

Under normal operating conditions, clamp devices appear

virtually as an open circuit, although a small amount of

leakage current is usually present. With increasing voltage

a point is reached where current increases rapidly with

voltage as shown by the curved portion of Figure 8. The

rapidly changing curved portion is called the “knee region.”

Further increases in current places operation in the

“breakdown” region.

Figure 8. Static Characteristics of a Clamp Device

In the knee region the V-I characteristic of clamping

devices can be approximated by the equation:

I = K Vs (1)

where K is a constant of proportionality and s is an exponent

which defines the “sharpness factor” of the knee. The

exponent s is 1 for a resistor and varies from 5 to over 100

for the clamping devices being used in TVS applications. A

high value of s i.e., a sharp knee, is beneficial. A TVS device

can be chosen whose breakdown voltage is just above the

worst case signal amplitude on the line without concern of

loading the line or causing excessive dissipation in the TVS.

As the current density in the clamp becomes high, the

incremental resistance as described by Equation 1 becomes

very small in comparison to the bulk resistance of the

material. The incremental resistance is therefore ohmic in

the high current region.

Unfortunately, a uniform terminology for all TVS devices

has not been developed; rather, the terms were developed in

conjunction with the appearance of each device in the

marketplace. The key characteristics normally specified

define operation at voltages below the knee and at currents

above the knee.

Leakage current is normally specified below the knee at

a voltage variously referred to as the stand-off voltage, peak

working voltage or rated dc voltage. Some devices are rated

in terms of an RMS voltage, if they are bidirectional. Normal

signal levels must not exceed this working voltage if the

device is to be transparent.

Breakdown voltage is normally specified at a fairly low

current, typically 1 mA, which places operation past the

knee region. Worst case signal levels should not exceed the

breakdown voltage to avoid the possibility of circuit

malfunction or TVS destruction.

The voltage in the high current region is called the

clamping voltage, VC. It is usually specified at the maximum

current rating for the device. To keep VC close to the

breakdown voltage, s must be high and the bulk resistance

low. A term called clamping factor, (FC) is sometimes used

to describe the sharpness of the breakdown characteristic.

FC is the ratio of clamping voltage to the breakdown voltage.

As the V-I characteristic curve of the TVS approaches a right

angle, the clamping factor approaches unity. Clamping

factor is not often specified, but it is useful to describe clamp

device behavior in general terms.

Table 1. Comparison of TVS Components

Type

Protection

Time

Protection

Voltage

Power

Dissipation

Reliable

Performance

Expected

Life

Other

Considerations

GAS TUBE > 1 μs 60 −100 V Nil No Limited Only 50 −2500 surges.

Can short power line.

MOV 10 −20 ns > 300 V Nil No Degrades Fusing required. Degrades.

Voltage level too high.

AVALANCHE TVS 50 ps 3−400 V Low Yes Long Low power dissipation.

Bidirectional requires dual.

THYRISTOR TVS < 3 ns 30 −400 V Nil Yes Long High capacitance.

Temperature sensitive.

http://onsemi.com

Clamp devices generally react with high speed and as a

result find applications over a wide frequency spectrum. No

delay is associated with restoration to the off state after

operation in the breakdown region.

CROWBAR DEVICES

Crowbar TVS devices have the general characteristics

shown in Figure 9. As with clamp devices, asymmetric

crowbars are available which may show a diode forward

characteristic or a high voltage breakdown in one direction.

Figure 9. Static Characteristics of a

Bidirectional Crowbar Device

VDV3O2

V3O1

The major difference between a crowbar and a clamp is

that, at some current in the breakdown region, the device

switches to a low voltage on-state. In the clamping region

from I1 to I2, the slope of the curve may be positive as shown

by segment 1, negative (segment 2) or exhibit both

characteristics as shown by segment 3. A slightly positive

slope is more desirable than the other two curves because a

negative resistance usually causes a burst of high frequency

oscillation which may cause malfunction in associated

circuitry. However, a number of performance and

manufacturing trade-offs affect the shape of the slope in the

clamping region.

A crowbar TVS has an important advantage over a clamp

TVS in that it can handle much larger transient surge current

densities because the voltage during the surge is

considerably lower. In a telephone line application, for

example, the clamping level must exceed the ring voltage

peak and will typically be in the vicinity of 300 volts during

a high current surge. The on-state level of the crowbar may

be as low as 3 volts for some types which allows about two

orders of magnitude increase in current density for the same

peak power dissipation.

However, a crowbar TVS becomes “latched” in the

on-state. In order to turn off its current flow the driving

voltage must be reduced below a critical level called the

holding or extinguishing level. Consequently, in any

application where the on-state level is below the normal

system voltage, a follow-on current occurs. In a dc circuit

crowbars will not turn off unless some means is provided to

interrupt the current. In an ac application crowbars will turn

off near the zero crossing of the ac signal, but a time delay

is associated with turn-off which limits crowbars to

relatively low frequency applications. In a data line or

telecom application the turn-off delay causes a loss of

intelligence after the transient surge has subsided.

A telephone line has both ac and dc signals present.

Crowbars can be successfully used to protect telecom lines

from high current surges. They must be carefully chosen to

ensure that the minimum holding current is safely above the

maximum dc current available from the lines.

TVS DEVICES

A description of the various types of TVS devices follows

in the chronological order in which they became available.

Used appropriately, sometimes in combination, any

transient protection problem can be suitably resolved. Their

symbols are shown in Figure 10.

Figure 10. TVS Devices and Their Symbols

Air-Gap

Carbon Block

2- and 3-Element

Gas Tubes

Heat Coil

Switch

Metal Oxide

Varistor (MOV)

Zener

Regulator

Unidirectional

Avalanche TVS De-

vice

Bidirectional

Avalanche TVS Device

Thyristor TVS Devices Dual Thyristor TVS Devices

AIR GAP ARRESTORS

The air gap is formed by a pair of metal points rigidly fixed

at a precise distance. The air ionizes at a particular voltage

depending upon the gap width between the points. As the air

ionizes breakover occurs and the ionized air provides a low

impedance conductive path between the points.

The breakover threshold voltage is a function of the air’s

relative humidity; consequently, open air gaps are used

mainly on high voltage power lines where precise

performance is not necessary. For more predictable

behavior, air gaps sealed in glass and metal packages are also

available.

Because a finite time is required to ionize the air, the actual

breakover voltage of the gap depends upon the rate of rise of

the transient overvoltage. Typically, an arrestor designed for

a 120 V ac line breaks over at 2200 volts.

Air gaps handle high currents in the range of 10,000

amperes. Unfortunately, the arc current pits the tips which

causes the breakover voltage and on-state resistance to

increase with usage.

http://onsemi.com

CARBON BLOCK ARRESTORS

The carbon block arrestor, developed around the turn of

the century to protect telephone circuits, is still in place in

many older installations. The arrestor consists of two carbon

block electrodes separated by a 3 to 4 mil air gap. The gap

breaks over at a fairly high level − approximately 1 kV − and

cannot be used as a sole protection element for modern

telecom equipment. The voltage breakdown level is

irregular. With use, the surface of the carbon block is burned

which increases the unit’s resistance. In addition, the burned

material forms carbon tracks between the blocks causing a

leakage current path which generates noise. Consequently,

many of the carbon blocks in service are being replaced by

gas tubes and are seldom used in new installations.

SILICON CARBIDE VARISTORS

The first non-linear resistor to be developed was called a

“varistor.” It was made from specially processed silicon

carbide and found wide use in high power, high voltage TVS

applications. It is not used on telecom lines because its

clamping factor is too high: s is only about 5.

GAS SURGE ARRESTORS

Gas surge arrestors are a sophisticated modification of the

air gap more suited to telecom circuit protection. Most often

used is the “communication” type gap which measures

about 3/8 inch in diameter and 1/4 inch thick. A cross section

is shown in Figure 11. They consist of a glass or ceramic

envelope filled with a low pressure inert gas with specialized

electrodes at each end. Most types contain a minute quantity

of radioactive material to stabilize breakover voltage.

Otherwise, breakover is sensitive to the level of ambient

light.

Because of their small size and fairly wide gap,

capacitance is very low, only a few picofarads. When not

activated, their off-state impedance or insulation resistance

is virtually infinite.

ÇÇÇÇÇ

ÇÇÇÇ

Figure 11. Gas Arrestor Cross Section

ELECTRODES

DISCHARGE

REGION

INSULATOR

(GLASS OR

CERAMIC MATERIAL)

ACTIVATING

COMPOUND

IGNITION AID

Key electrical specifications for this TVS type include

breakover voltage (dc & pulse), maximum holdover

voltage, arc voltage, and maximum surge current.

The breakover voltage is rated at a slow rate of rise,

5000 V/s, essentially dc to a gas arrestor. Typical dc voltage

ratings range from 75 V through 300 V to provide for most

communication systems protection requirements. The

maximum pulse voltage rating is that level at which the

device fires and goes into conduction when subjected to a

fairly rapid rate of voltage rise, (dv/dt) usually 100 V/μs.

Maximum rated pulse voltages typically range from 400 V

to 600 V, depending on device type.

A typical waveform of a gas surge arrestor responding to

a high voltage pulse is shown in Figure 12. From the

waveform, it can be seen that the dv/dt of the wave is 100

v/μs and the peak voltage (the breakover voltage) is 520 V.

Figure 12. Voltage Waveform of Gas Surge Arrestor

Responding to a Surge Voltage

ARRESTOR VOLTAGE (VOLTS)

TIME (500 ns/DIVISION)

200

400

600

0123

Gas surge arrestors fire faster but firing voltage increases

as the transient wave fronts increase in slope as illustrated in

Figure 13. The near vertical lines represent the incident

transient rise time. Note that the response time is greater than

0.1s at slow rise times but decreases to less than 0.1 μs for

risetimes of 20 kV/μs. However, the firing voltage has

increased to greater than 1000 V for the gas tube which

breaks over at 250 Vdc.

The driving circuit voltage must be below the holdover

voltage for the gap to extinguish after the transient voltage

has passed. Holdover voltage levels are typically 60% to

70% of the rated dc breakdown voltage.

Arc voltage is the voltage across the device during

conduction. It is typically specified at 5 to 10 V under a low

current condition, but can exceed 30 V under maximum

rated pulse current.

The maximum surge current rating for a 8/20 μs

waveform is typically in the 10 kA to 20 kA range for

communication type devices. For repetitive surges with a

10/1000 wave, current ratings are typically 100 A,

comfortably above the typical exposure levels in a telephone

subscriber loop.

http://onsemi.com

1.2

1.0

0.8

0.6

0.4

0.2

1.4

87 65 43 21

110

RESPONSE TIME (SECONDS)

SPARKOVER VOLTAGE (kV)

100 V/ sμ

100 V/ms

100 V/SEC

Figure 13. Typical Response Time of a Gas Surge Arrestor

470 Vdc

350 Vdc

250 Vdc

150 Vdc

500 V/sμ

1KV/sμ

5KV/sμ

10KV/sμ

20KV/sμ

Gas tubes normally provide long life under typical

operating conditions, however; wear-out does occur.

Wear-out is characterized by increased leakage current and

firing voltage. An examination of gas tubes in service for six

to eight years revealed that 15% were firing outside of their

specified voltage limits.9 Because firing voltage increases

with use, protectors often use an air gap backup in parallel

with the gas tube. End-of-life is often specified by

manufacturers as an increase of greater than 50% of

breakover or firing voltage. Other limits include a decrease

in leakage resistance to less than 1 mW.

The features and limitations of gas tube surge arrestors are

listed below.

Advantages:

•High current capability

•Low capacitance

•Very high off-state impedance

Disadvantages:

•Slow response time

•Limited life

•High let-through voltage

•Open circuit failure mode

Principally because of their high firing voltage, gas surge

arrestors are not suitable for use as the sole element to

protect modern equipment connected to a data or telecom

line. However, they are often part of a protection network

where they are used as the primary protector at the building

interface with the outside world.

SELENIUM CELLS

Polycrystalline diodes formed from a combination of

selenium and iron were the forerunners of monocrystalline

semiconductor diodes. The TVS cells are built by depositing

the polycrystalline material on a metal plate to increase their

thermal mass thereby raising energy dissipation. The cells

exhibit typical diode characteristics and a non-linear reverse

breakdown which is useful for transient suppression. Cells

can be made which are “self-healing”; that is, the damage

which occurs when subjecting them to excessive transient

current is repaired with time.

Selenium cells are still used in high power ac line

protection applications because of their self-healing

characteristic; however, their high capacitance and poor

clamping factor (s » 8) rule them out for data or telecom line

applications.

METAL OXIDE VARISTORS

The metal oxide varistor (MOV) is composed of zinc

oxide granules in a matrix of bismuth and other metal

oxides. The interface between the zinc oxide and the matrix

material exhibits characteristics similar to that of a p-n

junction having a voltage breakdown of about 2.6 V. With

this structure the electrical equivalent is that of groups of

diodes in parallel which are stacked in series with similar

parallel groups to provide the desired electrical parameters.

The taller the stack, the higher the breakdown and operating

voltage. Larger cross-sections provide higher current

capability. The structure of an MOV is shown in Figure 14.

MICROVARISTOR

ZINC OXIDE

INTERGRANULAR

Figure 14. MOV Cross Section

http://onsemi.com

MOV (27 V)

Vert: 10 V/div

Horiz: 0.5 ms/div

Transient Source Impedance: 0.55 Ω

Vpeak = 62.5 V

Figure 15. MOV Clamping Voltage Waveform

MOVs, formed from a ceramic-like material, are usually

produced in the shape of discs with most widely used MOVs

having diameters of 7 mm, 14 mm, and 20 mm. The disc

surfaces are coated with a highly conductive metal such as

silver to assure uniform conduction through the cross

sectional area of the device. After terminal attachment the

parts are coated with a durable plastic material.

The typical voltage spectrum of MOVs ranges from 8 V

through 1000 V for individual elements. Pulse current

capability (8/20 μs) ranges from a few amperes to several

thousands of amperes depending on the element’s size. The

V-I characteristic of MOVs is similar to Figure 8. Their

clamping factor is fairly good; s is in the vicinity of 25.

Key electrical specifications include: operating voltage,

breakdown voltage, peak current maximum clamping

voltage, and leakage current.

The maximum operating voltage specified is chosen to be

below the breakdown voltage by a margin sufficient to

produce negligible heating under normal operating

conditions. Breakdown voltage is the transition point at

which a small increase in voltage results in a significant

increase in current producing a clamping action. Maximum

limits for breakdown voltage are typically specified at

1 mA with upper end limits ranging from 20% to 40%

greater than the minimum breakdown voltage.

Maximum peak current is a function of element area and

ranges from tens of amperes to tens of thousands of amperes.

MOVs are typically pulse rated with an 8/20 μs waveform

since they are intended primarily for use across power lines.

The clamping characteristics of a 27 V ac rated MOV, with

a 4 joule maximum pulse capability is shown in Figure 15.

The transient energy is derived from an exponentially

decreasing pulse having a peak amplitude of 90 V. The pulse

generator source impedance is 0.55 W. Peak clamping

voltage is 62.5 V while the developed current is 50 A. The

clamping factor calculates to be 2.3.

Leakage currents are listed for MOVs intended for use in

sensitive protection applications but are not normally listed

for devices most often used on power lines. Leakage current

behavior is similar to that of a p-n junction. It roughly

doubles for every 10°C increase in temperature and also

shows an exponential dependence upon applied voltage. At

a voltage of 80% of breakdown, leakage currents are several

microamperes at a temperature of 50°C.

Although the theory of MOV operation is not fully

developed, behavior is similar to a bidirectional avalanche

diode. Consequently its response time is very fast.

Life expectancy is an important characteristic generated

under pulse conditions. A typical example is shown in

Figure 16. The data applies to 20 mm diameter disc types

having rated rms voltage from 130 V to 320 V. Lifetime

rating curves are usually given for each device family.

IMPULSE DURATION  μs

RATED PULSE CURRENT − AMPERES

10,000

5000

2000

1000

500

200

100

Figure 16. MOV Pulse Life Curve

10,000100010020

INDEFINITE

105

104

103

102

106

For a single 8/20 μs pulse, the device described in Figure

16 is rated at 6500 A; however, it must be derated by more

than two orders of magnitude for large numbers of pulses.

Longer duration pulses also require further derating. For

example, for a 10/1000 μs duration pulse, this family of

devices has a maximum pulse rating of about 100 A on a

single shot basis and devices must be derated to less than

10A for long lifetimes in excess of 100,000 pulses.

http://onsemi.com

End-of-life for an MOV is defined as the voltage

breakdown degrading beyond the limits of ±10%. As MOVs

are pulsed, they degrade incrementally as granular

interfaces are overheated and changed to a highly

conductive state. Failure occurs in power line applications

when the breakdown voltage has degraded to the point

where the MOV attempts to clip the powerline peaks. In

telecom applications, their breakdown must be above the

peaks of the impressed ac line during a ring cycle or a power

cross; otherwise an immediate catastrophic failure will

occur.

When MOVs fail catastrophically they initially fail short.

However, if a source of high energy is present as might occur

with a power cross, the follow-on current may cause the part

to rupture resulting in an open circuit.

The advantages and shortcomings of using an MOV for

general purpose protection in microprocessor based

circuitry include the following:

Advantages:

•High current capability

•Broad voltage spectrum

•Broad current spectrum

•Fast response

•Short circuit failure mode

Disadvantages:

•Gradual decrease of breakdown voltage

•High capacitance

The capacitance of MOVs is fairly high because a large

device is required in order to achieve a low clamping factor;

consequently, they are seldom used across telecom lines.

ZENER TVS

Zener TVS devices are constructed with large area silicon

p-n junctions designed to operate in avalanche and handle

much higher currents than their cousins, zener voltage

regulator diodes. Some manufacturers use small area mesa

chips with metal heatsinks to achieve high peak power

capability. However, ON Semiconductor has determined

that large area planar die produce lower leakage current and

clamping factor. The planar construction cross section is

shown in Figure 17 and several packages are shown in

Figure 18.

Figure 17. Zener TVS Cross-Section

ÈÈ

PLASTIC

ENCAPSULATION

PLANAR DIE SOLDER

Figure 18. Typical Insertion and Surface Mount Silicon

TVS Packages — Zeners and Thyristors

59-0317-02 41A

318-07 403A403

Key electrical parameters include maximum operating

voltage, maximum reverse breakdown voltage, peak pulse

current, peak clamping voltage, peak pulse power, and

leakage current.

The normal operating or working voltage is usually called

the reverse standoff voltage in specification sheets. Devices

are generally available over the range of 5 V through 250 V.

Standoff voltage defines the maximum peak ac or dc voltage

which the device can handle. Standoff voltage is typically

10% to 15% below minimum reverse breakdown voltage. A

listing of TVS products available from ON Semiconductor

is shown in Table 2.

Table 2. ON Semiconductor Zener TVS Series

DEVICE

SERIES

RANGE

PULSE POWER

RATING

(100/1000

PULSE) PACKAGE

*SA5.0A-

SA170A 6.8-200 500 W Axial

*P6KE6.8A -

P6KE200A 6.8-200 600 W Axial

*1.5KE6.8A -

1.5KE250A 6.8-250 1500 W Axial

1SMB5.0AT3 -

1SMB170AT3 6.8-200 600 W SMB

P6SMB6.8AT3 -

P6SMB200T3 6.8-200 600 W SMB

1SMC5.0AT3 -

1SMC78AT3 6.8-91 1500 W SMC

1.5SMC6.8AT3

1.5SMC91AT3 6.8-91 1500 W SMC

MMBZ15VDLT1 15 ESD Protection

>15 kV SOT-23

* Available in bidirectional configurations

http://onsemi.com

The reverse breakdown voltage is specified at a bias level

at which the device begins to conduct in the avalanche mode.

Test current levels typically are 1 mA for diodes which

breakdown above 10 V and 10 mA for lower voltage

devices. Softening of the breakdown knee, that is, lower s,

for lower voltage p-n junction devices requires a higher test

current for accurate measurements of reverse breakdown

voltage. Diodes that break down above 10 V display a very

sharp knee; s is over 100.

Peak pulse current is the maximum upper limit at which

the device is expected to survive. Silicon p-n junctions are

rated for constant power using a particular transient

waveform; consequently, current is a function of the peak

clamping voltage. For example, a 6.8 V device handles

about 28 times the pulse current that a 220 V device will

withstand; however, both the 6.8 V and 220 V types dissipate

the same peak power under the same pulse waveform

conditions. Most Zener TVS diodes are rated for 10/1000 μs

waveform pulses which are common in the telecom industry.

The clamping voltage waveform of a 27 V Zener TVS

having a 1.5 joule capability is illustrated in Figure 19. Its

peak voltage is 30.2 V. The transient energy source is the

same as applied to the MOV whose response is shown in

Figure 10. However, the current through the Zener TVS is

over 100 A, much higher than occurs with the MOV because

the clamping voltage is significantly lower. Despite the

higher pulse current, the Zener displays much better

clamping action; its clamping factor is 1.1.

Figure 14

Figure 19. Zener TVS Clamping Voltage Waveform

TIME (500 μs/DIVISION)

V , CLAMPING VOLTAGE (VOLTS)

Peak pulse power is the instantaneous power dissipated at

the rated pulse condition. Common peak pulse power ratings

are 500 W, 600 W, and 1500 W for 10/1000 μs waveforms.

As the pulse width decreases, the peak power capability

increases in a logarithmic relationship. An example of a

curve depicting peak pulse power versus pulse width is

shown in Figure 20. The graph applies to the 1.5 kW series

(10/1000 pulse) of TVS diodes and can be interpolated to

determine power ratings over a broad range of pulse widths.

At 50 μs, the maximum rated power shown in the curve is

6 kW, which is four times greater than the rating at 1ms. The

current handling capability is also increased roughly by this

same factor of four.

