19-3141; Rev 1; 6/05 KIT ATION EVALU E L B A IL AVA Dual-Synchronous Buck Controllers for Point-ofLoad, Tracking, and DDR Memory Power Supplies Features The MAX8537/MAX8539 controllers provide a complete power-management solution for both double-data-rate (DDR) and combiner supplies. The MAX8537 and MAX8539 are configured for out-of-phase and in-phase DDR power-supply operations, respectively, and generate three outputs: the main memory voltage (VDDQ), the tracking sinking/sourcing termination voltage (VTT), and the termination reference voltage (VTTR). The MAX8538 is configured as a dual out-of-phase controller for pointof-load supplies. Each buck controller can source or sink up to 25A of current, while the termination reference can supply up to 15mA output. MAX8537/MAX8539: Complete DDR Supplies MAX8538: Dual Nontracking Controller Out-of-Phase (MAX8537/MAX8538) or In-Phase (MAX8539) Operation 4.5V to 23V Wide Input Range (Operate Down to 1.8V Input with External 5V Supply) Tracking Supply Maintains VTT = VTTR = 1/2 VDDQ Adjustable Output from 0.8V to 3.6V with 1% Accuracy The MAX8537/MAX8538/MAX8539 use constantfrequency voltage-mode architecture with operating frequencies of 200kHz to 1.4MHz. An internal highbandwidth (25MHz) operational amplifier is used as an error amplifier to regulate the output voltage. This allows fast transient response, reducing the number of output capacitors. An all-N-FET design optimizes efficiency and cost. The MAX8537/MAX8538/MAX8539 have a 1% accurate reference. The second synchronous buck controller in the MAX8537/MAX8539 and the VTTR amplifier generate 1/2 VDDQ voltage for VTT and VTTR, and track the VDDQ within 1%. This family of controllers uses a high-side current-sense architecture for current limiting. ILIM pins allow the setting of an adjustable, lossless current limit for different combinations of load current and RDSON. Alternately, more accurate overcurrent limit is achieved by using a sense resistor in series with the high-side FET. Overvoltage protection is achieved by latching off the high-side MOSFET and latching on the low-side MOSFET when the output voltage exceeds 17% of its set output. Independent enable, power-good, and soft-start features enhance flexibility. VTTR Reference Sources and Sinks Up to 15mA 200kHz to 1.4MHz Adjustable Switching Frequency All-Ceramic Design Achievable >90% Efficiency Independent POK_ and EN_ Adjustable Soft-Start and Soft-Stop for Each Output Lossless Adjustable-Hiccup Current Limit Output Overvoltage Protection 28-Pin QSOP Package Ordering Information PART TEMP RANGE PINPACKAGE OPERATION MAX8537EEI -40C to +85C 28 QSOP Out-of-phase tracking MAX8537EEI+ -40C to +85C 28 QSOP Out-of-phase tracking MAX8538EEI -40C to +85C 28 QSOP Out-of-phase nontracking DDR Memory Power Supplies Notebooks and Desknotes Servers and Storage Systems Broadband Routers XDSL Modems and Routers MAX8538EEI+ -40C to +85C 28 QSOP Out-of-phase nontracking MAX8539EEI -40C to +85C 28 QSOP In-phase tracking MAX8539EEI+ -40C to +85C 28 QSOP In-phase tracking Power DSP Core Supplies Power Combiner in Advanced VGA Cards Networking Systems RAMBUS Memory Power Supplies +Denotes lead-free package. Applications Pin Configurations appear at end of data sheet. ________________________________________________________________ Maxim Integrated Products For pricing, delivery, and ordering information, please contact Maxim/Dallas Direct! at 1-888-629-4642, or visit Maxim's website at www.maxim-ic.com. 1 MAX8537/MAX8538/MAX8539 General Description MAX8537/MAX8538/MAX8539 Dual-Synchronous Buck Controllers for Point-ofLoad, Tracking, and DDR Memory Power Supplies ABSOLUTE MAXIMUM RATINGS V+ to GND ..............................................................-0.3V to +25V AVL, VL to GND........................................................