10ISO120/121
because IMV-induced errors behave like input-referred error
signals. To predict the total IMR, divide the isolation voltage
by the IMR shown in IMR vs Frequency performance curve
and compute the amplifier response to this input-referred
error signal from the data given in the Signal Response vs
Carrier Frequency performance curve. Due to effects of very
high-frequency signals, typical IMV performance can be
achieved only when dV/dT of the isolation mode voltage
falls below 1000V/µs. For convenience, this is plotted in the
typical performance curves for the ISO120 and ISO121 as a
function of voltage and frequency for sinusoidal voltages.
When dV/dT exceeds 1000V/µs but falls below 20kV/µs,
performance may be degraded. At rates of change above
20kV/µs, the amplifier may be damaged, but the barrier
retains its full integrity. Lowering the power supply voltages
below ±15V may decrease the dV/dT to 500V/µs for typical
performance, but the maximum dV/dT of 20kV/µs remains
unchanged.
Leakage current is determined solely by the impedance of
the 2pF barrier capacitance and is plotted in the Isolation
Leakage Current vs Frequency curve.
ISOLATION VOLTAGE RATINGS
Because a long-term test is impractical in a manufacturing
situation, the generally accepted practice is to perform a
production test at a higher voltage for some shorter time. The
relationship between actual test voltage and the continuous
derated maximum specification is an important one. Histori-
cally, Burr-Brown has chosen a deliberately conservative
one: VTEST = (2 X ACrms continuous rating) + 1000V for 10
seconds, followed by a test at rated ACrms voltage for one
minute. This choice was appropriate for conditions where
system transients are not well defined.
Recent improvements in high-voltage stress testing have
produced a more meaningful test for determining maximum
permissible voltage ratings, and Burr-Brown has chosen to
apply this new technology in the manufacture and testing of
the ISO120 and ISO121.
Partial Discharge
When an insulation defect such as a void occurs within an
insulation system, the defect will display localized corona or
ionization during exposure to high-voltage stress. This ion-
ization requires a higher applied voltage to start the dis-
charge and lower voltage to maintain it or extinguish it once
started. The higher start voltage is known as the inception
voltage, while the extinction voltage is that level of voltage
stress at which the discharge ceases. Just as the total insula-
tion system has an inception voltage, so do the individual
voids. A voltage will build up across a void until its incep-
tion voltage is reached, at which point the void will ionize,
effectively shorting itself out. This action redistributes elec-
trical charge within the dielectric and is known as partial
discharge. If, as is the case with AC, the applied voltage
gradient across the device continues to rise, another partial
discharge cycle begins. The importance of this phenomenon
is that, if the discharge does not occur, the insulation system
retains its integrity. If the discharge begins, and is allowed
to continue, the action of the ions and electrons within the
defect will eventually degrade any organic insulation system
in which they occur. The measurement of partial discharge
is still useful in rating the devices and providing quality
control of the manufacturing process. Since the ISO120 and
ISO121 do not use organic insulation, partial discharge is
non-destructive.
The inception voltage for these voids tends to be constant, so
that the measurement of total charge being redistributed
within the dielectric is a very good indicator of the size of
the voids and their likelihood of becoming an incipient
failure. The bulk inception voltage, on the other hand, varies
with the insulation system, and the number of ionization
defects and directly establishes the absolute maximum volt-
age (transient) that can be applied across the test device
before destructive partial discharge can begin. Measuring
the bulk extinction voltage provides a lower, more conserva-
tive voltage from which to derive a safe continuous rating.
In production, measuring at a level somewhat below the
expected inception voltage and then derating by a factor
related to expectations about system transients is an
accepted practice.
Partial Discharge Testing
Not only does this test method provide far more qualitative
information about stress-withstand levels than did previous
stress tests, but it provides quantitative measurements from
which quality assurance and control measures can be based.
Tests similar to this test have been used by some manufac-
turers, such as those of high-voltage power distribution
equipment, for some time, but they employed a simple
measurement of RF noise to detect ionization. This method
was not quantitative with regard to energy of the discharge,
and was not sensitive enough for small components such as
isolation amplifiers. Now, however, manufacturers of HV
test equipment have developed means to quantify partial
discharge. VDE, the national standards group in Germany
and an acknowledged leader in high-voltage test standards,
has developed a standard test method to apply this powerful
technique. Use of partial discharge testing is an improved
method for measuring the integrity of an isolation barrier.
To accommodate poorly-defined transients, the part under
test is exposed to voltage that is 1.6 times the continuous-
rated voltage and must display ≤5pC partial discharge level
in a 100% production test.