Figure 20. Peak Pulse Power Rating

for a Popular Zener TVS Family

1μs 10μs 100μs 1 ms 10 ms

100

, PULSE WIDTH

PP, PEAK P

WER (kW)

0.1μs

To increase power capability devices are stacked in series.

For example, doubling the power capability requirement for

a 100 V, 1.5 kW Zener TVS is easily done by placing two 50

V devices in series. Clamping factor is not significantly

affected by this arrangement.

Although leakage current limits are relatively high for the

industry low voltage types (500 μA to 1000 μA), dropping

off to 5 μA or less for voltages above 10 V, the planar die in

use by ON Semiconductor exhibit considerably less leakage

than the specified limits of the industry types.

Capacitance for the popular 1500 W family exceeds

10,000 pF at zero bias for a 6.8 V part, dropping

exponentially to less than 100 pF for a 200 V device.

Capacitance drops exponentially with a linear increase in

bias. The capacitance of a 6.8 V device is 7000 pF, while the

200 V part measures under 60 pF, at their respective standoff

voltages.

Capacitance loads the signal line at high frequencies. For

high speed data transmission circuits, low capacitance is

achieved by placing two diodes in a series stack as shown in

Figure 21. Under normal operation the top diode (DS)

operates at essentially zero bias current. Since its power

dissipation requirement is small, its area can be much

smaller than that of the TVS diode (DZ) in order to provide

low capacitance. The top diode normally is not intended to

be used in avalanche. Consequently if a negative voltage

exceeding the reverse rating of the stack could occur, the low

capacitance diode must be protected by another diode (DP)

shown connected by dotted lines on Figure 21. The

arrangement of Figure 21 is satisfactory for situations where

the signal on the line is always positive. When the signal is

ac, diode DP is replaced by another low capacitance stack,

connected in anti-parallel.

http://onsemi.com

Figure 21. A Series Stack to Achieve Low

Capacitance with Zener TVS Diodes

Switching speed is a prime attribute of the zener TVS.

Avalanche action occurs in picoseconds but performing tests

to substantiate the theory is extremely difficult. As a

practical matter, the device may be regarded as responding

instantaneously. Voltage overshoots which may appear on

protected lines are the result of poor layout and packaging

or faulty measurement techniques.

The p-n junction diode is a unidirectional device. For use

on ac signal lines, bidirectional devices are available which

are based upon stacking two diodes back to back. Most

manufacturers use monolithic NPN and PNP structures. The

center region is made relatively wide compared to a

transistor base to minimize transistor action which can cause

increased leakage current.

No wearout mechanism exists for properly manufactured

Zener diode chips. They are normally in one of two states;

good, or shorted out from over-stress. Long-term life studies

show no evidence of degradation of any electrical

parameters prior to failure. Failures result from stress which

causes separation of the metal heat sink from the silicon chip

with subsequent overheating and then failure. Like MOVs,

silicon chips quickly fail short under steady state or long

duration pulses which exceed their capabilities.

The strengths and weaknesses of Zener TVS devices are

listed below.

Advantages:

•High repetitive pulse power ratings

•Low clamping factor

•Sub-nanosecond turn-on

•No wearout

•Broad voltage spectrum

•Short circuit failure mode

Disadvantages:

•Low non-repetitive pulse current

•High capacitance for low voltage types

Because of their fast response and low clamping factor,

silicon devices are used extensively for protecting

microprocessor based equipment from voltage surges on dc

power buses and I/O ports.

THYRISTOR DIODES

The most recent addition to the TVS arsenal is the

thyristor surge suppressor (TSS). The device has the low

clamping factor and virtually instantaneous response

characteristic of a silicon avalanche (Zener) diode but, in

addition, it switches to a low voltage on-state when

sufficient avalanche current flows. Because the on-state

voltage is only a few volts, the TSS can handle much higher

currents than a silicon diode TVS having the same chip area

and breakdown voltage. Furthermore, the TSS does not

exhibit the large overshoot voltage of the gas tube.

Thyristor TVS diodes are available with unidirectional or

bidirectional characteristics. The unidirectional type

behaves somewhat like an SCR with a Zener diode

connected from anode to gate. The bidirectional type

behaves similarly to a triac having a bidirectional diode

(Diac) from main terminal to gate.

Packaged TSS chips are shown in Figure 18. Figure 22a

shows a typical positive switching resistance bidirectional

TSS chip. Construction of the device starts with an n

material wafer into which the p-bases and n-emitters are

diffused. There are four layers from top to bottom on each

side of the chip, forming an equivalent SCR. Only half the

device conducts for a particular transient polarity.

http://onsemi.com

Figure 22a. Chip Construction

ÈÈÈÈ

Figure 22c. Circuit Model

of Left Side

P EMITTER

N BASE

N EMITTER

P BASE

Figure 22b. Cross Section

of Left Side

MT1 MT2

Figure 22d. Circuit Symbol

The “gate” does not trigger the SCR, instead, operation in

the Zener mode begins when the collector junction

avalanches. Note that the p-bases pass through the

n-emitters in a dot pattern and connect to the contact metal

covering both halves of the chip. This construction

technique provides a low resistance path for current flow

and prevents it from turning on the NPN transistor.

Therefore, at relatively low currents, the device acts like a

low gain PNP transistor in breakdown. The Zener diode is

the collector base junction of the PNP transistor. Negative

resistance TSS devices are similarly constructed but start

with p-type material wafer, allowing the fabrication of a

high-gain NPN transistor. The switchback in voltage with

increasing current is caused by the gain of the NPN.

Both device types switch on completely when the current

flow through the base emitter shunt resistance causes

enough voltage drop to turn on the emitter and begin four

layer action. Now the device acts like an SCR. The collector

current of the PNP transistor (Figure 22c) provides the base

drive for the NPN transistor. Likewise, the collector current

of the NPN transistor drives the base of the PNP causing the

two devices to hold one another on. Both the p and n emitters

flood the chip with carriers resulting in high electrical

conductivity and surge current capability.

When driven with high voltage ac, which occurs during a

power cross, positive resistance TSS devices act like a Zener

diode until the ac voltage drives the load line through the

point where regeneration occurs. Then it abruptly switches

to a low voltage. When the peak ac current is just below the

current required for breakover, the device operates mainly

as a Zener and power dissipation is high although the current

is low. When the ac current peak is well above breakover,

(>10 A), the device operates mainly as an SCR, and the low

on-state voltage causes power dissipation to be relatively

low.

Negative resistance devices operate in a similar fashion.

However, their behavior is dependent upon the “load line,”

that is, the equivalent resistance which the device “sees.”

When the load resistance is high (>1000 W) behavior is

similar to that of a positive resistance TSS in that high

http://onsemi.com

instantaneous power dissipation occurs as the load line is

driven along the high voltage region of the TSS prior to

switching.

When a TSS with a negative resistance characteristic is

driven with a low “load” resistance, switching occurs when

the load line is tangent to the peak of the negative resistance

curve. Thus, complete turn-on can occur at a very low

current if the load resistance is low and the device has a

“sharp” switchback characteristic.

Leakage and the Zener knee voltage increase with

temperature at eight percent and 0.11% per °C respectively

for 200 V positive resistance types. But the current required

to cause regeneration falls with temperature, causing less

Zener impedance contribution to the breakover voltage,

resulting in a large reduction in the breakover voltage

temperature coefficient to as little as 0.05%/°C. Negative

resistance types can show positive or negative breakover

voltage coefficients depending on temperature and the

sharpness of the negative switchback.

The response of both positive and negative switching

resistance units to fast transients involves a race between

their Zener and regenerative attributes. At first the device

conducts only in the small chip area where breakdown is

occurring. Time is required for conduction to spread across

the chip and to establish the currents and temperatures

leading to complete turn-on. The net result is that both types

exhibit increasing breakover voltage with fast transients.

However, this effect is very small compared to gas discharge

tubes, being less than 25% of the breakover voltage.

Negative resistance types are more sensitive to unwanted

turn-on by voltage rates (dv/dt) at peak voltages below the

avalanche value. The transient current that flows to charge

the self-capacitance of the device sets up an operating point

on the negative resistance slope leading to turn-on.

Reduction of dv/dt capability becomes significant when the

signal voltage exceeds 80% of the avalanche value.

Complete turn-off following a transient requires the load

line to intersect the device leakage characteristic at a point

below the avalanche knee. During turn-off the load line must

not meet an intermediate conducting state which can occur

with a negative resistance device. Positive resistance types

are free of states causing turn-off “sticking.” Both types

have temperature sensitive holding currents that lie between

1 and 4 mA/°C.

Recent product developments and published studies have

generated much interest. Based on a study sponsored by Bell

South,9 the authors concluded that these new devices offered

the highest level of surge protection available.

Key electrical parameters for the Thyristor TVS include

operating voltage, clamping voltage, pulse current, on-state

voltage, capacitance, and holding current.

Operating voltage is defined as the maximum normal

voltage which the device should experience. Operating

voltages from 60 V to 200 V are available.

Clamping voltage is the maximum voltage level attained

before thyristor turn-on and subsequent transition to the

on-state conduction mode. The transition stage to

conduction may have any of the slopes shown in Figure 9.

The important voltages which define the thyristor

operating characteristics are also shown in Figure 9. VD is

operating voltage, VC is the clamping voltage and VT is

on-state voltage.

On-state voltage for most devices is approximately 3 V.

Consequently, transient power dissipation is much lower for

the thyristor TVS than for other TVS devices because of its

low on-state voltage. For example, under power cross

conditions Bell South Services reported their tests showed

that the thyristor TVS devices handled short bursts of

commercial power with far less heating than arc type surge

arrestors.9

Capacitance is also a key parameter since in many cases

the TSS is a replacement for gas surge arrestors which have

low capacitance. Values for the TSS range from 100 pF to

200 pF at zero volts, but drop to about half of these values

at a 50 Vdc bias.

Holding current (IH) is defined as the current required to

maintain the on-state condition. Device thru-current must

drop below IH before it will restore to the non-conducting

state. Turn-off time is usually not specified but it can be

expected to be several milliseconds in a telecom application

where the dc follow-on current is just slightly below the

holding current.

The major advantages and limitations of the thyristor are:

Advantages:

•Fast response

•No wearout

•Produces no noise

•Short circuit failure mode

Disadvantages:

•Narrow voltage spectrum

•Non-restoring in dc circuits unless current is below IH

•Turn-off delay time

The thyristor TVS is finding wide acceptance in telecom

applications because its characteristics uniquely match

telecom requirements. It handles the difficult “power cross”

requirements with less stress than other TVS devices while

providing the total protection needed.

SURGE PROTECTOR MODULES

TYPES OF SURGE PROTECTOR MODULES

Several component technologies have been implemented

either singly or in combination in surge protector modules

and devices. The simplest surge protectors contain nothing

more than a single transient voltage suppression (TVS)

component in a larger package. Others contain two or more

in a series, parallel or series-parallel arrangement. Still

others contain two or more varieties of TVS elements in

combination, providing multiple levels of protection.

Many surge protectors contain non-semiconductor

elements such as carbon blocks and varistors. If required,

http://onsemi.com

other modes of protection components may be incorporated,

such as circuit breakers or EMI noise filters.

Surge protector modules are one solution to the

overvoltage problem. Alternatives include:

•Uninterruptible power systems (UPS), whose main

duty is to provide power during a blackout, but

secondarily provide protection from surges, sags and

spikes.

•Power line conditioners, which are designed to isolate

equipment from raw utility power and regulated

voltages within narrow limits.

Both UPS and power line conditioners are far more

expensive than surge protector modules.

THE 6 MAJOR CATEGORIES OF

SURGE PROTECTION MODULES

Plug-in

Hardwired

Utility

Datacom

Telecom

RF and microwave

PLUG-IN MODULES

Plug-in modules come in a variety of sizes and shapes, and

are intended for general purpose use. They permit the

protection of vulnerable electronic equipment, such as home

computers, from overvoltage transients on the 115 vac line.

These products are sold in retail outlets, computer stores and

via mail order. Most models incorporate a circuit breaker or

fuse, and an on/off switch with a neon indicator. The module

may have any number of receptacles, with common models

having from two to six. These products comply with UL

1449, and are generally rated to withstand the application of

multiple transients, as specified in IEEE 587. Plug-in

modules generally provide their protection through the use

of these devices which are typically connected between line

and neutral, and between neutral and ground.

HARDWIRED

Hardwired modules take on a wide variety of styles,

depending upon their designed application. They provide

protection for instrumentation, computers, automatic test

equipment, industrial controls, motor controls, and for

certain telecom situations.

Many of these modules provide snubbing networks

employing resistors and capacitors to produce an RC time

constant. Snubbers provide common mode and differential

mode low-pass filtering to reduce interference from line to

equipment, and are effective in reducing equipment

generated noise from being propogated onto the line.

Snubbers leak current however, and many of these modules

are designed with heat sinks and require mounting to a

chassis. The surge protection is performed in a similar

manner to the plug-in modules mentioned earlier.

Hardwired products, therefore, present a prime opportunity

for avalanche TVS components.

UTILITY

The power transmission and distribution equipment

industry has an obvious need for heavy duty protection

against overvoltage transients. Many utility situations

require a combination of techniques to provide the necessary

solution to their particular problems. This industry utilizes

many forms of transient suppression outside the realm of

semiconductors.

DATACOM

Local area networks and other computer links require

protection against high energy transients originating on their

data lines. In addition, transients on adjacent power lines

produce electromagnetic fields that can be coupled onto

unprotected signal lines. Datacom protectors have a ground

terminal or pigtail which must be tied to the local equipment

ground with as short a lead as possible. Datacom protectors

should be installed on both ends of a data link, or at all nodes

in a network. This protection is in addition to the ac line

transient protection, which is served by the plug-in or

hardwired protection modules. Some datacom protector

modules contain multi-stage hybrid circuits, specially

tailored for specific applications, such as 4−20 mA analog

current loops.

TELECOM

Included here are devices used to protect central office and

station telecommunications (telecom) equipment against

voltage surges. None of these devices are grounded through

an ac power receptacle. Those that are grounded through an

ac power receptacle are categorized as plug-in modules. Not

only can overvoltages cause disruptions of telecom service,

but they can destroy the sophisticated equipment connected

to the network. Also, users or technicians working on the

equipment can be injured should lightning strike nearby. It

is estimated that 10 to 15 people are killed in the U.S. each

year while talking on the telephone during lightning storms.

For these reasons, surge protectors are used both in central

offices and in customer premises.

There are three types of telecom surge protectors now in

service: air-gap carbon block, gas tube, and solid state. The

desire of the telecom market is to convert as many of the

non-solid state implementations into solid state as cost will

permit.

SELECTING TVS COMPONENTS

From the foregoing discussion it should be clear that the

silicon junction avalanche diode offers more desirable

characteristics than any other TVS component. Its ability to

clamp fast rising transients without overshoot, low clamping

factor, non-latching behavior, and lack of a wearout

mechanism are the overriding considerations. Its one-shot

surge capability is lower than most other TVS devices but is

normally adequate for the application. Should an unusually

severe event occur, it will short yet still protect the

equipment.

http://onsemi.com

For example, an RS-232 data line is specified to operate

with a maximum signal level of ± 25 V. Failure analysis

studies11 have shown that the transmitters and receivers

used on RS-232 links tolerate 40 V transients. A

1.5KE27CA diode will handle the maximum signal level

while holding the peak transient voltage to less than 40 V

with a 40 A 10/1000 pulse which is adequate for all indoor

and most outdoor data line runs. As a practical matter, few

data links use 25 V signals; 5 V is most common.

Consequently, much lower voltage silicon diodes may be

used which will allow a corresponding increase in surge

current capability. For example, a 10 V breakdown device

from the same 1500 W family will clamp to under 15 V

(typically 12 V) when subjected to a 100 A pulse.

Telecommunications lines which must accommodate the

ring voltage have much more severe requirements. For

example, one specification11 from Bellcore suggests that

leakage current be under 20 mA over the temperature range

from −40°C to +65°C with 265 V peak ac applied. To meet

this specification using Zener TVS parts, devices must be

stacked. Devices which breakdown at 160 V are chosen to

accommodate tolerances and the temperature coefficient. A

part number with a 10% tolerance on breakdown could

supply a unit which breaks down at 144 V. At −40°C

breakdown could be 133 V. The breakdown of two devices

stacked just barely exceeds the worst case ring peak of

265 V. A 1500 W unit has a surge capability of 6.5 A

(10/1000) which is too low to be satisfactory while higher

power units are too expensive as a rule.

Another problem which telephone line service presents to

a surge suppressor is survival during a power cross. An

avalanche diode is impractical to use because the energy

delivered by a power cross will produce diode failure before

any overcurrent protective element can react.

As indicated by the Bell South studies, the thyristor TVS

is ideal for telephone line applications. Suppliers offer

unidirectional and bidirectional units which meet the FCC

impulse wave requirements as shown in Table 1. In addition,

the thyristor can handle several cycles of 50/60 Hz power

before failure. The ON Semiconductor MKT1V200 series,

for example, can handle 10 A for 4 cycles, which is enough

time for a low current fuse or other current activated

protective device to react.

APPLICATION CONSIDERATIONS

In most cases, it is not advisable to place a zener TVS

directly across a data line because of its relatively high

capacitance. The arrangement previously discussed and

shown in Figure 21 works well for an unbalanced line such

as RS-232. When using discrete steering diodes, they should

have low capacitance and low turn-on impedance to avoid

causing an overshoot on the clamped voltage level.

Most noise and transient surge voltages occur on lines

with respect to ground. A signal line such as RS-232 which

uses ground as a signal reference is thus very vulnerable to

noise and transients. It is, however, easy to protect using a

single TVS at each end of the line.

Telephone and RS-422 lines are called balanced lines

because the signal is placed between two lines which are

“floating” with respect to ground. A signal appears between

each signal line and ground but is rejected by the receiver;

only the difference in potential between the two signal lines

is recognized as the transmitted signal. This system has been

in use for decades as a means of providing improved noise

immunity, but protection from transient surge voltages is

more complex.

A cost-effective means of protecting a balanced line is

shown in Figure 23. The bridge diode arrangement allows

protection against both positive and negative transients to be

achieved, an essential requirement but the TVS devices Z1

and Z2 need only be unidirectional. The diodes are chosen

to have low capacitance to reduce loading on the line and

low turn-on impedance to avoid causing an overshoot on the

clamped voltage level. Although a zener TVS is shown, a

TSS is more appropriate when a telephone line having ring

voltages is to be protected.

D3D4

Figure 23. A Method of Reducing Capacitance and

Protecting a Balanced Line

Since transients are usually common-mode, it is important

that the TVS circuit operate in a balanced fashion;

otherwise, common mode transients can cause differential

mode disturbances which can be devastating to the line

receiver. For example, suppose that an identical positive

common mode surge voltage appears from each

line-to-ground. Diodes D2 and D4 will conduct the transients

to Z2. However, if one of these diodes has a slower turn-on

or higher dynamic impedance than the other, the voltage

difference caused by the differing diode response appears

across the signal lines. Consequently, the bridge diodes must

be chosen to be as nearly identical as possible.

Should a differential mode transient appear on the signal

lines, it will be held to twice that of the line-to-ground

clamping level. In many cases a lower clamping level is

needed which can be achieved by placing another TVS

across the signal lines. It must be a bidirectional low

http://onsemi.com

capacitance device. With a line-to-line TVS in the circuit

diode matching is not required.

Other schemes appearing in the literature use two

bidirectional TVS devices from each line-to-ground as

shown in Figure 24 that of the line-to-ground voltage. To

avoid generating a differential mode signal, the TVS must

be closely matched or a third TVS must be placed

line-to-line. By using a third TVS differential mode

transients can be held to a low level.

TIP

RING

Figure 24. Protecting a Balanced Line with

Bidirectional TSS Devices

The arrangement of Figure 25 offers the advantage of

lower capacitance when differential mode transient

protection is required. If all three TSS devices have the same

capacitance (C), the line-to-line and line-to-ground

capacitance of Figure 24 is 1.5C. However, the arrangement

of Figure 25 exhibits a capacitance of only C/2. To design the

circuit to handle the same simultaneous common mode

energy as the circuit of Figure 24, T3 must be twice as large

as T1 and T2. In this case the capacitance of T3 is doubled

which causes the line-to-ground capacitance to be 2C/3, still

a considerable improvement over the arrangement of

Figure 24.

TIP

RING

Figure 25. Preferred Method of Protecting a

Balanced Line Using Bidirectional TSS Devices

Protectors are usually designed to be “fail safe” if their

energy ratings are exceeded, but the definition of “fail safe”

is often dependent upon the application. The most common

requirement is that the surge voltage protective element

should fail short and remain shorted regardless of the

resulting current flow. To insure that this occurs,

semiconductor devices use heavy gauge clips or bonding

wires between the chip and terminals. In addition, parts are

available in plastic packages having a spring type shorting

bar which shorts the terminals when the package softens at

the very high temperatures generated during a severe

overload.

The shorted TVS protects the equipment, but the line

feeding it could be destroyed if the source of energy which

shorted the TVS is from a power cross. Therefore, it is wise

and necessary for a UL listing to provide a series element

such as a fuse or PTC device to either open the circuit or

restrict its value to a safe level.

The design of circuit boards is critical and layout must be

done to minimize any lead or wiring inductance in series

with the TVS. Significant voltage is developed in any loop

subject to transients because of their high current amplitudes

and fast risetimes.

REFERENCES

1. D. W. Bodle and P. A. Gresh, “Lightning Surges in

Paired Telephone Cable Facilities,” The Bell System

Technical Journal, Vol. 4, March 1961, pp. 547−576.

2. D. G. Stroh, “Static Electricity Can Kill Transistors,”

Electronics, Vol. 35, 1962, pp. 90−91.