-0.3V to +6V PGND to GND .......................................................-0.3V to +0.3V FB_, EN_, POK_ to GND...........................................-0.3V to +6V REFIN, VTTR, FREQ, SS_, COMP_ to GND....-0.3V to (AVL + 0.3V) BST_, ILIM_ to GND ...............................................-0.3V to +30V DH1 to LX1 ...............................................-0.3V to (BST1 + 0.3V) DH2 to LX2 ...............................................-0.3V to (BST2 + 0.3V) LX_ to BST_ ..............................................................-6V to +0.3V LX_ to GND................................................................-2V to +25V DL_ to PGND ................................................-0.3V to (VL + 0.3V) Continuous Power Dissipation (TA = +70C) 28-Pin QSOP (derate 10.8mW/C above +70C)........860mW Operating Temperature Range ...........................-40C to +85C Junction Temperature ......................................................+150C Storage Temperature Range .............................-65C to +150C Lead Temperature (soldering, 10s) ................................+300C Stresses beyond those listed under "Absolute Maximum Ratings" may cause permanent damage to the device. These are stress ratings only, and functional operation of the device at these or any other conditions beyond those indicated in the operational sections of the specifications is not implied. Exposure to absolute maximum rating conditions for extended periods may affect device reliability. ELECTRICAL CHARACTERISTICS (V+ = 12V, EN_ = VL, BST_ = 6V, LX_ = 1V, VL load = 0mA, CVL = 10F (ceramic), REFIN = 1.25V, PGND = AGND = FB_ = ILIM_ = 0V, CSS = 10nF, CVTTR = 1F, RFREQ = 20k, DH_ = open, DL_ = open, POK_ = open, circuit of Figure 1, TA = 0C to +85C, unless otherwise noted.) PARAMETER CONDITIONS MIN TYP MAX UNITS 23.0 V GENERAL V+ Operating Range VL regulator drops out below 5.5V (Note 1) 4.5 V+/ VL Operating Range VL is externally generated (Note 1) 4.5 5.5 V V+ Operating Supply Current IL(VL) = 0, FB_ forced 50mV above threshold 3.5 7 mA V+ Standby Supply Current IL(VL) = 0, BST_ = VL, EN = LX_ = FB_ = 0V 350 700 A VL REGULATOR Output Voltage 5.5V < V+ < 23V, 1mA < ILOAD < 70mA 4.75 5 5.25 V VL Undervoltage-Lockout Trip Level Rising edge, hysteresis = 550mV (typ) (trip level is typically 85% of VL) 4.18 4.3 4.42 V Output Current This is for gate current of DL_ /DH_ drivers, C(VL) = 1F/10mA ceramic capacitor 70 mA Thermal Shutdown Rising temperature, typical hysteresis = 10C +160 C CURRENT-LIMIT THRESHOLD (all current limits are tested at V+ = VL = 4.5V and 5.5V) ILIM Sink Current ILIM_ = LX - 200mV, 1.8V < LX < 23V, BST = LX +5V 180 200 220 A SOFT-START Soft-Start Source Current SS_ = 100mV -7 -5 -3 A Soft-Start Sink Current SS_ = 0.8 or REFIN 3 5 7 A 0.8 or REFIN Soft-Start Full-Scale Voltage V FREQUENCY Low End of Range RFREQ = 100k, V+ = VL = 5V 160 200 240 kHz Intermediate Range RFREQ = 20k, V+ = VL = 5V 800 1000 1200 kHz High End of Range RFREQ = 14.3k, V+ = VL = 5V 1120 1400 1680 kHz Maximum Duty Cycle 2 RFREQ = 100k 95 RFREQ = 20k 80 RFREQ = 14.3k 72 _______________________________________________________________________________________ % Dual-Synchronous Buck Controllers for Point-ofLoad, Tracking, and DDR Memory Power Supplies (V+ = 12V, EN_ = VL, BST_ = 6V, LX_ = 1V, VL load = 0mA, CVL = 10F (ceramic), REFIN = 1.25V, PGND = AGND = FB_ = ILIM_ = 0V, CSS = 10nF, CVTTR = 1F, RFREQ = 20k, DH_ = open, DL_ = open, POK_ = open, circuit of Figure 1, TA = 0C to +85C, unless otherwise noted.) PARAMETER Minimum Duty Cycle TYP MAX RFREQ = 100k CONDITIONS MIN 2.4 4 RFREQ = 20k 12 18 16 25 DH_ Minimum Off-Time RFREQ = 14.3k 140 200 DH_ Minimum On-Time 120 UNITS % ns ns ERROR AMPLIFIER FB_ Input Bias Current VFB_ = 0.8V 250 nA FB1 