3. J. D. Norgard and C. L. Chen, “Lightning-Induced

Transients on Buried Shielded Transmission Lines,”

IEEE Transactions on EMC, Vol. EMC28, No. 3,

August 1986, pp. 168−171.

4. O. M. Clark, “Transient Voltage Suppression (TVS),”

1989, pp. 6−7.

5. Clark, p. 7.

6. World Information Technologies, “U. S. Electrical and

Electronic Surge Protection Markets,” 1989, p. 5.

7. T. J. Tucker, “Spark Initiation Requirements of a

Secondary Explosive,” Annals of the New York

Academy of Sciences, Vol. 152, Article I, 1968, pp.

643−653.

8. H. K. Florig, “The Future Battlefield: A Blast of

Gigawatts?,” IEEE Spectrum, Vol. 25, No. 3, March

1988, pp. 50−54.

9. Mel Thrasher, “A Solid State Solution,” Telephony,

June 1989, pp. 48−52.

10. A. Urbieta, “Sensitivity Study to EOS/SSD of Bipolar

Integrated Circuits,” EOS-8, 1986.

11. M. Tetreault and F.D. Martzloff, “Characterization of

Disturbing Transient Waveforms on Computer Data

Communication Lines,” EMC Proceedings, Zurich,

March 1985, pp. 423−428.

12. F. Martzloff, “Coupling, Propagation, and Side Effects

of Surges in an Industrial Building Wiring System,”

http://onsemi.com

Conference Record of IEEE IAS Meeting, 1988, pp.

1467−1475.

IMPORTANT REGULATORY REQUIREMENTS

AND GUIDELINES

GENERAL

DOD-STD-1399, MIL-STD-704, MIL-STD-1275,

MIL-STD-461C. These military specifications are

important, if we intend to target devices for military or

commercial aviation markets.

IEEE 587. This specification describes multiple transient

waveforms.

UL1449. This is a compulsory test which demonstrates

performance to criteria establishing the maximum voltage

that can pass through a device after clamping has taken

place. It is important that we comply, and say so on our data

sheets.

INDUSTRIAL

ANSI/IEEE C62.41. Established by the American

National Standards Institute (ANSI) and the Institute of

Electrical and Electronic Engineers (IEEE), this guideline

tests the effectiveness of devices to typical power

disturbances. To meet the most rigorous category of this

spec, a device or module must be capable of withstanding a

maximum repetitive surge current pulse of 3000 amps with

a 8/20 μs waveform.

IEC TC 102 D. Requirements for remote control

receivers for industrial applications are detailed in this

International Electrotechnical Commission (IEC)

document.

IEC 255-4 and IEC TC 41. These documents describe

testing for static relays for industrial use.

IEC 801-1 thru -3. These are specifications for various

industrial control applications.

IEC 801-4. The IEC specifies transient voltage impulses

which occur from the switching of inductive loads. We must

be aware of the importance of this specification, especially

in Europe, and characterize our devices’ performance to it.

IEEE 472/ANSI C 37.90.1. Requirements for protective

relays, including 10/1000 nS waveform testing is described.

UL943. This requirement defines a 0.5 μs/100 kHz

waveform for ground fault and other switching applications.

VDE 0420. Industrial remote control receivers are

detailed, and test procedures defined.

VDE 0860/Part 1/II. This includes a description of

0.2/200 μs, 10kV test requirements.

TELECOM

CCITT IX K.17, K.20, K.15. These documents relate to

repeaters.

EIA PM-1361. This document covers requirements for

telephone terminals and data processing equipment.

FTZ 4391 TV1. This is a general German specification

for telecom equipment.

FCC Part 68. The Federal Communications Commission

(FCC) requirements for communications equipment is

defined. Of special note is §68.302, dealing with telecom

power lines.

NT/DAS/PRL/003. Telephone instrument, subscriber

equipment and line requirements are documented.

PTT 692.01. This is a general Swiss specification for

telecom exchange equipment.

REA PE-60. The Rural Electrification Administration

(REA) has documented the predominant waveform for

induced lightning transients. This test is now commonly

known as the 10/1000 μs pulse test.

REA PE-80. The REA defines requirements for gas tubes

and similar devices for telecom applications.

TA-TSY-000974. This technical advisory by Bellcore

defines double exponential waveforms, which are the basis

for many telecom applications norms.

UL 1459 and UL 4978. These document detail tests for

standard telephone equipment and data transmission.

TA-TSY-000974. This technical advisory by Bellcore

defines double exponential waveforms, which are the basis

for many telecom applications norms.

UL 1459 and UL 4978. These document detail tests for

standard telephone equipment and data transmission.

March, 2001 − Rev.0

82 Publication Order Number:

AN784/D

Transient Power Capability

of Zener Diodes

Prepared by

Applications Engineering and

Jerry Wilhardt, Product Engineer —

Industrial and Hi-Rel Zener Diodes

INTRODUCTION

Because of the sensitivity of semiconductor components

to voltage transients in excess of their ratings, circuits are

often designed to inhibit voltage surges in order to protect

equipment from catastrophic failure. External voltage

transients are imposed on power lines as a result of lightning

strikes, motors, solenoids, relays or SCR switching circuits,

which share the same ac source with other equipment.

Internal transients can be generated within a piece of

equipment by rectifier reverse recovery transients,

switching of loads or transformer primaries, fuse blowing,

solenoids, etc. The basic relation, v = L di/dt, describes most

equipment developed transients.

ZENER DIODE CHARACTERISTICS

Zener diodes, being nearly ideal clippers (that is, they

exhibit close to an infinite impedance below the clipping

level and close to a short circuit above the clipping level), are

often used to suppress transients. In this type of application,

it is important to know the power capability of the zener for

short pulse durations, since they are intolerant of excessive

stress.

Some ON Semiconductor data sheets such as the ones for

devices shown in Table 1 contain short pulse surge

capability. However, there are many data sheets that do not

contain this data and Figure 1 is presented here to

supplement this information.

Table 1. Transient Suppressor Diodes

Series

Numbers

Steady

State Power Package Description

1N4728 1 W DO-41 Double Slug Glass

1N6267 5 W Case 41A Axial Lead Plastic

1N5333A 5 W Case 17 Surmetic 40

1N746/957

A/4371

400 mW DO-35 Double Slug Glass

1N5221A 500 mW DO-35 Double Slug Glass

Some data sheets have surge information which differs

slightly from the data shown in Figure 1. A variety of reasons

exist for this:

1. The surge data may be presented in terms of actual surge

power instead of nominal power.

2. Product improvements have occurred since the data

sheet was published.

Figure 1. Peak Power Ratings of Zener Diodes

Power is defined as VZ(NOM) x IZ(PK) where VZ(NOM) is the nominal

zener voltage measured at the low test current used for voltage

classification.

1N6267 SERIES

GLASS DO-35 & GLASS DO-41

250 mW TO 1 W TYPES

5 WATT TYPES

PULSE WIDTH (ms)

0.1

100

0.01 0.02

PPK(NOM), NOMINAL PEAK POWER (kW)

0.5

0.2

0.1

0.05

0.02

0.01 0.05 0.2 0.5 1 2 5 10

1 TO 3 W TYPES

PLASTIC DO-41

3. Larger dice are used, or special tests are imposed on the

product to guarantee higher ratings than those shown on

Figure 1.

4. The specifications may be based on a JEDEC

registration or part number of another manufacturer.

The data of Figure 1 applies for non-repetitive conditions

and at a lead temperature of 25°C. If the duty cycle increases,

the peak power must be reduced as indicated by the curves

of Figure 2. Average power must be derated as the lead or

ambient temperature rises above 25°C. The average power

derating curve normally given on data sheets may be

normalized and used for this purpose.

At first glance the derating curves of Figure 2 appear to be

in error as the 10 ms pulse has a higher derating factor than

the 10 μs pulse. However, when the derating factor for a

given pulse of Figure 2 is multiplied by the peak power value

of Figure 1 for the same pulse, the results follow the

expected trend.

http://onsemi.com

APPLICATION NOTE

AN784/D

http://onsemi.com

When it is necessary to use a zener close to surge ratings,

and a standard part having guaranteed surge limits is not

suitable, a special part number may be created having a surge

limit as part of the specification. Contact your nearest

ON Semiconductor OEM sales office for capability, price,

delivery, and minimum order criteria.

MATHEMATICAL MODEL

Figure 2. Typical Derating Factor for Duty Cycle

0.1

0.7

0.5

0.3

0.2

0.02

0.1

0.07

0.05

0.03

0.01 0.2 0.5 1 52 10 20 50 100

PULSE WIDTH

10 ms

1 ms

100 μs

10 μs

D, DUTY CYCLE (%)

DERATING FACTOR

Since the power shown on the curves is not the actual

transient power measured, but is the product of the peak

current measured and the nominal zener voltage measured

at the current used for voltage classification, the peak current

can be calculated from:

IZ(PK) = P(PK)

VZ(NOM)

(1)

The peak voltage at peak current can be calculated from:

(2)

VZ(PK) = FC × VZ(NOM)

where FC is the clamping factor. The clamping factor is

approximately 1.20 for all zener diodes when operated at

their pulse power limits. For example, a 5 watt, 20 volt zener

can be expected to show a peak voltage of 24 volts regardless

of whether it is handling 450 watts for 0.1 ms or 50 watts for

10 ms. This occurs because the voltage is a function of

junction temperature and IR drop. Heating of the junction is

more severe at the longer pulse width, causing a higher

voltage component due to temperature which is roughly

offset by the smaller IR voltage component.

For modeling purposes, an approximation of the zener

resistance is needed. It is obtained from:

RZ(NOM) = VZ(NOM)(FC-1)

PPK(NOM)/VZ(NOM)

(3)

The value is approximate because both the clamping

factor and the actual resistance are a function of temperature.

CIRCUIT CONSIDERATIONS

It is important that as much impedance as circuit

constraints allow be placed in series with the zener diode and

the components to be protected. The result will be a lower

clipping voltage and less zener stress. A capacitor in parallel

with the zener is also effective in reducing the stress imposed

by very short duration transients.

To illustrate use of the data, a common application will be

analyzed. The transistor in Figure 3 drives a 50 mH solenoid

which requires 5 amperes of current. Without some means

of clamping the voltage from the inductor when the

transistor turns off, it could be destroyed.

Figure 3. Circuit Example

Used to select a zener diode having the proper voltage

and power capability to protect the transistor.

10 ms

2 s

26 Vdc

50 mH, 5 Ω

The means most often used to solve the problem is to

connect an ordinary rectifier diode across the coil; however,

this technique may keep the current circulating through the

coil for too long a time. Faster switching is achieved by

allowing the voltage to rise to a level above the supply before

being clamped. The voltage rating of the transistor is 60 V,

indicating that approximately a 50 volt zener will be

required.

The peak current will equal the on-state transistor current

(5 amperes) and will decay exponentially as determined by

the coil L/R time constant (neglecting the zener impedance).

A rectangular pulse of width L/R (0.01 sec) and amplitude

of IPK (5 A) contains the same energy and may be used to

select a zener diode. The nominal zener power rating

therefore must exceed (5 A × 50) = 250 watts at 10 ms and

a duty cycle of 0.01/2 = 0.5%. From Figure 2, the duty cycle

factor is 0.62 making the single pulse power rating required

equal to 250/0.62 = 403 watts. From Figure 1, one of the

1N6267 series zeners has the required capability. The

1N6287 is specified nominally at 47 volts and should prove

satisfactory.

Although this series has specified maximum voltage

limits, equation 3 will be used to determine the maximum

zener voltage in order to demonstrate its use.

RZ = 47(1.20 − 1)

500/47

9.4

10.64

0.9Ω

At 5 amperes, the peak voltage will be 4.5 volts above

nominal or 51.5 volts total which is safely below the 60 volt

transistor rating.

March, 2001 − Rev.0

84 Publication Order Number:

AN843/D

A Review of Transients and

Their Means of Suppression

Prepared by

Steve Cherniak

Applications Engineering

INTRODUCTION

One problem that most, if not all electronic equipment

designers must deal with, is transient overvoltages.

Transients in electrical circuits result from the sudden

release of previously stored energy. Some transients may be

voluntary and created in the circuit due to inductive

switching, commutation voltage spikes, etc. and may be

easily suppressed since their energy content is known and

predictable. Other transients may be created outside the

circuit and then coupled into it. These can be caused by

lightning, substation problems, or other such phenomena.

These transients, unlike switching transients, are beyond the

control of the circuit designer and are more difficult to

identify, measure and suppress.

Effective transient suppression requires that the impulse

energy is dissipated in the added suppressor at a low enough

voltage so the capabilities of the circuit or device will not be

exceeded.

REOCCURRING TRANSIENTS

Transients may be formed from energy stored in circuit

inductance and capacitance when electrical conditions in the

circuit are abruptly changed.

Switching induced transients are a good example of this;

the change in current ǒdi

dtǓ in an inductor (L) will generate

a voltage equal to Ldi

dt. The energy (J) in the transient is equal

to 1/2Li2 and usually exists as a high power impulse for a

relatively short time (J = Pt).

If load 2 is shorted (Figure 1), devices parallel to it may

be destroyed. When the fuse opens and interrupts the fault

current, the slightly inductive power supply produces a

transient voltage spike of V+Ldi

dt with an energy content of

J = 1/2Li2. This transient might be beyond the voltage

limitations of the rectifiers and/or load 1. Switching out a

high current load will have a similar effect.

TRANSFORMER PRIMARY

BEING ENERGIZED

If a transformer is energized at the peak of the line voltage

(Figure 2), this voltage step function can couple to the stray

capacitance and inductance of the secondary winding and

generate an oscillating transient voltage whose oscillations

depend on circuit inductance and capacitance. This

transient’s peak voltage can be up to twice the peak

amplitude of the normal secondary voltage.

In addition to the above phenomena the capacitively

coupled (CS) voltage spike has no direct relationship with

the turns ratio, so it is possible for the secondary circuit to see

the peak applied primary voltage.

Figure 1. Load Dump with Inductive Power Supply

POWER

SUPPLY

BLOAD

LOAD

−

SHORT

ACROSS

LOAD 2

FUSE

VAB

−

http://onsemi.com

APPLICATION NOTE

AN843/D

http://onsemi.com

Figure 2. Situation Where Transformer Capacitance Causes a Transient

−

VLine

PEAK SHOULD

BE 30% LIGHTER

SWITCH

CLOSED

VAB

LOAD

MAY HAVE STRAY

INDUCTANCE OR

CAPACITANCE

SWITCH

TRANSFORMER PRIMARY

BEING DE-ENERGIZED

If the transformer is driving a high impedance load,

transients of more than ten times normal voltage can be

created at the secondary when the primary circuit of the

transformer is opened during zero-voltage crossing of the ac

line. This is due to the interruption of the transformer

magnetizing current which causes a rapid collapse of the

magnetic flux in the core. This, in turn, causes a high voltage

transient to be coupled into the transformer’s secondary

winding (Figure 3).

Transients produced by interrupting transformers

magnetizing current can be severe. These transients can

destroy a rectifier diode or filter capacitor if a low

impedance discharge path is not provided.

SWITCH “ARCING”

When a contact type switch opens and tries to interrupt

current in an inductive circuit, the inductance tries to keep

current flowing by charging stray capacitances. (See

Figure 4.)

Figure 3. Typical Situation Showing Possible Transient When Interrupting

Transformer Magnetizing Current

Figure 4. Transients Caused by Switch Opening

SWITCH

LOAD

LINE

SWITCH

OPENED

LINE

VOLTAGE

MAGNETIZING

CURRENT AND

FLUX

SECONDARY

VOLTAGE

VCap

VLine

TRANSIENT

SENSITIVE

LOAD

LINE

VOLTAGE

AN843/D

http://onsemi.com

This can also happen when the switch contacts bounce

open after its initial closing. When the switch is opened (or

bounces open momentarily) the current that the inductance

wants to keep flowing will oscillate between the stray

capacitance and the inductance. When the voltage due to the

oscillation rises at the contacts, breakdown of the contact

gap is possible, since the switch opens (or bounces open)

relatively slowly compared to the oscillation frequency, and

the distance may be small enough to permit “arcing.” The

arc will discontinue at the zero current point of the

oscillation, but as the oscillatory voltage builds up again and

the contacts move further apart, each arc will occur at a

higher voltage until the contacts are far enough apart to

interrupt the current.

WAVESHAPES OF SURGE VOLTAGES

Indoor Waveshapes

Measurements in the field, laboratory, and theoretical

calculations indicate that the majority of surge voltages in

indoor low-voltage power systems have an oscillatory

waveshape. This is because the voltage surge excites the

natural resonant frequency of the indoor wiring system. In

addition to being typically oscillatory, the surges can also

have different amplitudes and waveshapes in the various

places of the wiring system. The resonant frequency can

range from about 5 kHz to over 500 kHz. A 100 kHz

frequency is a realistic value for a typical surge voltage for

most residential and light industrial ac wire systems.

The waveshape shown in Figure 5 is known as an “0.5 μs

− 100 kHz ring wave.” This waveshape is reasonably

representative of indoor low-voltage (120 V − 240 V) wiring

system transients based on measurements conducted by

several independent organizations. The waveshape is

defined as rising from 10% to 90% of its final amplitude in

0.5 μs, then decays while oscillating at 100 kHz, each peak

being 60% of the preceding one.

The fast rise portion of the waveform can induce the

effects associated with non-linear voltage distribution in

windings or cause dv/dt problems in semiconductors.

Shorter rise times can be found in transients but they are

lengthened as they propagate into the wiring system or

reflected from wiring discontinuities.

Figure 5. 0.5 μs 100 kHz Ring Wave

0.9 Vpk

Vpk

t = 10 μs (f = 100 kHz)

0.1 Vpk

0.5 μs

60% OF Vpk

The oscillating portion of the waveform produces voltage

polarity reversal effects. Some semiconductors are sensitive

to polarity changes or can be damaged when forced into or

out of conduction (i.e. reverse recovery of rectifier devices).

The sensitivity of some semiconductors to the timing and

polarity of a surge is one of the reasons for selecting this

oscillatory waveform to represent actual conditions.

Outdoor Locations

Both oscillating and unidirectional transients have been

recorded in outdoor environments (service entrances and

other places nearby). In these locations substantial energy is

still available in the transient, so the waveform used to

model transient conditions outside buildings must contain

greater energy than one used to model indoor transient

surges.

Properly selected surge suppressors have a good

reputation of successful performance when chosen in

conjunction with the waveforms described in Figure 6. The

recommended waveshape of 1.2 × 50 μs (1.2 μs is associated

with the transients rise time and the 50 μs is the time it takes

for the voltage to drop to 1/2Vpk) for the open circuit voltage

and 8 × 20 μs for the short circuit current are as defined in

IEEE standard 28-ANSI Standard C62.1 and can be

considered a realistic representation of an outdoor transients

waveshape.

Figure 6. Unidirectional Wave Shapes

0.9 Vpk

0.3 Vpk

0.1 Vpk

Vpk

0.5 Vpk

50 μs

t1t1 × 1.67 = 1.2 μs

IIpk

0.9 Ipk

0.1 Ipk

0.5 Ipk

t220 μst2 × 1.25 = 8 μs

(a) Open-Circuit Voltage Waveform (b) Discharge Current Waveform

AN843/D

http://onsemi.com

The type of device under test determines which

waveshape in Figure 6 is more appropriate. The voltage

waveform is normally used for insulation voltage withstand

tests and the current waveform is usually used for discharge

current tests.

RANDOM TRANSIENTS

The source powering the circuit or system is frequently

the cause of transient induced problems or failures. These

transients are difficult to deal with due to their nature; they

are totally random and it is difficult to define their

amplitude, duration and energy content. These transients are

generally caused by switching parallel loads on the same

branch of a power distribution system and can also be caused

by lightning.

AC POWER LINE TRANSIENTS

Transients on the ac power line range from just above

normal voltage to several kV. The rate of occurrence of

transients varies widely from one branch of a power

distribution system to the next, although low-level transients

occur more often than high-level surges.

Data from surge counters and other sources is the basis for

the curves shown in Figure 7. This data was taken from

unprotected (no voltage limiting devices) circuits meaning

that the transient voltage is limited only by the sparkover

distance of the wires in the distribution system.

Figure 7. Peak Surge Voltage versus Surges per Year*

*EIA paper, P587.1/F, May, 1979, Page 10

0.7

0.2

0.1

0.5

0.3

0.9

0.8

0.4

0.6

0.01 10.1 10 100 1000

SURGES PER YEAR

(kV)

HIGH EXPOSURE

MEDIUM EXPOSURE

LOW EXPOSURE

AN843/D

http://onsemi.com

The low exposure portion of the set of curves is data

collected from systems with little load-switching activity

that are located in areas of light lightning activity.

Medium exposure systems are in areas of frequent

lightning activity with a severe switching transient problem.

High exposure systems are rare systems supplied by long

overhead lines which supply installations that have high

sparkover clearances and may be subject to reflections at

power line ends.

When using Figure 7 it is helpful to remember that peak

transient voltages will be limited to approximately 6 kV in

indoor locations due to the spacing between conductors

using standard wiring practices.

TRANSIENT ENERGY LEVELS AND

SOURCE IMPEDANCE

The energy contained in a transient will be divided between

the transient suppressor and the source impedance of the

transient in a way that is determined by the two impedances.

With a spark-gap type suppressor, the low impedance of the

Arc after breakdown forces most of the transient’s energy to

be dissipated elsewhere, e.g. in a current limiting resistor in

series with the spark-gap and/or in the transient’s source

impedance. Voltage clamping suppressors (e.g. zeners,

mov’s, rectifiers operating in the breakdown region) on the

other hand absorb a large portion of the transient’s surge

energy. So it is necessary that a realistic assumption of the

transient’s source impedance be made in order to be able to

select a device with an adequate surge capability.

The 100 kHz “Ring Wave” shown in Figure 5 is intended

to represent a transient’s waveshape across an open circuit.

The waveshape will change when a load is connected and the

amount of change will depend on the transient’s source

impedance. The surge suppressor must be able to withstand

the current passed through it from the surge source. An

assumption of too high a surge impedance (when testing the

suppressor) will not subject the device under test to

sufficient stresses, while an assumption of too low a surge

impedance may subject it to an unrealistically large stress;

there is a trade-off between the size (cost) of the suppressor

and the amount of protection obtained.

In a building, the transient’s source impedance increases

with the distance from the electrical service entrance, but

open circuit voltages do not change very much throughout

the structure since the wiring does not provide much

attenuation. There are three categories of service locations

that can represent the majority of locations from the

electrical service entrance to the most remote wall outlet.

These are listed below. Table 1 is intended as an aid for the

preliminary selection of surge suppression devices, since it

is very difficult to select a specific value of source

impedance.

Category I: Outlets and circuits a “long distance” from

electrical service entrance. Outlets more than 10 meters

from Category II or more than 20 meters from Category

III (wire gauge #14 − #10)

Category II: Major bus lines and circuits a “short distance”

from electrical service entrance. Bus system in industrial

plants. Outlets for heavy duty appliances that are “close”

to the service entrance.

Distribution panel devices.

Commercial building lighting systems.

Category III. Electrical service entrance and outdoor

locations.

Power line between pole and electrical service entrance.

Power line between distribution panel and meter.

Power line connection to additional near-by buildings.

Underground power lines leading to pumps, filters, etc.

Categories I and II in Table 1 correspond to the extreme

range of the “medium exposure” curve in Figure 7. The

surge voltage is limited to approximately 6 kV due to the

sparkover spacing of indoor wiring.

The discharge currents of Category II were obtained from

simulated lightning tests and field experience with

suppressor performance.

The surge currents in Category I are less than in Category

II because of the increase in surge impedance due to the fact

that Category I is further away from the service entrance.

Category III can be compared to the “High Exposure”

situation in Figure 7. The limiting effect of sparkover is not

available here so the transient voltage can be quite large.

Table 1. Surge Voltages and Currents Deemed to Represent the Indoor Environment Depending Upon Location

Energy (Joules) Dissipated in a

Suppressor with a Clamping Voltage of(3)

Category Waveform Surge Voltage(1) Surge Current(2) 250 V 500 V 1000 V

I0.5 μs 100 kHz

Ring Wave

6 kV 200 A 0.4 0.8 1.6

II 0.5 μs 100 kHz

Ring Wave

6 kV 500 A 1 2 4

1.2 × 50 μs

8 × 20 μs

6 kV

3 kA 20 40 80

III 1.2 × 50 μs

8 × 20 μs

10 kV or more

10 kA or more

Notes: 1. Open Circuit voltage

Notes: 2. Discharge current of the surge (not the short circuit current of the power system)

Notes: 3. The energy a suppressor will dissipate varies in proportion with the suppressor’s clamping voltage, which can be different with different system voltages (assuming the same

Notes: 3. discharge current).

AN843/D

http://onsemi.com

LIGHTNING TRANSIENTS

There are several mechanisms in which lightning can

produce surge voltages on power distribution lines. One of

them is a direct lightning strike to a primary (before the

substation) circuit. When this high current, that is injected

into the power line, flows through ground resistance and the

surge impedance of the conductors, very large transient

voltages will be produced. If the lightning misses the

primary power line but hits a nearby object the lightning

discharge may also induce large voltage transients on the

line. When a primary circuit surge arrester operates and

limits the primary voltage the rapid dv/dt produced will

effectively couple transients to the secondary circuit through

the capacitance of the transformer (substation) windings in

addition to those coupled into the secondary circuit by

normal transformer action. If lightning struck the secondary

circuit directly, very high currents may be involved which

would exceed the capability of conventional surge

suppressors. Lightning ground current flow resulting from

nearby direct to ground discharges can couple onto the

common ground impedance paths of the grounding

networks also causing transients.

AUTOMOTIVE TRANSIENTS

Transients in the automotive environment can range from

the noise generated by the ignition system and the various

accessories (radio, power window, etc.) to the potentially

destructive high energy transients caused by the charging

(alternator/regulator) system. The automotive “Load

Dump” can cause the most destructive transients; it is when

the battery becomes disconnected from the charging system

during high charging rates. This is not unlikely when one

considers bad battery connections due to corrosion or other

wiring problems. Other problems can exist such as steady

state overvoltages caused by regulator failure or 24 V

battery jump starts. There is even the possibility of incorrect

battery connection (reverse polarity).

Capacitive and/or inductive coupling in wire harnesses as

well as conductive coupling (common ground) can transmit

these transients to the inputs and outputs of automotive

electronics.

The Society of Automotive Engineers (SAE) documented

a table describing automotive transients (see Table 2) which

is useful when trying to provide transient protection.

Considerable variation has been observed while gathering

data on automobile transients. All automobiles have their

electrical systems set up differently and it is not the intent of

this paper to suggest a protection level that is required. There

will always be a trade-off between cost of the suppressor and

the level of protection obtained. The concept of one master

suppressor placed on the main power lines is the most

cost-effective scheme possible since individual suppressors

at the various electronic devices will each have to suppress

the largest transient that is likely to appear (Load Dump),

hence each individual suppressor would have to be the same

size as the one master suppressor since it is unlikely for

several suppressors to share the transient discharge.

Table 2. Typical Transients Encountered in the Automotive Environment

Length of

Transient Cause

Energy Capability

Voltage Amplitude

Possible Frequency

of Application

Steady State Failed Voltage Regulator

Booster starts with 24 V battery

Load Dump — i.e., disconnection of battery during

high charging rates

Inductive Load Switching Transient

Alternator Field Decay

Ignition Pulse

Disconnected Battery

Mutual Coupling in Harness

Ignition Pulse Normal

Accessory Noise

Transceiver Feedback

5 Minutes

4.5−100 ms

≤ 0.32 s

≤ 0.2 s

90 ms

1 ms

15 μs

∞

+18 V Infrequent

∞

±24 V

≥10 J

≤ 125 V

Infrequent

< 1 J

−300 V to +80 V Often

< 1 J

−100 V to −40 V

Each Turn-Off

≤ 500 Hz

Several Times in

vehicle life

< 0.5 J

≤ 75 V

< 1 J

≤ 200 V

< 0.001 J

3 V

≤ 1.5 V

≈ 20 mV

Often

3 500 Hz

Continuous

50 Hz to 10 kHz

R.F.

AN843/D

http://onsemi.com

There will, of course, be instances where a need for

individual suppressors at the individual accessories will be

required, depending on the particular wiring system or

situation.

TRANSIENT SUPPRESSOR TYPES

Carbon Block Spark Gap

This is the oldest and most commonly used transient

suppressor in power distribution and telecommunication

systems. The device consists of two carbon block electrodes

separated by an air gap, usually 3 to 4 mils apart. One

electrode is connected to the system ground and the other to

the signal cable conductor. When a transient over-voltage

appears, its energy is dissipated in the arc that forms between

the two electrodes, a resistor in series with the gap, and also

in the transient’s source impedance, which depends on

conductor length, material and other parameters.

The carbon block gap is a fairly inexpensive suppressor

but it has some serious problems. One is that it has a

relatively short service life and the other is that there are

large variations in its arcing voltage. This is the major

problem since a nominal 3 mil gap will arc anywhere from

300 to 1000 volts. This arcing voltage variation limits its use

mainly to primary transient suppression with more accurate

suppressors to keep transient voltages below an acceptable

level.

Gas Tubes

The gas tube is another common transient suppressor,

especially in telecommunication systems. It is made of two

metallic conductors usually separated by 10 to 15 mils

encapsulated in a glass envelope which is filled with several

gases at low pressure. Gas tubes have a higher current

carrying capability and longer life than carbon block gaps.

The possibility of seal leakage and the resultant of loss

protection has limited the use of these devices.

Selenium Rectifiers

Selenium transient suppressors are selenium rectifiers

used in the reverse breakdown mode to clamp voltage

transients. Some of these devices have self-healing

properties which allows the device to survive energy

discharges greater than their maximum capability for a

limited number of surges. Selenium rectifiers do not have

the voltage clamping capability of zener diodes. This is

causing their usage to become more and more limited.

METAL OXIDE VARISTORS (MOV’S)

An MOV is a non-linear resistor which is voltage

dependent and has electrical characteristics similar to

back-to-back zener diodes. As its name implies it is made up

of metal oxides, mostly zinc oxide with other oxides added

to control electrical characteristics. MOV characteristics are

compared to back-to-back zeners in Photos 2 through 7.

When constructing MOV’s the metal oxides are sintered

at high temperatures to produce a polycrystalline structure

of conductive zinc oxide separated by highly resistive

intergranular boundaries. These boundaries are the source of

the MOV’s non-linear electrical behavior.

MOV electrical characteristics are mainly controlled by

the physical dimensions of the polycrystalline structure

since conduction occurs between the zinc oxide grains and

the intergranular boundaries which are distributed

throughout the bulk of the device.

The MOV polycrystalline body is usually formed into the

shape of a disc. The energy rating is determined by the

device’s volume, voltage rating by its thickness, and current

handling capability by its area. Since the energy dissipated

in the device is spread throughout its entire metal oxide

volume, MOV’s are well suited for single shot high power

transient suppression applications where good clamping

capability is not required.

The major disadvantages with using MOV’s are that they

can only dissipate relatively small amounts of average

power and are not suitable for many repetitive applications.

Another drawback with MOV’s is that their voltage

clamping capability is not as good as zeners, and is

insufficient in many applications.

Perhaps the major difficulty with MOV’s is that they have

a limited life time even when used below their maximum

ratings. For example, a particular MOV with a peak current

handling capability of 1000 A has a lifetime of about 1 surge

at 1000 Apk, 100 surges at 100 Apk and approximately 1000

surges at 65 Apk.

TRANSIENT SUPPRESSION

USING ZENERS

Zener diodes exhibit a very high impedance below the

zener voltage (VZ), and a very low impedance above VZ.

Because of these excellent clipping characteristics, the zener

diode is often used to suppress transients. Zeners are

intolerant of excessive stress so it is important to know the

power handling capability for short pulse durations.

Most zeners handle less than their rated power during

normal applications and are designed to operate most

effectively at this low level. Zener transient suppressors

such as the ON Semiconductor 1N6267 Mosorb series are

designed to take large, short duration power pulses.

This is accomplished by enlarging the chip and the

effective junction area to withstand the high energy surges.

The package size is usually kept as small as possible to

provide space efficiency in the circuit layout, and since the

package does not differ greatly from other standard zener

packages, the steady state power dissipation does not differ

greatly.

Some data sheets contain information on short pulse surge

capability. When this information is not available for ON

Semiconductor devices, Figure 8 can be used. This data

applies for non-repetitive conditions with a lead temperature

of 25°C.

It is necessary to determine the pulse width and peak

power of the transient being suppressed when using

Figure 8. This can be done by taking whatever waveform

AN843/D

http://onsemi.com

the transient is and approximating it with a rectangular pulse

with the same peak power. For example, an exponential

discharge with a 1 ms time constant can be approximated by

a rectangular pulse 1 ms wide that has the same peak power

as the transient. This would be a better approximation than

a rectangular pulse 10 ms wide with a correspondingly lower

amplitude. This is because the heating effects of different

pulse width lengths affect the power handling capability, as

can be seen by Figure 8. This also represents a conservative

approach because the exponential discharge will contain ≈

1/2 the energy of a rectangular pulse with the same pulse

width and amplitude.

Figure 8. Peak Power Ratings of Zener Diodes

1N6267 SERIES

GLASS DO-35 & GLASS DO-41

250 mW TO 1 W TYPES

5 WATT TYPES

PULSE WIDTH (ms)

0.1

100

0.01 0.02

PPK(NOM), NOMINAL PEAK POWER (kW)

0.5

0.2

0.1

0.05

0.02

0.01

0.05 0.2 0.5 1 2 5 10

1 TO 3 W TYPES

PLASTIC DO-41

When used in repetitive applications, the peak power must

be reduced as indicated by the curves of Figure 9. Average

power must be derated as the lead or ambient temperature

exceeds 25°C. The power derating curve normally given on

data sheets can be normalized and used for this purpose.

Figure 9. Typical Derating Factor for Duty Cycle

0.1

0.7

0.5

0.3

0.2

0.02

0.1

0.07

0.05

0.03

0.01 0.2 0.5 1 52 10 20 50

PULSE WIDTH

10 ms

1 ms

100 μs

10 μs

D, DUTY CYCLE (%)

DERATING FACTOR

The peak zener voltage during the peak current of the

transient being suppressed can be related to the nominal

zener voltage (Eqtn 1) by the clamping factor (FC).

Eqtn 1: VZ(pK) = FC (VZ(nom))

Unless otherwise specified FC is approximately 1.20 for

zener diodes when operated at their pulse power limits.

For example, a 5 watt, 20 volt zener can be expected to

show a peak voltage of 24 volts regardless of whether it is

handling 450 watts for 0.1 ms or 50 watts for 10 ms. (See

Figure 8.)

This occurs because the zener voltage is a function of both

junction temperature and IR drop. Longer pulse widths

cause a greater junction temperature rise than short ones; the

increase in junction temperature slightly increases the zener

voltage. This increase in zener voltage due to heating is

roughly offset by the fact that longer pulse widths of

identical energy content have lower peak currents. This

results in a lower IR drop (zener voltage drop) keeping the

clamping factor relatively constant with various pulse

widths of identical energy content.

An approximation of zener impedance is also helpful in

the design of transient protection circuits. The value of

RZ(nom) (Eqtn 2) is approximate because both the clamping

factor and the actual resistance is a function of temperature.

Eqtn 2: RZ(nom) =

V2Z(nom) (FC −1)

PpK(nom)

VZ(nom) = Nominal Zener Voltage

PpK(nom) = Found from Figure 8 when device type and

pulse width are known. For example, from Figure 8 a

1N6267 zener suppressor has a PpK(nom) of 1.5 kW at a

pulse width of 1 ms.

As seen from equation 2, zeners with a larger PpK(nom)

capability will have a lower RZ(nom).

ZENER versus MOV TRADEOFFS

The clamping characteristics of Zeners and MOV’s are

best compared by measuring their voltages under transient

conditions. Photos 1 through 9 are the result of an

experiment that was done to compare the clamping

characteristics of a Zener (ON Semiconductor 1N6281,

approximately 1.5J capability) with those of an MOV (G.E.

V27ZA4, 4J capability); both are 27 V devices.

Photo 1 shows the pulse generator output voltage. This

generator synthesizes a transient pulse that is characteristic

of those that may appear in the real world.

Photos 2 and 3 are clamping voltages of the MOV and

Zener, respectively with a surge source impedance of

500 Ω.

Photos 4 and 5 are the clamping voltages with a surge

source impedance of 50 Ω.

Photos 6 and 7 simulate a condition where the surge

source impedance is 5 Ω.

Photos 8 and 9 show a surge source impedance of 0.55 Ω,

which is at the limits of the Zener suppressor’s capability.

AN843/D

http://onsemi.com

PHOTO 1

Open Circuit Transient Pulse

Vert: 20 V/div

Horiz: 0.5 ms/div

Vpeak = 90 V

PHOTO 2

MOV (27 V)

Vert: 10 V/div

Horiz: 0.5 ms/div

Transient Source Impedance: 500 Ω

Vpeak = 39.9 V

AN843/D

http://onsemi.com

PHOTO 3

Zener (27 V)

Vert: 10 V/div

Horiz: 0.5 ms/div

Transient Source Impedance: 500 Ω

Vpeak = 27 V

PHOTO 4

MOV (27 V)

Vert: 10 V/div

Horiz: 0.5 ms/div

Transient Source Impedance: 50 Ω

Vpeak = 44.7 V

PHOTO 5

Zener (27 V)

Vert: 10 V/div

Horiz: 0.5 ms/div

Transient Source Impedance: 50 Ω

Vpeak = 27 V

AN843/D

http://onsemi.com

PHOTO 6

MOV (27 V)

Vert: 10 V/div

Horiz: 0.5 ms/div

Transient Source Impedance: 5 Ω

Vpeak = 52 V

PHOTO 7

Zener (27 V)

Vert: 10 V/div

Horiz: 0.5 ms/div

Transient Source Impedance: 5 Ω

Vpeak = 28 V

PHOTO 8

MOV (27 V)

Vert: 10 V/div

Horiz: 0.5 ms/div

Transient Source Impedance: 0.55 Ω

Vpeak = 62.5 V

AN843/D

http://onsemi.com

PHOTO 9

Zener (27 V)

Vert: 10 V/div

Horiz: 0.5 ms/div

Transient Source Impedance: 0.55 Ω

Vpeak: 30.2 V

Peak Power: Approx 2000 Wpeak

(The limit of this device’s capability)

As can be seen by the photographs, the Zener suppressor

has significantly better voltage clamping characteristics

than the MOV even though that particular Zener has less

than one-fourth the energy capability of the MOV it was

compared with. However, the energy rating can be

misleading because it is based on the clamp voltage times the

surge current, and when using an MOV, the high impedance

results in a fairly high clamp voltage. The major tradeoff

with using a zener type suppressor is its cost versus power

handling capability, but since it would take an “oversized”

MOV to clamp voltages (suppress transients) as well as the

zener, the MOV begins to lose its cost advantage.

If a transient should come along that exceeds the

capabilities of the particular Zener, or MOV, suppressor that

was chosen, the load will still be protected, since they both

fail short.

The theoretical reaction time for Zeners is in the

picosecond range, but this is slowed down somewhat with

lead and package inductance. The 1N6267 Mosorb series of

transient suppressors have a typical response time of less

than one nanosecond. For very fast rising transients it is

important to minimize external inductances (due to wiring,

etc.) which will minimize overshoot.

Connecting Zeners in a back-to-back arrangement will

enable bidirectional voltage clamping characteristics. (See

Figure 10.)

If Zeners A and B are the same voltage, a transient of

either polarity will be clamped at approximately that voltage

since one Zener will be in reverse bias mode while the other

will be in the forward bias mode. When clamping low

voltage it may be necessary to consider the forward drop of

the forward biased Zener.

The typical protection circuit is shown in Figure 11a. In

almost every application, the transient suppression device is

placed in parallel with the load, or component to be

protected. Since the main purpose of the circuit is to clamp

the voltage appearing across the load, the suppressor should

be placed as close to the load as possible to minimize

overshoot due to wiring (or any inductive) effect. (See

Figure 11b.)

Figure 10. Zener Arrangement for

Bidirectional Clamping

Figure 11a. Using Zener to Protect Load

Against Transients

ÎÎÎ

Vin

Zin

LOAD VL

Figure 11b. Overshoot Due to Inductive Effect

ZENER

VOLTAGE

PEAK VOLTAGE

DUE TO OVERSHOOT

TRANSIENT

INPUT

AN843/D

http://onsemi.com

Zener capacitance prior to breakdown is quite small (for

example, the 1N6281 27 Volt Mosorb has a typical

capacitance of 800 pF). Capacitance this small is desirable

in the off-state since it will not attenuate wide-band signals.

When the Zener is in the breakdown mode of operation

(e.g. when suppressing a transient) its effective capacitance

increases drastically from what it was in the off-state. This

makes the Zener ideal for parallel protection schemes since,

during transient suppression, its large effective capacitance

will tend to hold the voltage across the protected element

constant; while in the off-state (normal conditions, no

transient present), its low off-state capacitance will not

attenuate high frequency signals.

Input impedance (Zin) always exists due to wiring and

transient source impedance, but Zin should be increased as

much as possible with an external resistor, if circuit

constraints allow. This will minimize Zener stress.

CONCLUSION

The reliable use of semiconductor devices requires that

the circuit designer consider the possibility of transient

overvoltages destroying these transient-sensitive

components.

These transients may be generated by normal circuit

operations such as inductive switching circuits, energizing

and deenergizing transformer primaries, etc. They do not

present much of a problem since their energy content,

duration and effect may easily be obtained and dealt with.

Random transients found on power lines, or lightning

transients, present a greater threat to electronic components

since there is no way to be sure when or how severe they will

be. General guidelines were discussed to aid the circuit

designer in deciding what size (capability and cost)

suppressor to choose for a certain level of protection. There

will always be a tradeoff between suppressor price and

protection obtained.

Several different suppression devices were discussed with

emphasis on Zeners and MOV’s, since these are the most

popular devices to use in most applications.

REFERENCES

1. GE Transient Voltage Suppression Manual, 2nd

edition.

2. ON Semiconductor Zener Diode Manual.

http://onsemi.com

DESIGN CONSIDERATIONS AND PERFORMANCE

OF ON SEMICONDUCTOR TEMPERATURE-COMPENSATED

ZENER (REFERENCE) DIODES

Prepared by

Zener Diode Engineering

and

Ronald N. Racino

Reliability and Quality Assurance

INTRODUCTION

This application note defines ON Semiconductor

temperature-compensated zener (reference) diodes,

explains the device characteristics, describes electrical

testing, and discusses the advanced concepts of device

reliability and quality assurance. It is a valuable aid to those

who contemplate designing circuits requiring the use of

these devices.

Zener diodes fall into three general classifications:

Regulator diodes, reference diodes and transient voltage

suppressors. Regulator diodes are normally employed in

power supplies where a nearly constant dc output voltage is

required despite relatively large changes in input voltage or

load resistance. Such devices are available with a wide range

of voltage and power ratings, making them suitable for a

wide variety of electronic equipments.

Regulator diodes, however, have one limitation: They are

temperature-sensitive. Therefore, in applications in which

the output voltage must remain within narrow limits during

input-voltage, load-current, and temperature changes, a

temperature-compensated regulator diode, called a

reference diode, is required.

The reference diode is made possible by taking advantage

of the differing thermal characteristics of forward- and

reverse-biased silicon p-n junctions. A forward-biased

junction has a negative temperature coefficient of

approximately 2 mV/°C, while reverse-biased junctions

have positive temperature coefficients ranging from about

2 mV/°C at 5.5 V to 6 mV/°C at 10 V. Therefore it is

possible, by judicious combination of forward- and

reverse-biased junctions, to fabricate a device with a very

low overall temperature coefficient (Figure 1).

The principle of temperature compensation is further

illustrated in Figure 2, which shows the voltage-current

characteristics at two temperature points (25 and 100°C) for

both a forward- and a reverse-biased junction. The diagram

shows that, at the specified test current (IZT), the absolute

value of voltage change (ΔV) for the temperature change

between 25 and 100°C is the same for both junctions.

Therefore, the total voltage across the combination of these

two junctions is also the same at these temperature points,

since one ΔV is negative and the other is positive. However,

the rate of voltage change with temperature over the

Figure 1. Temperature Compensation

of a 6.2 Volt Reference Diode (1N821 Series)

DIODE VOLTAGE DROP (V)

(REFERENCED TO IZT = 7.5 mA)

6.4

6.2

5.8

5.6

5.4

0.8

0.6

0.4

0.2

−75 −50 −25 0 +25 +50 +75 +100

TEMPERATURE (°C)

6.2 − VOLT REFERENCE DIODE

(COMBINATION OF ZENER

AND FORWARD DICE)

ZENER DIE

FORWARD-BIASED COMPENSATING DIE

temperature range defined by these points is not necessarily

the same for both junctions, thus the temperature

compensation may not be linear over the entire range.

Figure 2 also indicates that the voltage changes of the two

junctions are equal and opposite only at the specified test

current. For any other value of current, the temperature

compensation may not be complete.

http://onsemi.com

Figure 2. Temperature Compensation of

P-N Junctions

DIRECTION OF CURRENT FLOW

PACKAGE

OUTLINE

25°C

100°C

IZT

−ΔV

+ΔV

25°C

VRI

100°C

IZT −

FORWARD-BIASED

P-N JUNCTION

REVERSE-BIASED

ZENER JUNCTION

IMPORTANT ELECTRICAL CHARACTERISTICS

OF REFERENCE DIODES

The three most important characteristics of reference

diodes are 1) reference voltage, 2) voltage-temperature

stability, and 3) voltage-time stability.

1. Reference Voltage. This characteristic is defined as the

voltage drop measured across the diode when the specified

test current passes through it in the zener direction. It is also

called the zener voltage (VZ, Figure 3). On the data sheets,

the reference voltage is given as a nominal voltage for each

family of reference diodes.

The nominal voltages are normally specified to a

tolerance of ±5%, but devices with tighter tolerances, such

as ±2% and ±1%, are available on special order.

2. Voltage-Temperature Stability. The temperature

stability of zener voltage is sometimes expressed by means

of the temperature coefficient. This parameter is usually

defined as the percent voltage change across the device per

degree centigrade. This method of indicating voltage

stability accurately reflects the voltage deviation at the test

temperature extremes but not necessarily at other points

within the specified temperature range. This fact is due to

variations in the rate of voltage change with temperature for

the forward- and reverse-biased dice of the reference diode.

Therefore, the temperature coefficient is given in

ON Semiconductor data sheets only as a quick reference,

for designers who are accustomed to this method of

specification.

A more meaningful way of defining temperature stability

is the “box method.” This method, used by

ON Semiconductor, guarantees that the zener voltage will

not vary by more than a specified amount over a specified

temperature range at the indicated test current, as verified by

tests at several temperatures within this range.

Some devices are accurately compensated over a wide

temperature range (−55°C to 100°C), others over a narrower

range (0 to 75°C). The wide-range devices are, as a rule,

more expensive. Therefore, it would be economically

wasteful for the designer to specify devices with a

temperature range much wider than actually required for the

specific device application.

During actual production of reference diodes, it is difficult

to predict the compensation accuracy. In the interest of

maximum economy, it is common practice to test all devices

coming off the production line, and to divide the production

lot into groups, each with a specified maximum ΔVZ. Each

group, then, is given a different device type number.

Figure 3. Typical Voltage − Current Characteristic of Reference Diodes

IF(mA)

0.3

0.2

0.1

−6 −4 −2

−0.1

−0.2

−0.3

2 4 6 8 10 12 14 16

VF (V)

IR(mA)

VR (V)

http://onsemi.com

On the data sheet, the voltage-temperature characteristics

of the most widely used device types are illustrated in a

graph similar to the one shown in Figure 4. The particular

production line represented in this figure produces 6.2 volt

devices, but the line yields five different device type

numbers (1N821 through 1N829), each with a different

temperature coefficient. The 1N829, for example, has a

maximum voltage change of less than 5 mV over a

temperature range of −55 to +100°C, while the 1N821 may

have a voltage change of up to 96 mV over the same

temperature range.

Figure 4. Temperature Dependence

of Zener Voltage (1N821 Series)

ΔVZ = +31 mV

ΔVZ = −31 mV

IZT = 7.5 mA

1N821,A

1N823,A

1N825,A

1N829,A

1N827,A

1N825,A

1N823,A

1N821,A

100

−25

−50

−75

−100

−55 0 50 100

ΔVZ, MAXIMUM VOLTAGE CHANGE (mV)

(Referenced to −55°C)

In the past, design data and characteristic curves on data sheets fo

reference diodes have been somewhat limited: The devices hav

been characterized principally at the recommended operating point

ON Semiconductor has introduced a data sheet, providing devic

data previously not available, and showing limit curves that permi

worst-case circuit design without the need for associated tests re

quired in conjunction with the conventional data sheets.

Graphs such as these permit the selection of the

lowest-cost device that meets a particular requirement. They

also permit the designer to determine the maximum voltage

change of a particular reference diode for a relatively small

change in temperature. This is done by drawing vertical lines

from the desired temperature points at the abscissa of the

graph to intersect with each the positive- and negative-going

curves of the particular device of interest. Horizontal lines

are then drawn from these intersects to the ordinate of the

graph. The difference between the intersections of these

horizontal lines with the ordinate yields the maximum

voltage change over the temperature increment. For

example, for the 1N821, a change in ambient temperature

from 0 to 50°C results in a voltage change of no more than

about ±31 mV.

The reason that the device reference voltage may change

in either the negative or positive direction is that after

assembly, some of the devices within a lot may be

overcompensated while others may be undercompensated.

In any design, the “worst-case” condition must be

considered. Therefore, in the above example, it can be

assumed that the maximum voltage change will not exceed

31 mV.

It should be understood, however, that the above

calculations give the maximum possible voltage change for

the device type, and by no means the actual voltage change

for the individual unit.

3. Voltage-Time Stability. The voltage-time stability of

a reference diode is defined by the voltage change during

operating time at the standard test current (IZT) and test

temperature (TA). In general, the voltage stability of a

reference diode is better than 100 ppm per 1000 hours of

operation.

Figure 5. Current Dependence of Zener Voltage

at Various Temperatures

(1N821 Series)

IZ, ZENER CURRENT (mA)

7.5

−75 −50 −25 0 25 50

+100°C

IZT

+25°C

−55°C

+25°C

+100°C

−55°C

ΔVZ, MAXIMUM VOLTAGE CHANGE (mV)

(Referenced to IZT = 7.5 mA)

THE EFFECT OF CURRENT VARIATION ON

ZENER VOLTAGE

The nominal zener voltage of a reference diode is

specified at a particular value of current, called the zener test

current (IZT). All measurements of voltage change with

temperature are referenced to this test current. If the

operating current is varied, all these specifications will

change.

The effect of current variation on zener voltage, at various

temperatures, is graphically illustrated on the 1N821 data

sheet as “Zener Current versus Maximum Voltage Change.”

A typical example of such a graph is shown for the 1N821

series in Figure 5. The voltage change shown is due entirely

to the impedance of the device at the fixed temperature. It

does not reflect the change in reference voltage due to the

change in temperature since each curve is referenced to IZT

= 7.5 mA at the indicated temperature. As shown, the

http://onsemi.com

100

greatest voltage change occurs at the highest temperature

represented in the diagram. (See “Dynamic Impedance”

under the next section).

Figure 5 shows that, at 25°C, a change in zener current

from 4 to 10 mA causes a voltage shift of about 90 mV.

Comparing this value with the voltage-change example in

Figure 4 (31 mV), it is apparent that, in general, a greater

voltage variation may be due to current fluctuations than to

temperature change. Therefore, good current regulation of

the source should be a major consideration when using

reference diodes in critical applications.

It is not essential, however, that a reference diode be

operated at the specified test current. The new

voltage-temperature characteristics for a change in current

can be obtained by superimposing the data of Figure 5 on

that of Figure 4. A new set of characteristics, at a test current

of 4 mA, is shown for the 1N823 in Figure 6, together with

the original characteristics at 7.5 mA.

Figure 6. Voltage Change with Temperature

for 1N823 at Two Different Current Levels

+100

+50

−50

−100

−150

−200

−50 0 50 100

7.5 mA

4 mA

TEMPERATURE (°C)

VZ(mV) (REFERENCED TO −55 C)°Δ

From these characteristics, it is evident that the voltage

change with temperature for the new curves is different from

that for the original ones. It is also apparent that if the test

current varies between 7.5 and 4 mA, the voltage changes

would lie along the dashed lines belonging to the given

temperature points. This clearly shows the need for a

well-regulated current source.

It should be noted, however, that even when a

well-regulated current supply is available, other factors

might influence the current flowing through a reference

diode. For example, to minimize the effects of

temperature-sensitive passive elements in the load circuit on

current regulation, it is desirable that the load in parallel with

the reference diode have an impedance much higher than the

dynamic impedance of the reference diode.

OTHER CHARACTERISTICS

In addition to the three major characteristics discussed

earlier, the following parameters and ratings of reference

diodes may be considered in some applications.

Power Dissipation

The maximum dc power dissipation indicates the power

level which, if exceeded, may result in the destruction of the

device. Normally a device will be operated near the

specified test current for which the data-sheet specifications

are applicable. This test current is usually much below the

current level associated with the maximum power

dissipation.

Dynamic Impedance

Zener impedance may be construed as composed of a

current-dependent resistance shunted by a

voltage-dependent capacitance. Figure 7 indicates the

typical variations of dynamic zener impedance (ZZ) with

current and temperature for the 1N821 reference diode

series. These diagrams are given in the 1N821 data sheet. As

shown, the zener impedance decreases with current but

increases with ambient temperature.

Figure 7. Variation of Zener Impedance

With Current and Temperature (1N821 Series)

1000

800

600

400

200

100

11 2 4681020406080100

IZ, ZENER CURRENT (mA)

ZZ, MAXIMUM ZENER IMPEDANCE (OHMS)

−55°C

25°C

100°C

The impedance of a reference diode is normally specified

at the test current (IZT). It is determined by measuring the ac

voltage drop across the device when a 60 Hz ac current with

an rms value equal to 10% of the dc zener current is

superimposed on the zener current (IZT). Figure 8 shows the

block diagram of a circuit used for testing zener impedance.

http://onsemi.com

101

Figure 8. Block Diagram of Test Circuit for Measuring Dynamic Zener Impedance

DC POWER

SUPPLY

ǒ88i8.8

8888 Ǔ

712A

Aǒ88i8.8

8888 Ǔ

650A

SIGNAL

GENERATOR

ǒ88i8.8

8888 Ǔ

400H

VTVM

ǒ88i8.8

8888 Ǔ

412A

VTVM

METER

DUMMY

LOAD

600

0.1μF

1k1k

Rx=Zz

READ

SET

READ

SET (B)

(A) 10M

100pF

Rx= E1−E2

ELECTRICAL TESTING

All devices are tested electrically as a last step in the

manufacturing process.

The subsequent final test procedures represent an

automated and accurate method of electrically classifying

reference diodes. First, an electrical test is performed on all

devices to insure the correct voltage-breakdown and

stability characteristics. Next, the breakdown voltage and

dynamic impedance are measured. Finally, the devices are

placed in an automatic data acquisition system that

automatically cycles them through the complete

temperature range specified. The actual voltage

measurements at the various temperature points are retained

in the system computer memory until completion of the full

temperature excursion. The computer then calculates the

changes in voltage for each device at each test temperature

and classifies all units on test into the proper category. The

system provides a printed readout for every device,

including the voltage changes to five digits during

temperature cycling, and the corresponding EIA type

number, as well as the data referring to test conditions such

as device position, lot number, and date.

DEVICE RELIABILITY AND

QUALITY ASSURANCE

Insuring a very low failure rate requires maximum

performance in all areas effecting device reliability: Device

design, manufacturing processes, quality control, and

reliability testing. ON Semiconductor’s basic reliability

concept is based on the belief that reference diode reliability

is a complex yet controllable function of all these variables.

Under this “total reliability” concept, ON Semiconductor

can mass-produce high-reliability reference diodes.

The reliability of a reference diode fundamentally

depends upon the device design, regardless of the degree of

effort put into device screening and circuit designing.

Therefore, reliability measures must be incorporated at the

device design and process development stages to establish

a firm foundation for a comprehensive reliability program.

The design is then evaluated by thorough reliability testing,

and the results are supplied to the Design Engineering

department. This closed-loop feedback procedure provides

valuable information necessary to improve important design

features such as electrical instability due to surface effects,

mechanical strength, and uniformly low thermal resistance

between the die and ambient environment.

Process Control

There are more than 2000 variables that must be kept

under control to fabricate a reliable reference diode. The

in-process quality control group controls most of these

variables. It places a strict controls on all aspects of

manufacturing from materials procurement to the finished

product. Included in this broad spectrum of controls are:

• Materials Control. All materials purchased or

fabricated in-plant are checked against rigid specifications.

A quality check on vendors’ products is kept up to date to

insure that only materials of a proven quality level will be

purchased.

• In-Process Inspection and Control. Numerous on-line

inspection stations maintain a statistical process control

program on specific manufacturing processes. If any of

these processes are found to be out of control, the discrepant

material is diverted from the normal production flow and the

cognizant design engineer notified. Corrective action is

initiated to remedy the cause of the discrepancy.

Reliability Testing

The Reliability Engineering group evaluates all new

products and gives final conclusions and recommendations

to the device design engineer. The Reliability Engineering

group also performs independent testing of all products and

includes, as part of this testing program,

step-stress-to-failure testing to determine the maximum

capabilities of the product.

http://onsemi.com

102

SOME STRAIGHT TALK ABOUT MOSORBS

TRANSIENT VOLTAGE SUPPRESSORS

INTRODUCTION

Distinction is sometimes made between devices

trademarked Mosorb (by ON Semiconductor Inc.), and

standard zener/avalanche diodes used for reference,

low-level regulation and low-level protection purposes. It

must be emphasized from the beginning that Mosorb

devices are, in fact, zener diodes. The basic semiconductor

technology and processing are identical. The primary

difference is in the applications for which they are designed.

Mosorb devices are intended specifically for transient

protection purposes and are designed, therefore, with a large

effective junction area that provides high pulse power

capability while minimizing the total silicon use. Thus,

Mosorb pulse power ratings begin at 500 watts − well in

excess of low power conventional zener diodes which in

many cases do not even include pulse power ratings among

their specifications.

MOVs, like Mosorbs, do have the pulse power

capabilities for transient suppression. They are metal oxide

varistors (not semiconductors) that exhibit bidirectional

avalanche characteristics, similar to those of back-to-back

connected zeners. The main attributes of such devices are

low manufacturing cost, the ability to absorb high energy

surges (up to 600 joules) and symmetrical bidirectional

“breakdown” characteristics. Major disadvantages are: high

clamping factor, an internal wear-out mechanism and an

absence of low-end voltage capability. These limitations

restrict the use of MOVs primarily to the protection of

insensitive electronic components against high energy

transients in application above 20 volts, whereas, Mosorbs

are best suited for precise protection of sensitive equipment

even in the low voltage range − the same range covered by

conventional zener diodes. The relative features of the two

device types are covered in Table 1.

IMPORTANT SPECIFICATIONS FOR

MOSORB PROTECTIVE DEVICES

Typically, a Mosorb suppressor is used in parallel with the

components or circuits being protected (Figure 1), in order

to shunt the destructive energy spike, or surge, around the

more sensitive components. It does this by avalanching at its

“breakdown” level, ideally representing an infinite

impedance at voltages below its rated breakdown voltage,

and essentially zero impedance at voltages above this level.

In the more practical case, there are three voltage

specifications of significance, as shown in Figure 1a.

a) VRWM is the maximum reverse stand-off voltage at

which the Mosorb is cut off and its impedance is at its

highest value − that is, the current through the device is

essentially the leakage current of a back-biased diode.

b) V(BR) is the breakdown voltage − a voltage at which the

device is entering the avalanche region, as indicated by

a slight (specified) rise in current beyond the leakage

current.

c) VRSM is the maximum reverse voltage (clamping

voltage) which is defined and specified in conjunction

with the maximum reverse surge current so as not to

exceed the maximum peak power rating at a pulse

width (tp) of 1 ms (industry std time for measuring

surge capability).

RELATIVE FEATURES OF MOVs and MOSORBS

Table 1.

MOV Mosorb/Zener Transient Suppressor

•High clamping factor.

•Symmetrically bidirectional.

•Energy capability per dollar usually higher than a silicon device.

However, if good clamping is required the energy capability would

have to be grossly overspecified resulting in higher cost.

•Inherent wear out mechanism clamp voltage degrades after every

pulse, even when pulsed below rated value.

•Ideally suited for crude ac line protection.

•High single-pulse current capability.

•Degrades with overstress.

•Good high voltage capability.

•Limited low voltage capability.

•Very good clamping close to the operating voltage.

•Standard parts perform like standard zeners. Symmetrical bidirec-

tional devices available for many voltages.

•Good clamping characteristic could reduce overall system cost.

•No inherent wear out mechanism.

•Ideally suited for precise DC protection.

•Medium multiple-pulse current capability.

•Fails short with overstress.

•Limited high voltage capability unless series devices are used.

•Good low voltage capability.

http://onsemi.com

103

In practice, the Mosorb is selected so that its VRWM is

equal to or somewhat higher than the highest operating

voltage required by the load (the circuits or components to

be protected). Under normal conditions, the Mosorb is

inoperative and dissipates very little power. When a

transient occurs, the Mosorb converts to a very low dynamic

impedance and the voltage across the Mosorb becomes the

clamping voltage at some level above V(BR). The actual

clamping level will depend on the surge current through the

Mosorb. The maximum reverse surge current (IRSM) is

specified on the Mosorb data sheets at 1 ms and for a

logrithmically decaying pulse waveform. The data sheet

also contains curves to determine the maximum surge

current rating at other time intervals.

Typically, Mosorb devices have a built-in safety margin at

the maximum rated surge current because the clamp voltage,

VRSM, is itself, guardbanded. Thus, the parts will be

operating below their maximum pulse-power (Ppk) rating

even when operated at maximum reverse surge current).

If the transients are random in nature (and in many cases

they are), determining the surge-current level can be a

problem. The circuit designer must make a reasonable

estimate of the proper device to be used, based on his

knowledge of the system and the possible transients to be

encountered. (e.g., transient voltage, source impedance and

time, or transient energy and time are some characteristics

that must be estimated). Because of the very low dynamic

impedance of Mosorb devices in the region between V(BR)

and VRSM, the maximum surge current is dependent on, and

limited by the external circuitry.

In cases where this surge current is relatively low, a

conventional zener diode could be used in place of a Mosorb

or other dedicated protective device with some possible

savings in cost. The surge capabilities most of

ON Semiconductor’s zener diode lines are discussed in

ON Semiconductor’s Application Note AN784.

In the data sheets of some protective devices, the

parameter for response time is emphasized. Response time

on these data sheets is defined as the time required for the

voltage across the protective device to rise from 0 to V(BR),

and relates primarily to the effective series impedance

associated with the device. This effective impedance is

somewhat complex and changes drastically from the

blocking mode to the avalanche mode. In most applications

(where the protective device shunts the load) this response

time parameter becomes virtually meaningless as indicated

by the waveforms in Figures 1b and 1c. If the response time

as defined is very long, it still would not affect the

performance of the surge suppressor.

However, if the series inductance becomes appreciable, it

could result in “overshoot” as shown in Figure 1d that would

be detrimental to circuit protection. In Mosorb devices,

series inductance is negligible compared to the inductive

effects of the external circuitry (primarily lead lengths).

Hence, Mosorbs contribute little or nothing to overshoot

and, in essence, the parameter of response time has very little

significance. However, care must be exercised in the design

of the external circuitry to minimize overshoot.

SUMMARY

In selecting a protective device, it is important to know as

much as possible about the transient conditions to be

encountered. The most important device parameters are

reverse working voltage (VRWM), surge current (IRSM), and

clamp voltage (VRSM). the product of VRSM and IRSM yields

the peak power dissipation, which is one of the prime

categories for device selection.

The selector guide, in this book, gives a broad overview

of the Mosorb transient suppressors now available from ON

Semiconductor. For more detailed information, please

contact your ON Semiconductor sales representative or

distributor.

http://onsemi.com

104

Figure 1a. Figure 1b.

Figure 1.

Figure 1c. Figure 1d.

MOSORB

PROTECTED

LOAD

OVERSHOOT

VOLTAGE

VRSM

TIME

V(BR)

VRWM

Vin

Vout

tclamping, VERY SHORT

VOLTAGE

TIME

VRSM

V(BR)

VRWM

Vout

Vin

VOLTAGE

TIME

VOLTAGE

TIME

VRSM

V(BR)

VRWM

VRSM

V(BR)

VRWM

Vout

Vin

tclamping, VERY LONG tclamping, WITH OVERSHOOT

Vout

Vin

tp = PULSE WIDTH OF INCOMING TRANSIENT

Vin Vout

http://onsemi.com

105

TYPICAL MOSORB APPLICATIONS

−

MOSORB

MOSORBS

INPUT

POWER

SUPPLY

DC Power Supplies Input/Output Regulator Protection

MOSORB MOSORB

VOLTAGE

REGULATOR

MOSORB

OP AMP

−B

Op Amp Protection

−

MOSORB

Inductive Switching Transistor Protection

MOTOR

DC Motors − Reduces EMI Memory Protection

Microprocessor Protection

Computer Interface Protection

MOS

MEMORY

+5 V

MOSORBS

I/O

ADDRESS BUS

RAM ROM

DATA BUS

CONTROL BUS

MOSORB MOSORB

VDD

VGG

Gnd

CPU

CLOCK

I/O

KEYBOARD

TERMINAL

PRINTER

ETC.

Gnd

FUNCTIONAL

DECODER

MOSORB

−8 V

MOSORB

−10 V

MOSORBS

Vout

http://onsemi.com

106

AR450 − CHARACTERIZING OVERVOLTAGE TRANSIENT SUPPRESSORS

Prepared by

Al Pshaenich

ON Semiconductor Power Products Division

The use of overvoltage transient suppressors for

protecting electronic equipment is prudent and

economically justified. For relatively low cost, expensive

circuits can be safely protected by one or even several of the

transient suppressors on the market today. Dictated by the

type and energy of the transient, these suppressors can take

on several forms.

For example, in the telecommunication field, where

lightning induced transients are a problem, such primary

suppressors as gas tubes are often used followed by

secondary, lower energy suppressors. In an industrial or

automotive environment, where transients are

systematically generated by inductive switching, the

transient energy is more well-defined and thus adequately

suppressed by relatively low energy suppressors. These

lower energy suppressors can be zener diodes, rectifiers with

defined reverse voltage ratings, metal oxide varistors

(MOVs), thyristors, and trigger devices, among others. Each

device has its own niche: some offer better clamping factors

than others, some have tighter voltage tolerances, some are

higher voltage devices, others can sustain more energy and

still others, like the thyristor family, have low on-voltages.

The designer’s problem is selecting the best device for the

application.

Thus, the intent of this article is twofold:

1. To describe the operation of the surge current test

circuits used in characterizing lower energy transient

suppressors.

2. To define the attributes of the various suppressors,

allowing the circuit designer to make the

cost/performance tradeoffs.

Surge suppressors are generally specified with

exponentially decaying and/or rectangular current pulses.

The exponential surge more nearly simulates actual surge

current conditions − capacitor discharges, line and switching

transients, lightning induced transients, etc., whereas

rectangular surge currents are usually easier to implement

and control.

To generate an exponential rating, a charged capacitor is

simply dumped into the device under test (DUT) and the

energy of each successive pulse increased until the device

ultimately fails. The simplified circuit of Figure 1a describes

the circuit. By varying the size of the capacitor C, the

limiting resistor R2, and the voltage to which C is charged

to, various peak currents and pulse widths (defined to the

10% discharge point in this paper) can be obtained. To

Figure 1a. Simplified Exponential Tester

Figure 1b. Simplified Rectangular Tester

Using PNP Switch

Figure 1. Basic Surge Current Testers

VIN

R1 R2

DUT 10%

DUT

VEE

automate this circuit, the series switches S1 and S2 can be

replaced with appropriately controlled transistors or SCRs.

One method of easily implementing a rectangular surge

current pulse is shown in the simplified schematic of

Figure 1b. A PNP transistor switch connected to the

positive supply VEE applies power to the DUT. The current

is obviously set by varying either VEE and/or RL. If however,

the transistor switch were replaced with a variable, constant

current source, measurement procedures are simplified as

how the limiting resistor need not be selected for various

current conditions.

As in most surge current evaluations, the DUT is

ultimately subjected to destructive energy (current, voltage,

pulse width), the failure points noted, and the derated points

plotted to produce the energy limitation curve. Of particular

interest is the junction temperature at which the DUTs are

operated, be it near failure or at the specified derated point.

This measurement relates to the overall reliability of the

suppressor, i.e., can the suppressor sustain one surge current

pulse or a thousand, and will it be degraded when operated

above the specified maximum operating temperature?

The Rectangular Current Surge Suppressor Test Circuit to

be described addresses these questions by implementing and

measuring the rectangular current capability of the

http://onsemi.com

107

suppressor and determining the device junction temperature

TJ shortly after the end of the surge current pulse. Knowing

TJ, the energy to the DUT can be limited just short of failure

and thus a complete surge curve generated with only one, or

a few DUTs (Figure 6). Second, with the junction

temperature known, a reliability factor can be determined

for a practical application.

CIRCUIT OPERATION FOR THE

RECTANGULAR CURRENT TESTER

The Surge Suppressor Test Circuit block diagram is

shown in Figure 2 with the main blocks being the Constant

Current Amplifier supplying IZ to the DUT (a zener diode

in this instance) during the power pulse and the Diode

Forward Current Switch supplying IF during the

temperature sensing time. These two pulses are applied

sequentially, first the much larger IZ, and then the very small

sense current IF. During the IF time, the forward voltage VF

of the diode is measured from which the junction

temperature of the zener diode can be determined. This is

simply done by calibrating the forward biased DUT with a

specified low value of IF in a temperature chamber, one point

at 25°C and a second point at some elevated temperature.

The result is the familiar diode forward voltage versus

temperature linear plot with a slope of about −2 mV/°C for

typical diodes (Figure 7a). Comparing the plot with the test

circuit measured VF yields the DUT junction temperature

for that particular pulse width and IZ (Figure 7b).

Figure 2. Surge Suppressor Test

Circuit Block Diagram

25 μS

BLANKING

MV GATE

SAMPLE

PULSE

300 μS

SENSE

SHORT

DETECTOR

OPEN

DETECTOR

S/H

DUT

PULSE

GEN

CLOCK

The System Clock, Pulse Generator, the several

monostable multivibrators (25 μs Blanking, Sample Pulse

and 300 μs Sense MVs) and Gate are fashioned from three

CMOS gate ICs. The remaining blocks are the Sample and

Hold (S/H) circuit and two detectors for determining the

status of failed DUTs, either shorted devices or open.

Shown in Figures 3 and 4 respectively, are the complete

circuit and significant waveforms. Clocking for the system

is derived from a CMOS, two inverters, astable MV (gates

1A and 1B) whose output triggers the two input NOR gate

configured monostable MV (gates 1C and 1D) to produce

the Pulse Generator output pulse (Figure 4b). Alternatively,

a single pulse can be obtained by setting switch S2 to the One

Shot position and depressing the pushbutton Start switch S1.

Contact bounce is suppressed by the 100 ms MV (gates 4C

and D). Frequency of the astable MV, set by potentiometer

R1, can vary from about 200 Hz to 0.9 Hz and the pulse

width, controlled by R2 and the capacitor timing selector

switch S3, from about 300 μs to 1.3 s.

The positive going Pulse Generator output feeds the

Constant Current Amplifier IZ and turns on, in order, NPN

transistor Q1, PNP transistor Q3, NPN Darlington Q4, PNP

Power Darlington Q6 and parallel connected PNP Power

Transistors Q8 and Q9. Transistor Q4 is configured as a

constant current source whose current is set by the variable

base voltage potentiometer R3. Thus, the voltage to the

bases of Q6, Q8 and Q9 are also accordingly varied.

Transistors Q8 and Q9 (MJ14003, IC continuous of 60 A),

also connected as constant current sources with their 0.1 Ω

emitter ballasting resistors, consequently can produce a

rectangular current pulse from a minimum of about 0.5 A

and still have adequate gain for 1 ms pulses of 150A peak.

Due to propagation delays of this amplifier, the IZ current

waveform is as shown in Figure 4f. Since Q8 and Q9 must

be in the linear region for constant current operation, these

transistors are power dissipation limited at high currents to

the externally connected power supply V+ of 60 V. Thus the

maximum DUT voltage, taking into account the clamping

factor of the device, should be limited to about 50 V. At

wider pulse widths and consequently lower currents before

the DUT fails, the V+ supply should be proportionally

reduced to minimize Q8, Q9 dissipation. As an example, a

28 V surge suppressor operating at 100 ms pulse widths can

be tested to destructive limits with V+ of about 40 V.

Although a zener diode is shown as the DUT in the

schematic, the test devices can be any rectifier with defined

reverse voltage, e.g., surge suppressors.

Immediately after the power pulse is applied to the DUT,

the negative going sense pulse from the 300 μs MV (Gate

2A, Figure 4e) turns on series connected PNP transistor Q10

and NPN transistor Q11 of the Diode Forward Current

Switch IF. Sense current, set by current limiting resistor

selector switch S4, thus flows up from ground through the

forward biased DUT, the limiting resistor, and Q11 to the

−15 V supply. The result, by monitoring the cathode of the

DUT, is a 300 μs wide, approximately −0.6 V pulse.

http://onsemi.com

108

P.W. M.V.

SW S4

5 mA

RED

13 11

S1 START SW

+15 V 0.001 μF0.1 μF

47 k 100 k

100 mS MV

CONTACT BOUNCE

1.8M

+15 V

1A 1

U4C,D

(1/2) MC14001

47 k 100 k

1 SHOT SWS2

22 k

1B 3

270 pF

FREE

RUN

FREQ CONT

0.01 μF

0.1

CLOCK U1

MC14572

5 ms < T1 < 1.2S

47 k

1 k

SYNC

0.0033 μF

0.4 μF

39 k

PW CONT

+15 V

I2 SWITCH

3.9 k

5.6 k

22 k

10 k

47 k

270 pF

100 k

8 μs

SAMPLE

PULSE

270 pF

1.2M

390 pF

TRIAD

F90X

1D 91F 11

910

13 14

910

−

+15 V

60 μs < T2N < 8.5 mS

15 1E

+15 V

47 k

330 pF

100 k

MC14001

25 μs M.V.

300 μs

SENSE M.V. 200 μF

25 V

+−

120 V

7 ms < T2W < 1.1 S

MDA101A

0.1 μF

PWR ON

LED

7815

100 k 10 k

Q18

2N3904

390 k

100 k

270 pF

MC14001

μC

7915

RED

1.2 k

1/2 W

200 μF

25 V

(2)

10 k

2 W

0.1 Ω

25 W

(2)

+15 V

1N914

220 pF

+15 V

390 pF

200 μF

25 V

DUT

1N914

+15 V

−15 V

0.1 μF

+15 V

100 μS SAMPLE DELAY M.V.

P.W.

4,7 k

1 W

25 μs

+15 V

Q14

MPS8099

2N3906

Q15

10 k

1 k

SHORT LED #1

RED

OPEN LED

100

470

1 W

470

1 W

SHORT

DETECTOR #1 Q16

2N5060

0.1 μF

43 k

10 k

4.7 k

−15 V

+15 V

0.1 μF

4.7 k

−15 V

1 k

0.1 μF

1 k

0.1 μF

4.7 k

10 k

0.1 μF

−15 V

Q13

2N4858

25 k

100 k

(1/2) MC14001

U4A,B

10 k 100 k

150 k

+15 V

47 k

1 k

5 pF

+15 V

0.1 μF

1N914

(2)

MR821

1.3 k

4.7 k

−15 V

IF SWITCH 0.1 μF

10 k

180

300

1/2 W

820

1.2 k

10 k

390

1/2 W

1 k

3.3 k

12 k

10 mΩ

100 k

47 k 1 k

1N914

10 k

1 k

1/2 W

10 k

1N914

0.1 μF

GATE

300 μS

100 μS

8 μS

SENSE

DELAY

SAMPLE

DUT

−0.7 +3 V

+15 V

2N5060

Q17

OPEN DETECTOR

RESET SW S5

0.1 μF

0.001 μF

U5A

(1/2) MC1458

0.1 μF

+15 V

−

157

S/H

U7 LF355

−15 V

−

U6 LF355

CI R4

+15 V

1 k 0.1 μF

0.001 μF

0.001

Q12

2N3906

SHORT

LED #2

47 k 22 k

10 k

Q8 Q9

2N6668

V + ≤ 60 V 10.000 μF

75 V

Q5 8599

MPS

15 k

1/2 W

+15 V

0.1 μF

680

1/2 W

1 k

2.7 k

10 mA

560

25 mA

2N6287

MPSA29

2N3906

2N3904

2N3906

+15 V

Q10

2N3906

Q11

MPS8099

SHORT

DET #2

+15 V

U5B

(1/2) MC1458

ure 3. Sur

e Su

ressors Sur

e Current Fixture

http://onsemi.com

109

For accurate measurements of this pulse amplitude,

sample and hold circuitry is employed. This consists of unity

gain buffer amp U6, series FET switch Q13 and capacitor

hold buffer amp U7. The sample pulse (Figure 4H) to the

gate of the FET is delayed about 100 μs (by monostable MV

G-2C and G-2D) to allow for switching and thermal

transients to settle down. This pulse is derived from the

negative going, trailing edge output pulse of Gate 2D cutting

off transistor Q18 for the RC time constant in its base circuit.

The result is an approximate 8 μs wide sample pulse.

Consequently, the DC output voltage from hold amplifier

U7 is a measure of the DUT junction temperature.

Invariably, most DUTs will fail short. When the surge

suppressor tester is in the Free-run Mode, the power pulse

subsequent to the DUT shorting could excessively stress the

constant current drivers Q8 and Q9. To prevent this

occurrence, the Short Detector circuit was implemented.

This circuit consists of comparator U5A, 2 input NOR gate

configured 25 μs monostable MV (G1E and G1F), Gate

Circuit G3A, 3B and 3C, and SCR Q16. The 25 μs MV

(Figure 4D) is required to blank out turn-on switching

transients to produce the waveform shown in Figure 4I.

During the power pulse, U5A is normally high for a good

DUT (Figure 4J). This waveform is NOR’d with gate 3B

(inverted waveform of Figure 4I) to produce a low level

(0 V) gate 3C output (Figure 4K).

If, however, the DUT is shorted, U5A output switches low

resulting in a positive pulse output from G3C. This pulse

triggers the SCR on, lighting the LED in its anode circuit and

turning on the PNP transistor Q2 across the emitter-base of

Q3, thus clamping off the IZ power pulse. The circuit (Q16)

can be reset by opening switch S5.

By and large, this Short Detector circuit was found

adequate to protect transistors Q8 and Q9. However, for

some wide pulse widths, relatively high current conditions,

the propagation delay through the Short Detector was too

great, resulting in excessive FBSOA (Forward Bias Safe

Operating Area) stress on Q8 and Q9. Consequently, a faster

Short Detector #2 was implemented.

Figure 4. Surge Suppressor Test Circuit Waveforms

+15 V

−14 V

CLOCK

G1B

PULSE GEN

G1D

PULSE GEN

G1C

25 μS MV

G1E

300 μS

SENSE MV

G2A

100 μS SAMPLE

DELAY MV G2D

8 μS

SAMPLE PULSE Q8

GATE

G3C

SHORT

COMPARATOR U6A

OPEN SCR

TRIGGER U5B

SHORT SCR

TRIGGER G3C

−0.6 V

−14 V

GOOD

SHORT

GOOD

SHORT

GOOD

OPEN

(4A)

(4B)

(4C)

(4D)

(4E)

(4F)

(4G)

(4H)

(4I)

(4J)

(4K)

(4L)

(4M)

http://onsemi.com

110

This circuit, connected to the collectors of Q8 and Q9, uses

a differentiating network (R4, C1) to discriminate between

the normally relatively slow fall time of the voltage pulse on

the DUT, and the exceedingly fast fall time when the device

fails. Thus, the R4-C1 time constant (5 ns) will only generate

a negative going trigger to PNP transistor Q12 when the DUT

voltage collapses during device failure. The positive going

output from Q12 resets the flip-flop (gates 4A and 4B), which

turns on the NPN transistor Q14. This transistor supplies

drive to the two PNP clamp transistors (Q5 and Q7) placed

respectively across the emitter-bases of the high, constant

current stages Q6 and Q8 and Q9. Propagation delay is thus

minimized, providing greater protection to the power stages

of the tester. As an added safety feature, the positive going

output from Gate 3C when Short Detector #1 is activated is

also used to trigger the flip-flop.

On the few occasions when the DUT fails open, then the

Open Detector consisting of comparator U5B and SCR Q17

comes into play. This circuit measures the DUT integrity

during the sense time. For a good DUT (VF < −1 V), U5B

output remains low (see Figures 4L and 4M). However for

an open DUT, VF switches to the negative rail and U5B goes

high, turning on Q17. As in the Short Detector, Q2 clamps

off the IZ power amplifier.

All of the circuitry including the +15 V and −15 V

regulated power supplies are self-contained, with the

exception of the V+ supply. For high current, narrow pulse

width testing, this external supply should have 10 to 15 A

capability. If not, additional energy storing capacitors across

the supply output may be required.

CIRCUIT OPERATION FOR THE

EXPONENTIAL SURGE CURRENT TESTER

To generate the surge current curve of peak current versus

exponential discharge pulse width, the test circuit of

Figure 5 was designed. This tester is an implementation of

the simplified capacitor discharge circuit shown in

Figure 1A, with the PNP high voltage transistor Q2

allowing the capacitor C to charge through limiting resistor

R1 and a triggered SCR discharging the capacitor. As shown

in Figure 5, the DUTs can be of any technology, although the

device connected to the capacitor and discharge limiting

resistor RS is shown as an MOS SCR. It could just as well

have been an SCR as the DUT or as the switch for the zener

diode, rectifier, SIDAC, etc., DUTs.

System timing for this Exponential Surge Current Tester

is derived from a CMOS quad 2 input NOR gate with gates

1A and 1B comprising a non-symmetrical astable MV of

about 13 seconds on and about one second off (switch S3

open). The positive On pulse from gate 1B turns on the 500

V power MOSFET Q1 and the following PNP transistor Q2.

The extremely high current gain FET allows for the large

base current variation of Q2 with varying supply voltage

(V+). This capacitor charging circuit has a 400 V blocking

capability (limited by the VCEO of Q2) and thus the capacitor

C1 used should be comparably rated. When operating with

high voltage (V+ = 200 to 350 V) and large capacitors

(>3000 μF), the power dissipated in the current limiting

resistor R1 can be substantial, thus necessitating the

illustrated 20 W rating. For longer charging times, switch S3

is closed, doubling the timing capacitor and the astable MV

on time.

To discharge capacitor C1 and thereby generate the

exponential surge current, the SCR must be fired. This

trigger is generated by the positive going one second pulse

from gate 1A being integrated by the R2C2 network, and

then shaped by gates 1C and 1D. The net result of about 100

μs time delay from gate 1D ensures noncoincident timing

conditions. This pulse output is then differentiated by C3-R3

with the positive going leading edge turning on Q3, Q4 and

finally the SCR with about a 4 ms wide, 15 mA gate pulse.

Consequently, the DUT is subjected to a surge current pulse

whose magnitude is dictated by the voltage on the capacitor

C1 and value of resistor RS, and also whose pulse width to

the 10% point is 2.3 RSC1. For a fixed pulse width, the DUT

is then stressed with increasing charge (by increasing V+)

until failure occurs, usually a shorted device.

If the DUT is the SCR (or MOS SCR), the failed condition

is obvious as the capacitor C1 will not be allowed to charge

for subsequent timing cycles. However, when the DUT is the

zener, rectifier, SIDAC or even an MOV, and the SCR is an

adequately rated switch, the circuit will still discharge

through the shorted DUT, but now the SCR alone will be

stressed by the surge current. A shorted DUT can be detected

by noting the voltage across the device during testing.

One problem encountered when stressing SCRs with high

voltage is when the DUT fails short. The limiting resistor

R1, which is only rated for 20 W, would now experience

continuous power dissipation for the full On time − as much

as 123 W ([350 V]2/1K). To prevent this occurrence, the PR1

Short Protection Circuit is incorporated. Since this is only a

problem when high V+’s (>100 V) are used, the circuit can

be switched in or out by means of switch S2. When

activated, this circuit monitors the voltage on capacitor C1

some time after the charging cycle begins. If the capacitor is

charging, normal operation occurs. However, if the SCR

DUT is shorted, the absence of voltage on the capacitor is

detected and the system is disabled.

The circuit consists of one CMOS IC with NAND gates

2A and 2B comprising a one second monostable time delay

MV and gates 2C and 2D forming a comparator and NAND

gate, respectively.

The negative going, trailing edge of gate 2A is

differentiated by R4-C4, and amplified by Q5 to form a

positive, 10 ms wide pulse (delayed by 1 sec) to gate 2D

input. If the capacitor C1 is shorted, gate 2C output is high,

allowing the now negative pulse from gate 2D to turn on

PNP transistor Q6 and SCR Q7. This latches the input to the

astable MV gate 1A low, disabling the timing and

consequently removing the power from R1. Resetting the

tester for a new device is accomplished by depressing the

pushbutton switch S1.

http://onsemi.com

111

Figure 5. Exponential Surge Current Tester

13 2D

MJE5852

GRN

LED

2N3906

2N3904

CAP

DISCHARGE

SIDAC DUT

RECTIFIER DUT

13 1B

+15 V

1N914

20 W

47K

1N914

47K

62B

21C

10K

+15 V

1.5K

1/2 W

15M

10K

ZENER DUT

1N914

22K

10K

0.1 μF

10K

15K

20 W

+15 V

0.001 μF

91A

+15 V

MC14011

100 k

+15 V

1.8M

18M

1N914

22M 0.47 μF

2 W

+15 V

22K

92C

+15 V

2N3906 2N5060

SW S1

1/2 W

+15 V

RED

LED

RESET

0.1 μF

6.8K

2N3906

22K

+15 V

0.01 μF

22A

1 SEC DELAY MV

MC14011

0.1 μF

+15 V

0.1 μF

100K +15 V

10K

1N4005

SCR

DUT/SW

MOS SCR

DUT

DUT SHORT

INDICATOR

V+ ≤350 V

MTP2N50

61D

47K

22K

+15 V

150K

2 W

0.47 μF

13S

25S

2N3904

ON TIME

Exponential surge current curves, as well as rectangular,

are generated by destructive testing of at least several DUTs

at various pulse widths and derating the final curve by

perhaps 20−30%. These tests were conducted at low duty

cycles (<2%). To ensure multicycle operation, the DUTs are

then tested for about 1000 surges at a derated point on the

curve.

TEST RESULTS

In trying to make a comparison of the several different

technologies of transient suppressors, some common

denominator has to be chosen, otherwise, the amount of

testing and data reduction becomes unwieldy. For this

exercise, voltage was used, generally in the 20 V to 30 V

range, although some of the more unique suppressors

(SIDACs, MOS SCRs, SCRs) were tested at their operating

voltage. As an example, the SIDAC trigger families of

devices were tested with voltages greater than their

breakover voltage (104 V to 280 V) and the SCRs were

subjected to exponential surge currents derived from

voltages generally greater than 30 V. Also, since energy

capability is related to die size, this parameter is also listed.

For several devices, both rectangular and exponential surge

current pulses are listed. Other devices were tested with only

rectangular pulses (where the junction temperature can be

determined) and still others, whose applications include

crowbars, with exponential current only.

AVALANCHE RECTIFIER

The Rectangular Surge Current Tester was originally

designed for characterizing rectifier surge suppressors used

in automotive applications. For this operation, where

temperatures under the hood can reach well over 125°C, it

is important to know the device junction temperature at

elevated ambient temperature. Figures 6 and 7 describe the

results of such testing on a typical suppressor, the 24 V−32 V

MR2520L. It should be noted that these axial lead

suppressors, as well as all other axial lead devices tested,

were mounted between two spring loaded clips spaced 1

inch apart.

http://onsemi.com

112

As shown in Figure 6 of the actual current failure points

of the DUTs, at least four devices were tested at the various

pulse widths, tw (in this example from 0.5 ms to 100 ms).

Also shown in Figure 6 is the curve derived with a single

DUT at an energy level just short of failure. This

measurement was obtained by maintaining a constant

rectifier forward voltage drop, VF (0.25 V) for all pulse

widths (junction temperature, TJ of 230°C) by varying the

avalanche current. Thus, one device can be used,

non-destructively, to generate the complete rectangular

surge current curve.

It should also be pointed out that the definition for the

exponential tw in this article is the current discharge point to

the 10% value of the peak test current IZM. Expressed in time

constant τ, this would be 2.3 RC. Some data sheets describe

tw to the 50% point of IZM (0.69 τ) and others to 5 τ. To

normalized these time scales (abscissa of curves) simply

change the scales accordingly; i.e., IZM/2 pulse widths would

be multiplied by 2.3/0.69 = 3.33 for tw at 10% current pulses.

Figure 7a describes the actual temperature calibration

curve (measured in a temperature chamber) of the

MR2520L and Figure 7b, the junction temperature of the

DUT at various 10 ms rectangular pulse current amplitudes.

These temperatures are taken from the calibration curve (in

actuality, an extremely linear curve), knowing the rectifier

forward voltage drop immediately (within 100 μs) after

cessation of power. Note that the junction temperature just

prior to device failure is about 290°C.

Figure 6. Experimental Rectangular Surge

Current Capability Of The MR2520L Rectifier

Surge Suppressor

I , PEAK SURGE CURRENT (AMPS)

ACTUAL DUTS FAILURE POINTS

ONE DUT WITH VF = 0.25 V

MR2520L RECTIFIER

SURGE SUPPRESSOR,

RECTANGULAR PULSE

VZ = 28 V TYP

TA = 25°C

100

200

10 20 50 100

0.5 1 52

tW, RECTANGULAR PULSE WIDTH (ms)

TJ = 230°C

ZENER OVERVOLTAGE TRANSIENT

SUPPRESSOR

Illustrated in Figure 8 are the actual rectangular and

exponential surge current curves of the P6KE30 overvoltage

transient suppressor, an axial lead, Case 17, 30 V zener

diode characterized and specified for surge currents. This

device is specified for 600 W peak for a 1 ms exponential

pulse measured at IZM/2. From the exponential curve, it is

Figure 7a. Temperature Calibration Curve

Of The MR2520L

V , FORWARD VOLTAGE (VOLTS)

MR2520L AVALANCHE RECTIFIER

SURGE SUPPRESSOR

IF = 10 mA

0.2

0.6

0.8

0.4

50 100

0150 200 250 300 350

TJ, JUNCTION TEMPERATURE (°C)

RECTANGULAR

PULSE

tW = 10 ms

IF = 10 mA

(A)

(V)

(°C)

0.64

0.57

0.48

0.36

0.25

0.15

225

120

180

230

290

55 0.10 DUT

FAILED

Figure 7b. Measured

Forward Voltage

Figure 7. Calculated Junction Temperature

Of The MR2520L Surge Suppressor

At Various Avalanche Currents

apparent that the device is very conservatively specified.

Also, the relative magnitudes of the two curves reflect the

differences in the rms values of the two respective pulses.

SIDAC

SIDACs are increasingly being used as overvoltage

transient suppressors, particularly in telephone applications.

Being a high voltage bilateral trigger device with relatively

high current capabilities, they serve as a costeffective

overvoltage protection device. As in other trigger devices,

when the SIDACs breakover voltage is exceeded, the device

switches to a low voltage conduction state, allowing an

inordinate amount of surge current to be passed. This is well

illustrated by the surge current curves of Figure 9 which

describe the small die size ([37]2mil) axial lead, Case 59-04,

MKP9V240 SIDAC. The curves show that this 240 V device

was able to handle, to failure, as much as 31 A and 15 A,

respectively, for 1 ms exponential and rectangular current

pulses. Under the same pulse conditions, the large die

([78]2mil) MK1V270 SIDAC handled 170 A and 60 A,

respectively, as shown in Table 2.

http://onsemi.com

113

Figure 8. Surge Current Capability Of The P6KE30

Overvoltage Transient Suppressor As A Function Of

Exponential & Rectangular Pulse Widths

I , PEAK SURGE CURRENT (AMPS)

RECTANGULAR

EXPONENTIAL

tW, PULSE WIDTH (ms)

100

10 50 100

0.5 1 50.1 500 1000

P6KE30 OVERVOLTAGE

TRANSIENT SUPPRESSOR

VZ = 30 V

PSPEC = 600 W pk

Figure 9. Measured Surge Current To Failure

Of A SIDAC MKP9V240

I , PEAK SURGE CURRENT (AMPS)

100

10 20 50 100

0.2 0.5 1 52

SIDAC MKP9V240

240 V

CASE 59-04

372 MILS

EXPONENTIAL

RECTANGULAR

tW, PULSE WIDTH (ms)

OVERALL RATINGS

The compilation of all of the testing to date on the various

transient suppressors is shown in Tables 1 and 2. Table 1

describes the zener suppressors, avalanche rectifiers and

MOVs, comparing the die size and normalized costs

(referenced to the MOV V39MA2A). From this data, the

designer can make a cost/performance judgment.

Of interest is that the small pellet MOV is not the least

expensive device. The P6KE30 overvoltage transient

suppressor costs about 85% of the MOV, yet it can handle

about three times the current (2.5 A to 0.7 A) for a 100 ms

rectangular pulse. Under these conditions, the resultant

clamping voltages for the zener and MOV were 32 V and

60 V respectively.

Also shown in the table is a 1.5 W zener diode specified

for zener applications. This low surge current device costs

three times the MOV, illustrating that tight tolerance zener

diodes are not cost effective and that the user should use

devices designed and priced specifically for the suppressor

application.

Thyristor type surge suppressors are shown in Table 2.

They include four SIDAC series, two SCRs designed and

characterized specifically for crowbar applications and also

the MOS SCR MCR1000. The MOS SCR, a process

variation of the vertical structure power MOSFET,

combines the input characteristics of the FET with the

latching action of an SCR.

All devices were surge current tested with the resultant

peak currents being impressively high. The TO-220 (150)2

mil SCR MCR69 for example, reached peak current levels

approaching 700 A for a 1 ms exponential pulse. The

guaranteed, derated, time base translated curves for the

crowbar SCR family of devices are shown in Figure 10, as

is the MK1V SIDAC in Figure 11.

Figures 12A−C describe the guaranteed, reverse surge

design limits for the avalanche rectifier devices. These three

figures illustrate, respectively, the peak current, power and

energy capabilities of these overvoltage transient

suppressors derived from exponential testing. The peak

power, Ppk, ordinate of the curve is simply the product of the

derated IZ and VZ and the energy curve, the product of Ppk

and tw.

http://onsemi.com

114

Figure 10. SCR Crowbar Derating Curves

Figure 11. Exponential Surge Current

Capability Of The MK1V SIDAC, Pulse Width

versus Peak Current

Figure 12b. Peak Power

Ipk

5 TC

I , PEAK CURRENT (AMPS)

C = 8400 μFT

A = 25°C

ESR ≈ 25 mΩN = 2000 PULSES

VC ≤ 60 V f = 3 PULSES/MIN.

a. Peak Surge Current versus Pulse Width

MCR70

MCR69 MCR68

MCR67

MCR71

100

3000

1000

300

0.5 1 50.1 5010 100

tW, BASE PULSE WIDTH (ms)

SIDAC MK1V115

V(BO) = 115 V MAX

TA = 25°C

tW, PULSE WIDTH (ms)

I , PEAK SURGE CURRENT (AMPS)

100

10 30 50 100

0.3 0.5 1 3 5 300

10%

TC = 25°C, DUTY CYCLE ≤ 1%

SEE NOTE FOR TIME

CONSTANT DEFINITION

MR2530L

MR2525L

MR2520L

, PEAK REVERSE POWER (WATTS)

RSM

700

500

7000

5000

3000

2000

1000

10 20 50 100

152200 500 100

τ, TIME CONSTANT (ms)

10000

b. Peak Surge Current versus

Ambient Temperature

25 7550 100

0125

0.6

0.8

0.4

NORMALIZED PEAK SURGE CURRENT

N = 2000 PULSES

Figure 12a. Peak Current

TC = 25°C, DUTY CYCLE ≤ 1%

SEE NOTE FOR TIME

CONSTANT DEFINITION

MR2530L

MR2525L

MR2520L

200

300

100

, PEAK REVERSE CURRENT (AMPS)

RSM

10 20 50 100

152200 500 1000

τ, TIME CONSTANT (ms)

TC = 25°C, DUTY CYCLE ≤ 1%

SEE NOTE FOR TIME

CONSTANT DEFINITION

MR2530L

MR2525L

MR2520L

τ, TIME CONSTANT (ms)

100

200

300

, PEAK REVERSE ENERGY (JOULES)

RSM

10 20 50 100

152200 500 1000

Figure 12c. Energy

Figure 12. Guaranteed Reverse Surge Design Limits for the

MR2525L & MR2530L Overload Transient Suppressors

TA, AMBIENT TEMPERATURE (°C)

NOTE: τ = RC

http://onsemi.com

115

Table 1. Measured Surge Current Capability of Transient Suppressors

Spec. Peak Current at Pulse Widths, I

(Amps)

Clamping

Dev

Powe

Die

1 ms 10 ms 20 ms 100 ms

amp

Factor

V1ms

Nor

Cos

Ty p

Titl

Part No. Ca

Volt

(En

Die

Siz Exp.Rect.Exp.Rect.Exp.Rect.Exp.Rect.

1ms

V100ms

Avalanche

Surge Supp.,

Overvolta

MR2520L

194 05

24−32 V 2.5KW

Peak

1502

mil 85 A 40 30 18 27V

22V +1.2

Avalanche

Rectifier

Overvoltage

Transient

Suppressor MR2525L

194−05

24−32 V 10KW

Peak

1962

mil 150 A 70 54 37 31

23 +1.3 4.0

1.5 W Zener 1N5936A

DO 41

30 V 1.5 W 37212 A 5 6 2.5 5 2 3 1.3 41

30 +1.4

Zener

Diode 1N5932A

DO−41

20 V

Cont.

mil 23 A 6 10 2.8 7 2.3 5 1.4 28

23 +1.2

3.2

Zener

Overvoltage

Transient

P6KE30

30 V 600 W 60243 A 14 14 5 10 4.5 5 2.5 41

32 +1.3

085

Zener Transient

Suppressor P6KE10

10 V

600

Peak

mil 24 A 12 9 5.5 16

13 +1.2

0.85

MOSORB

1.5KE30

41A 02

30 V 1500 W 104235 A 10 4 35

33 +1.1

MOSORB

1.5KE24

41A−02

24 V

1500

Peak

104

mil 45 A 14 6 30V

28V +1.1

1.8

MOV**

Metal

Oxide

V39MA2A Axial

Lead 28 V ǒ0.16

JoulesǓ3 mm 9 A 5 0.7 80V

60V

6A

0.7A 1.0

MOV** Oxide

Varistor V33ZA1 Radial

Lead 26 V ǒ1.0

JoulesǓ7 mm 35 4 A 105V

80V

35A

4A 1.4

**G.E.

Table 2. Measured Surge Current Of Thyristor Type Devices

Ipk @ tW

Voltage

Die

1 ms 10 ms Norm

Cost

Technology Device

age

Ratings Case

Size Exponent. Rectang. Exponent. Rectang.

ost

MKP9V130 Series 104 V−135 V

59 04

372 mil

40 A 13 A 16 A 8A

087

SIDAC

MKP9V240 Series 220 V−280 V 59−04 372 mil 31 A 15 A 20 A 8A 0.87

SIDAC MK1V135 Series 120 V−135 V

267 01

782 mil

140 A 80 A 55 A 30 A

MK1V270 Series 220 V−280 V 267−01 782 mil 170 A 60 A 90 A 28 A 1.1

SCR

MCR68 Series

25 V 400 V

922mil 300 A 170 A 1.2

SCR MCR69 Series 25 V−400 V

1502mil 700 A 400 A 1.9

MOS SCR MCR1000 Series 200 V−600 V

TO−220 127 mil

183 mil

250 A 170 A 9.3

*Normalized to G.E. MOV V39MA2A, Qty 1-99, 1984 Price

Additionally, the published non-repetitive peak power

ratings of the various zener diode packages are illustrated in

Figure 13. Figure 14 describes the typical derating factor for

repetitive conditions of duty cycles up to 20%. Using these

two empirically derived curves, the designer can then

determine the proper zener for the repetitive peak current

conditions.

At first glance the derating of curves of Figure 14 appear

to be in error as the 10 ms pulse has a higher derating factor

than the 10 μs pulse. However, when the mathematics of

multiplying the derating factor of Figure 14 by the peak

power value of Figure 13 is performed, the resultant

respective power and current capability of the device

follows the expected trend. For example, for a 5 W, 20 V

zener operating at a 1.0% duty cycle, the respective derating

factors for 10 μs and 10 ms pulses are 0.08 and 0.47. The

non-repetitive peak power capabilities for these two pulses

(10 μs and 10 ms) are about 1300 W and 50 W respectively,

resulting in repetitive power and current capabilities of

about 104 W and 24 W and consequently 5.2 A and 1.2 A.

MOV

All of the surge suppressors tested with the exception of

the MOV are semiconductors. The MOV is fabricated from

a ceramic (Zn0), non-linear resistor. This device has wide

acceptance for a number of reasons, but for many

applications, particularly those requiring good clamping

http://onsemi.com

116

Figure 13. Peak Power Ratings of Zener Diodes

Power is defined as VZ(NOM) x IZ(PK) where VZ(NOM) is the

nominal zener voltage measured at the low test current used for

voltage classification.

1N6267 SERIES

GLASS DO-35 & GLASS DO-41

250 mW TO 1 W TYPES

5 WATT TYPES

PULSE WIDTH (ms)

0.1

100

0.01 0.02

PPK(NOM)

, NOMINAL PEAK POWER (kW)

0.5

0.2

0.1

0.05

0.02

0.01 0.05 0.2 0.5 1 2 5 10

1 TO 3 W TYPES

PLASTIC DO-41

Figure 14. Typical Derating Factor for Duty Cycle

0.1 0.2 0.5 1 52 10 20 50 100

PULSE WIDTH

10 ms

1 ms

100 μs

10 μs

D, DUTY CYCLE (%)

DERATING FACTOR

0.7

0.5

0.3

0.2

0.02

0.1

0.07

0.05

0.03

0.01

factors, the MOV is found lacking; (clamping factor is

defined as the ratio of VZ at the test current to that at 1.0 mA).

This is photographically illustrated in Figure 15 which

compares a 27 V zener (1N6281) with a 27 V MOV

(V27ZA4). The input waveform, through a source

impedance resistance to the DUTs, was an exponentially

decaying voltage waveform of 90 V peak. Figures 15A and

B compare the output waveforms (across the DUTs) when

the source impedance was 500 Ω and Figures 15C and D for

a 50 Ω condition. The zener clamped at about 27 V for both

impedances whereas the MOV was about 40 V and 45 V

respectively.

Surge current capabilities of a comparably powered MOV

were also determined, as shown in the curve of Figure 16.

Although the MOV, a V39MA2A, is specified as a 28 V

Figure 15a.

Figure 15b.

SOURCE IMPEDANCE RS = 500 Ω

27 V MOV

G.E. V27ZA4, 4 JOULES CAPABILITY

27 V ZENER DIODE

ON Semiconductor 1N6281, APPROX. 1.5 JOULES

SOURCE IMPEDANCE RS = 500 Ω

continuous device (39 V ±10% at 1 mA) at the pulse widths

and currents tested, the resultant voltage VZ across the MOV

− 80 V at about 6 A − necessitated a high voltage fixture. This

was accomplished with a circuit similar to that of Figure 1B.

But MOVs do have their own niche in the marketplace, as

described in Table 3, the Relative Features of MOVs and

MOSORBs.

http://onsemi.com

117

Figure 15c.

Figure 15d.

27 V MOV

27 V ZENER DIODE

Figure 15. Clamping Characteristics of a

27 V Zener Diode and 27 V MOV

SOURCE IMPEDANCE RS = 50 Ω

G.E. V39MA2A MOV

VDCM = 28 V

VNOM = 39 V @ 1 mA

TA = 25°C

0.3

0.5

0.1

10 30 50 100

135

I , PEAK SURGE CURRENT (AMPS)

tW, PULSE WIDTH (ms)

Figure 16. Rectangular Surge Current Capability

Of The V39MA2A MOV

Table 3. Relative Features of MOVs

and MOSORBs

MOV

MOSORB/Zener Transient

Suppressor

High Clamping Factor Very good clamping close to

the operating voltage.

Symmetrically bidirectional Standard parts perform like

standard zeners. Symmetrical

bidirectional devices available

for many voltages.

Energy capability per dollar

usually much greater than a

silicon device. However, if

good clamping is required a

higher energy device would be

needed, resulting in higher

cost.

Good clamping

characteristics could reduce

overall cost.

Inherent wear out mechanism,

clamp voltage degrades after

every pulse, even when

pulsed below rated value.

No inherent wear out

mechanism.

Ideally suited for crude AC line

protection.

Ideally suited for precise DC

protection.

High single-pulse current

capability.

Medium multiple-pulse

current capability.

Degrades with overstress. Fails short with overstress.

Good high voltage capability. Limited high voltage capability

unless series devices are

used.

Limited low voltage capability. Good low voltage capability.

http://onsemi.com

118

SUMMARY

The surge current capabilities of low energy overvoltage

transient suppressors have been demonstrated, including

cost/performance comparison of rectifiers, zeners, thyristor

type suppressors, and MOVs. Both rectangular and

exponential testing have been performed with the described

testers. Additionally, the Rectangular Current Surge Tester

has the capability of measuring the diode junction

temperature of zeners and rectifiers at various power levels,

thus establishing safe operating limits.

REFERENCES

1. Cherniak, S., A Review of Transients and Their Means

of Suppression, ON Semiconductor Application Note

AN843.

2. Wilhardt, J., Transient Power Capability of Zener

Diodes, ON Semiconductor Application Note AN784.

3. Pshanenich, A., Characterizing the SCR for Crowbar

Applications, ON Semiconductor Application Note

AN879.

4. Pshaenich, A., The SIDAC, A New High Voltage

Trigger that Replaces Circuit Complexity and Cost,

ON Semiconductor Engineering Bulletin EB-106.

5. General Electric, Transient Voltage Suppression

Manual, Second Edition.

http://onsemi.com

119

MEASUREMENT OF ZENER VOLTAGE TO THERMAL EQUILIBRIUM

WITH PULSED TEST CURRENT

Prepared by

Herb Saladin

Discrete Power Application Engineering

INTRODUCTION

This paper discusses the zener voltage correlation

problem which sometimes exists between the manufacturer

and the customer’s incoming inspection. A method is shown

to aid in the correlation of zener voltage between thermal

equilibrium and pulse testing. A unique double pulsed

sample and hold test circuit is presented which improves the

accuracy of correlation.1

Several zener voltages versus zener pulsed test current

curves are shown for four package styles. An appendix is

attached for incoming inspection groups giving detailed

information on tolerances involved in correlation.

For many years the major difficulty with zener diode

testing seemed to be correlation of tight tolerance voltage

specifications where accuracy between different test setups

was the main problem. The industry standard and the EIA

Registration system adopted thermal equilibrium testing of

zener diodes as the basic test condition unless otherwise

specified. Thermal equilibrium was chosen because it was

the most common condition in the final circuit design and it

was the condition that the design engineers needed for their

circuit design and device selection. Thermal equilibrium

testing was also fairly simple to set-up for sample testing at

incoming inspection of standard tolerance zeners.

In recent years with the advent of economical

computerized test systems many incoming inspection areas

have implemented computer testing of zener diodes which

has been generating a new wave of correlation problems

between customers and suppliers of zener diodes.

The computerized test system uses short duration pulse

test techniques for testing zener diodes which does not

directly match the industry standard thermal equilibrium

test specifications.

This paper was prepared in an attempt to clarify the

differences between thermal equilibrium and short duration

pulse testing of zener diodes, to provide a test circuit that

allows evaluation at various pulse widths and a suggested

procedure for incoming inspection areas that will allow

meaningful correlation between thermal equilibrium and

pulse testing.

In the measurement of zener voltage (VZ), the

temperature coefficient effect combined with test current

heating can present a problem if one is attempting to

correlate VZ measurements made by another party (Final

Test, Quality Assurance or Incoming Inspection).2 This

paper is intended as an aid in determining VZ at some test

current (IZT) pulse width other than the pulse width used by

the manufacturer.

Thermal equilibrium (TE) is reached when diode junction

temperature has stabilized and no further change will occur

in VZ if the IZT time is increased.2 This absolute value can

vary depending on the mounting method and amount of

heatsinking. Therefore, thermal equilibrium conditions

have to be defined before meaningful correlation can exist.

Normalized VZ curves are shown for four package styles

and for three to five voltage ratings per package. Pulse

widths from 1 ms up to 100 seconds were used to arrive at

or near thermal equilibrium for all packages with a given

method of mounting.

Mounting

There are five conditions that can affect the correlation of

VZ measurements and are: 1) instrumentation, 2) TA, 3) IZT

time, 4) PD and 5) mounting. The importance of the first

four conditions is obvious but the last one, mounting, can

make the difference between good and poor correlation. The

mounting can have a very important part in VZ correlation

as it controls the amount of heat and rate of heat removal

from the diode by the mass and material in contact with the

diode package.

Two glass axial lead packages (DO-35 and DO-41),

curves (Figures 5 and 6) were measured with standard

Grayhill clips and a modified version of the Grayhill clips to

permit lead length adjustment.

Test Circuit

The test circuit (Figure 8) consists of standard CMOS

logic for pulse generation, inverting and delaying. The logic

drives three bipolar transistors for generation of the power

pulse for IZT. VZ is fed into an unique sample and hold (S/H)

circuit consisting of two high input impedance operational

amplifiers and a field effect transistor switch.

For greater accuracy in VZ measurements using a single

pulse test current, the FET switch is double pulsed. Double

pulsing the FET switch for charging the S/H capacitor

increases accuracy of the charge on the capacitor as the

second pulse permits charging the capacitor closer to the

final value of VZ.

The timing required for the two pulse system is shown in

waveform G-3C whereby the initial sample pulse is delayed

from time zero by a fixed 100 μs to allow settling time and

the second pulse is variable in time to measure the analog

input at that particular point. The power pulse (waveform

G-2D) must also encompass the second sample pulse.

http://onsemi.com

120

To generate these waveforms, four time delay monostable

multivibrators (MV) are required. Also, an astable MV, is

required for free-running operation; single pulsing is simply

initiated by a push-button switch S1. All of the pulse

generators are fashioned from two input, CMOS NOR gates;

thus three quad gate packages (MC14001) are required.

Gates 1A and 1B form a classical CMOS astable MV clock

and the other gates (with the exception of Gate 2D) comprise

the two input NOR gate configured monostable MV’s. The

Pulse Width variable delay output (Gate 1D) positions the

second sample pulse and also triggers the 100 μs Delay MV

and the 200 μs Extended Power Pulse MV, The respective

positive going outputs from gates 3A and 2C are diode

NOR’ed to trigger the Sample Gate MV whose output will

consequently be the two sample pulses. These pulses then

turn on the PNP transistor Q1 level translator and the

following S/H N-channel FET series switch Q2. Op amps

U4 and U5, configured as voltage followers, respectively

provide the buffered low output impedance drive for the

input and output of the S/H. Finally, the pulse extended

Power Gate is derived by NORing (Gate 2D) the Pulse

Width Output (Gate 1D) with the 200 μs MV output (Gate

2C). This negative aging gate then drives the Power

Amplifier, which, in turn, powers the D.U.T. The power

amplifier configuration consists of cascaded transistors

Q3−Q5, scaled for test currents up to 2 A.

Push button switch (S4) is used to discharge the S/H

capacitor. To adjust the zero control potentiometer, ground

the non-inverting input (Pin 3) of U4 and discharge the S/H

capacitor.

Testing

The voltage VCC, should be about 50 volts higher than the

D.U.T. and with RC selected to limit the IZT pulse to a value

making VZT IZT = 1/4 PD (max), thus insuring a good current

source. All testing was performed at a normal room

temperature of 25°C. A single pulse (manual) was used and

at a low enough rate that very little heat remained from the

previous pulse.

The pulse width MV (1C and 1D) controls the width of the

test pulse with a selector switch S3 (see Table 1 for capacitor

values). Fixed widths in steps of 1, 3 and 5 from 1 ms to 10

seconds in either a repetitive mode or single pulse is

available. For pulse widths greater than 10 seconds, a stop

watch was used with push button switch (S1) and with the

mode switch (S2) in the > 10 seconds position.

For all diodes with VZ greater than about 6 volts a resistor

voltage divider is used to maintain an input of about 6 V to

the first op amp (U4) so as not to overload or saturate this

device. The divider consists of R5 and R6 with R6 being

10 kΩ and R5 is selected for about a 6 V input to U4.

Precision resistors or accurate known values are required for

accurate voltage readout.

Table 1. S3 — Pulse Width

Switch

Position *C(μF) t(ms)

0.001

0.004

0.006

0.01

0.04

0.06

0.1

0.4

0.6

1.0

1.2

6.0

100

300

500

10K

*Approximate Values

Using Curves

Normalized VZ versus IZT pulse width curves are shown

in Figure 1 through 6. The type of heatsink used is shown or

specified for each device package type. Obviously, it is

beyond the scope of this paper to show curves for every

voltage rating available for each package type. The object

was to have a representative showing of voltages including

when available, one diode with a negative temperature

coefficient (TC).

These curves are actually a plot of thermal response

versus time at one quarter of the rated power dissipation.

With a given heatsink mounting, VZ can be calculated at

some pulse width other than the pulse width used to specify

VZ.

For example, refer to Figure 5 which shows normalized

VZ curves for the axial lead DO-35 glass package. Three

mounting methods are shown to show how the mounting

effects device heating and thus VZ. Curves are shown for a

3.9 V diode (1N5228B) which has a negative TC and a 12 V

diode (1N5242B) having a positive TC.

In Figure 5, the two curves generated using the Grayhill

mountings are normalized to VZ at TE using the

ON Semiconductor fixture. There is very little difference in

VZ at pulse widths up to about 10 seconds and mounting only

causes a very small error in VZ. The maximum error occurs

at TE between mountings and can be excessive if VZ is

specified at TE and a customer measures VZ at some narrow

pulse width and does not use a correction factor.

Using the curves of Figure 5, VZ can be calculated at any

pulse width based upon the value of VZ at TE which is

represented by 1 on the normalized VZ scale. If the 1N5242B

diode is specified at 12 V ± 1.0% at 90 seconds which is at

TE, VZ at 100 ms using either of the Grayhill clips curves

would be 0.984 of the VZ value at TE or 1 using the

ON Semiconductor fixture curve. If the negative TC diode

is specified at 3.9 V ± 1.0% at TE (90 seconds), VZ at 100

ms would be 1.011 of VZ at TE (using ON Semiconductor

fixture curve) when using the Grayhill Clips curves.

http://onsemi.com

121

In using the curves of Figure 5 and 6, it should be kept in

mind that VZ can be different at TE for the three mountings

because diode junction temperature can be different for each

mounting at TE which is represented by 1 on the VZ

normalized scale. Therefore, when the correlation of VZ

between parties is attempted, they must use the same type of

mounting or know what the delta VZ is between the two

mountings involved.

The Grayhill clips curves in Figure 6 are normalized to the

ON Semiconductor fixture at TE as in Figure 5. Figures 1

through 4 are normalized to VZ at TE for each diode and

would be used as Figures 5 and 6.

Measurement accuracy can be affected by test equipment,

power dissipation of the D.U.T., ambient temperature and

accuracy of the voltage divider if used on the input of the first

op-amp (U4). The curves of Figures 1 through 6 are for an

ambient temperature of 25°C, at other ambients, θVZ has to

be considered and is shown on the data sheet for the

1N5221B series of diodes. θVZ is expressed in mV/°C and

for the 1N5228B diode is about −2 mV/°C and for the

1N5242B, about 1.6 mV/°C. These values are multiplied by

the difference in TA from the 25°C value and either

subtracted or added to the calculated VZ depending upon

whether the diode has a negative or positive TC.

General Discussion

The TC of zener diodes can be either negative or positive,

depending upon die processing. Generally, devices with a

breakdown voltage greater than about 5 V have a positive

TC and diodes under about 5 V have a negative TC.

Conclusion

Curves showing VZ versus IZT pulse width can be used to

calculate VZ at a pulse width other than the one used to

specify VZ. A test circuit and method is presented to obtain

VZ with a single pulse of test current to generate VZ curves

of interest.

References

1. Al Pshaenich, “Double Pulsing S/H Increases System

Accuracy”; Electronics, June 16, 1983.

2. ON Semiconductor Zener Diode Manual, Series A,

1980.

http://onsemi.com

122

FIGURES 1 thru 8 — Conditions: Single Pulse, TA = 25°C, VZ IZT = 1/4 PD (Max) Each device normalized to VZ at TE.

AXIAL LEAD PACKAGES: MOUNTING STANDARD GRAYHILL CLIPS

Figure 1. DO-35 (Glass) 500 mW Device Figure 2. DO-41 (Glass) 1 Watt Device

Figure 3. DO-41 (Plastic) 1.5 Watt Device Figure 4. Case 17 (Plastic) 5 Watt Device

V , ZENER VOLTAGE (NORMALIZED)

VZ = 3.9 V

6.2 V

12 V

1.06

1.04

1.02

0.98

0.96

0.94

0.92 10 40 100

14 400 1K 4K 10K 40K 100K

PW, PULSE WIDTH (ms)

75 V

1.06

1.04

1.02

0.98

0.96

0.94

0.92

V , ZENER VOLTAGE (NORMALIZED)

10 40 100

14 400 1K 4K 10K 40K100K

PW, PULSE WIDTH (ms)

VZ = 3.9 V

6.2 V

12 V

V , ZENER VOLTAGE (NORMALIZED)

VZ = 3.3 V

6.2 V

12 V

1.06

1.04

1.02

0.98

0.96

0.94

0.92 10 40 100

14 400 1K 4K 10K 40K 100K

PW, PULSE WIDTH (ms)

V , ZENER VOLTAGE (NORMALIZED)

VZ = 3.9 V

6 V

13 V

1.06

1.04

1.02

0.98

0.96

0.94

0.92 10 40 100

14 400 1K 4K 10K 40K100K

PW, PULSE WIDTH (ms)

68 V 150 V

http://onsemi.com

123

THREE MOUNTING METHODS: DO-35 AND DO-41

Figure 5. DO-35 (Glass) 500 mW Device

Figure 6. DO-41 (Glass) 1 Watt Device

Figure 7. Standard Grayhill Clips

MOUNTING FIXTURE

1N5242B (VZ = 12 V)

1N5228B (VZ = 3.9 V)

GRAYHILL CLIPS

STANDARD, L = 11/16″

ON SEMICONDUCTOR

FIXTURE L = 1/2″

MOUNTINGS:

MODIFIED

L = 3/8″

1.022

1.018

1.014

1.012

1.008

1.004

0.996

10 40 100

14 400 1K 4K 10K 40K 100K

0.992

0.988

0.984

0.98

0.976

0.972

0.968

V , ZENER V

LTA

E (N

RMALIZED)

PW, PULSE WIDTH (ms)

1.004

10 40 100

1400 1K 4K 10K 40K 100K

0.992

0.988

0.984

0.98

0.976 4

0.996

MOUNTINGS:

GRAYHILL CLIPS

STANDARD, L = 11/16″

MODIFIED, L = 3/8″

ON SEMICONDUCTOR

FIXTURE

L = 1/2″

1N4742A

VZ = 12 V

V , ZENER VOLTAGE (NORMALIZED)

PW, PULSE WIDTH (ms)

1.41

1.69

.78

.75

2.31

GRAYHILL CLIPS

MODIFIED, L = 3/8″

STANDARD,

L = 11/16″

http://onsemi.com

124

Figure 8. Zener Voltage Double Pulsing S/H Test Circuit

0.1 +12 V

1A 1B

2A 2B

100

100 k

22 pF

1N914

1/2 MC14001

U3 12 k

22 k

4.7 k

10 k

12 k

1 k

MPSA42

POWER AMPLIFER

RCR5

68 k

2 W

DUT

Q5A*

MJE350

R6 10 k

1N914

MC1741

VIN

−

−12 V0.1 μF

2N4856

0.1 μF

−12 V

0.1

μF

0.1 μF

VO = VIN

ZERO

CONTROL

+ 12 V

25 k

VIN = VIN (10 k)

R5 + 10 k

VIN

+12 V VCC ≤ 250 V

1N914

33 pF

−12 V

10 k

22 k

47 k

2N3906

+12 V

SAMPLE GATE MV

510 pF

330 k

1/2 MC14001

+12 V

EXTENDED POWER-PULSE MV

1N914

47 k 47 k

0.001

μF

47 k

1N914

0.001 μF

200 μs

+12 V

680 k

510 pF

27 k

+12 V U3

3/4 MC14001

0.001 μF

VDD

510 pF

330 k

+12 V

VDD

+12 V

0.001 μF

27 k

100 μs

DELAY MV

1/4 MC14001

S2B

≈2 R1 R1

VDD

T1 ≈ 2.2R1C1

MC14001

1N914

MODE

SEL SW

S2A

10 k

ONE

SHOT

0.001 μF

>10 SEC

START SW

+12 V

100 k

FREE

RUN

C15

100 k

P.W.

CONTROL

T2 ≈ 0.6C2 (R2 + R3)

SEE

TABLE 1

PULSE

WIDTH

27 k

TIMING WAVEFORMS

GATE

G-1D

G-2A

G-2C

G-2D

G-3A

G-3C

100 μs

100 μs100 μs

200 μs

3906

−

+12 V +12 V

LF155J

**Tek Current Probe

**P6302/AM503

1 k

12 k 68 k

2 W

*FOR DUT

CURRENTS:

200 mA ≤ IZT ≤ 2 A

VCC ≤ 250 V

DUT

Q5A

MJE35

Q5B

MJE

5850

VCC ≤ 250 V

http://onsemi.com

125

APPENDIX A

Recommended Incoming Inspection Procedures

Zener Voltage Testing

Pulsed versus Thermal Equilibrium

This section is primarily for use of incoming inspection

groups. The subject covered is the measurement of zener

voltage (VZ) and the inherent difficulty of establishing

correlation between supplier and buyer when using pulsed

test techniques. This difficulty, in part, is due to the

interpretation of the data taken from the variety of available

testers and in some cases even from the same model types.

It is therefore, our intent to define and reestablish a

standardized method of measurement to achieve correlation

no matter what test techniques are being used. This

standardization will guarantee your acceptance of good

product while maintaining reliable correlation.

DEFINITION OF TERMS

Temperature Coefficient (TC):

The temperature stability of zener voltages is sometimes

expressed by means of the temperature coefficient (TC).

This parameter is usually defined as the percent voltage

change across the device per degree centigrade, or as a

specific voltage change per degree centigrade. Temperature

changes during test are due to the self heating effects caused

by the dissipation of power in the zener junction. The VZ will

change due to this temperature change and will exhibit a

positive or negative TC, depending on the zener voltage.

Generally, devices with a zener voltage below five volts will

have a negative TC and devices above five volts will exhibit

a positive TC.

Thermal Equilibrium (TE)

Thermal equilibrium (TE) is reached when the diode

junction temperature has stabilized and no further change

will occur. In thermal equilibrium, the heat generated at the

junction is removed as rapidly as it is created, hence, no

further temperature changes.

MEASURING ZENER VOLTAGE

The zener voltage, being a temperature dependent

parameter, needs to be controlled for valid VZ correlation.

Therefore, so that a common base of comparison can be

established, a reliable measure of VZ can only occur when

all possible variables are held constant. This common base

is achieved when the device under test has had sufficient

time to reach thermal equilibrium (heatsinking is required to

stabilize the lead or case temperature to a specified value for

stable junction temperatures). The device should also be

powered from a constant current source to limit changes of

power dissipated and impedance.

All of the above leads us to an understanding of why

various pulse testers will give differing VZ readings; these

differences are, in part, due to the time duration of test (pulse

width), duty cycle when data logging, contact resistance,

tolerance, temperature, etc. To resolve all of this, one only

needs a reference standard to compare their pulsed results

against and then adjust their limits to reflect those

differences. It should be noted that in a large percentage of

applications the zener diode is used in thermal equilibrium.

ON Semiconductor guarantees all of it’s axial leaded

zener products (unless otherwise specified) to be within

specification ninety (90) seconds after the application of

power while holding the lead temperatures at 30 ± 1°C, 3/8

of an inch from the device body, any fixture that will meet

that criteria will correlate. 30°C was selected over the

normally specified 25°C because of its ease of maintenance

(no environmental chambers required) in a normal room

ambient. A few degrees variation should have negligible

effect in most cases. Hence, a moderate to large heatsink in

most room ambients should suffice.

Also, it is advisable to limit extraneous air movements

across the device under test as this could change thermal

equilibrium enough to affect correlation.

SETTING PULSED TESTER LIMITS

Pulsed test techniques do not allow a sufficient time for

zener junctions to reach TE. Hence, the limits need to be set

at different values to reflect the VZ at lower junction

temperatures. Since there are many varieties of test systems

and possible heatsinks, the way to establish these limits is to

actually measure both TE and pulsed VZ on a serialized

sample for correlation.

The following examples show typical delta changes in

pulsed versus TE readings. The actual values you use for

pulsed conditions will depend on your tester. Note, that there

are examples for both positive and negative temperature

coefficients. When setting the computer limits for a positive

TC device, the largest difference is subtracted from the

upper limit and the smallest difference is subtracted from the

lower limit. In the negative coefficient example the largest

change is added to the lower limit and the smallest change

is added to the upper limit.

http://onsemi.com

126

ON Semiconductor Zeners

•Thermal equilibrium specifications:

VZ at 10 mA, 9 V minimum, 11 V maximum:

(Positive TC)

TE Pulsed Difference

9.53 V

9.35 V

9.46 V

9.56 V

9.50 V

9.45 V

9.38 V

9.83 V

9.49 V

9.40 V

−0.08 V

−0.07 V

−0.08 V

−0.07 V

−0.10 V

Computer test limits:

Set VZ max. limit at 11 V − 0.10 V = 10.9 V

Set VZ min. limit at 9 V − 0.07 V = 8.93 V

•Thermal equilibrium specifications:

VZ at 10 mA, 2.7 V minimum, 3.3 V maximum:

(Negative TC)

TE Pulsed Difference

2.78 V

2.84 V

2.78 V

2.86 V

2.82 V

2.83 V

2.91 V

2.84 V

2.93 V

2.87 V

+0.05 V

+0.07 V

+0.05 V

+0.07 V

+0.05 V

Computer test limits:

Set VZ min. limit at 2.7 V + 0.07 V = 2.77 V

Set VZ max. limit at 3.3 V + 0.05 V = 3.35 V

April, 2005 − Rev. 1

127 Publication Order Number:

NLAS3158/D

ON Semiconductor and are registered trademarks of Semiconductor Components Industries, LLC (SCILLC). SCILLC reserves the right to make changes without further notice

to any products herein. SCILLC makes no warranty, representation or guarantee regarding the suitability of its products for any particular purpose, nor does SCILLC assume any

liability arising out of the application or use of any product or circuit, and specifically disclaims any and all liability, including without limitation special, consequential or incidental

damages. “Typical” parameters which may be provided in SCILLC data sheets and/or specifications can and do vary in different applications and actual performance may vary over

time. All operating parameters, including “Typicals” must be validated for each customer application by customer’s technical experts. SCILLC does not convey any license under

its patent rights nor the rights of others. SCILLC products are not designed, intended, or authorized for use as components in systems intended for surgical implant into the body,

or other applications intended to support or sustain life, or for any other application in which the failure of the SCILLC product could create a situation where personal injury or death

may occur. Should Buyer purchase or use SCILLC products for any such unintended or unauthorized application, Buyer shall indemnify and hold SCILLC and its officers, employees,

subsidiaries, affiliates, and distributors harmless against all claims, costs, damages, and expenses, and reasonable attorney fees arising out of, directly or indirectly, any claim of

personal injury or death associated with such unintended or unauthorized use, even if such claim alleges that SCILLC was negligent regarding the design or manufacture of the part.

SCILLC is an Equal Opportunity/Affirmative Action Employer. This literature is subject to all applicable copyright laws and is not for resale in any manner.

PUBLICATION ORDERING INFORMATION

N. American Technical Support: 800−282−9855 Toll Free

USA/Canada

Japan: ON Semiconductor, Japan Customer Focus Center

2−9−1 Kamimeguro, Meguro−ku, Tokyo, Japan 153−0051

Phone: 81−3−5773−3850

HBD854/D

LITERATURE FULFILLMENT:

Literature Distribution Center for ON Semiconductor

P.O. Box 61312, Phoenix, Arizona 85082−1312 USA

Phone: 480−829−7710 or 800−344−3860 Toll Free USA/Canada

Fax: 480−829−7709 or 800−344−3867 Toll Free USA/Canada

Email: orderlit@onsemi.com

ON Semiconductor Website: http://onsemi.com

Order Literature: http://www.onsemi.com/litorder

For additional information, please contact your

local Sales Representative.