Input-Voltage Set Point Over line and load 0.792 0.800 0.808 V MAX8538 0.792 0.800 0.808 MAX8537/MAX8539, REFIN = 0.9V 0.894 0.900 0.906 72 >80 dB Op-Amp Gain Bandwidth 25 MHz Op-Amp Output-Voltage Slew Rate 15 V/s Break-Before-Make Time 30 ns DH1, DH2 On-Resistance in Low State 0.9 2.5 DH1, DH2 On-Resistance in High State 1.3 2.5 DL1, DL2 On-Resistance in Low State 0.7 1.5 DL1, DL2 On-Resistance in High State 1.6 2.8 0.8 V FB2 Input-Voltage Set Point Op-Amp Open-Loop Voltage Gain COMP_ = 1.3V to 2.3V V DRIVERS LOGIC INPUTS (EN_) Input Low Level 4.5V < VL < 5.5V Input High Level 4.5V < VL < 5.5V 2.4 Input Bias Current 0V to 5.5V -1 VTTR Output Voltage Range Source or sink 15mA 0.5 VTTR Output Accuracy -15mA IVTTR +15mA, REFIN = 0.9V or 1.25V -1.0 REFIN = 0.9V or 1.25V -250 V +0.1 +1 A 2.5 V +1.0 % +250 nA VTTR REFIN REFIN REFIN Input Bias Current _______________________________________________________________________________________ 3 MAX8537/MAX8538/MAX8539 ELECTRICAL CHARACTERISTICS (continued) MAX8537/MAX8538/MAX8539 Dual-Synchronous Buck Controllers for Point-ofLoad, Tracking, and DDR Memory Power Supplies ELECTRICAL CHARACTERISTICS (continued) (V+ = 12V, EN_ = VL, BST_ = 6V, LX_ = 1V, VL load = 0mA, CVL = 10F (ceramic), REFIN = 1.25V, PGND = AGND = FB_ = ILIM_ = 0V, CSS = 10nF, CVTTR = 1F, RFREQ = 20k, DH_ = open, DL_ = open, POK_ = open, circuit of Figure 1, TA = 0C to +85C, unless otherwise noted.) PARAMETER CONDITIONS REFIN Input Voltage Range REFIN Undervoltage-Lockout Trip Level MIN TYP 0.5 Rising and falling edge, hysteresis = 15mV MAX UNITS 2.5 V 0.4 0.45 0.5 V OUTPUT-VOLTAGE FAULT COMPARATORS Upper FB2 Fault Threshold Rising voltage, hysteresis = 15mV 115 117 120 % of REFIN Lower FB2 Fault Threshold Falling voltage, hysteresis = 15mV 68 70 72 % of REFIN Upper FB1 Fault Threshold Rising voltage, hysteresis = 15mV 115 117 120 % of 0.8V Lower FB1 Fault Threshold Falling voltage, hysteresis = 15mV 68 70 72 % of 0.8V POWER-OK OUTPUT (POK_) POK_ Delay Clock cycles 64 Upper FB2 POK_ Threshold Rising voltage, hysteresis = 20mV 110 112 114 % of REFIN Lower FB2 POK_ Threshold Falling voltage, hysteresis = 20mV 86 88 90 % of REFIN Upper FB1 POK_ Threshold Rising voltage, hysteresis = 20mV 110 112 114 % of 0.8V Lower FB1 POK_ Threshold Falling voltage, hysteresis = 20mV 86 88 90 % of 0.8V POK_ Output Low Level ISINK = 2mA 0.4 V POK_ Output High Leakage POK_ = 5.5V 1 A ELECTRICAL CHARACTERISTICS (Note 2) (V+ = 12V, EN_ = VL, BST_ = 6V, LX_ = 1V, VL load = 0mA, CVL = 10F (ceramic), REFIN = 1.25V, PGND = AGND = FB_ = ILIM_ = 0V, CSS = 10nF, CVTTR = 1F, RFREQ = 20k, DH_ = open, DL_ = open, POK_ = open, circuit of Figure 1, TA = -40C to +85C, unless otherwise noted.) PARAMETER CONDITIONS MIN TYP MAX UNITS V+ Operating Range VL regulator drops out below 5.5V (Note 1) 4.75 23.00 V FB_ Input-Voltage Set Point Over line and load 0.788 0.800 0.812 V FB2 Input-Voltage Set Point MAX8537/MAX8539, REFIN = 0.9V 0.891 0.900 0.909 V VTTR Output Accuracy -15mA < IVTTR +15mA, REFIN = 0.9V or 1.25V -1 REFIN +1 % Note 1: Operating supply range is guaranteed by the VL line-regulation test. User must short V+ to VL if a fixed 5V supply is used (i.e., if V+ is less than 5.5V). Note 2: Specifications to -40C are guaranteed by design, not production tested. 4 _______________________________________________________________________________________ Dual-Synchronous Buck Controllers for Point-ofLoad, Tracking, and DDR Memory Power Supplies VDDQ = 1.8V 75 70 80 2.51 VDDQ = 1.8V 75 70 65 60 60 55 55 2.49 2.47 50 100 10 2.45 1 LOAD CURRENT (A) MAX8537 toc04 1.30 1.28 1.27 1.26 1.26 VTTR 1.27 1.25 1.25 1.24 1.24 1.23 1.23 1.22 1.22 1.21 1.21 1.20 OUTPUT VOLTAGE (V) 1.28 1.20 0 3 6 9 12 15 0 LOAD CURRENT (A) 5 10 15 20 25 20 VDDQ IOUT_VDDQ = 20A IOUT_VTT = 12A IOUT_VTTR = 15mA VTT AND VTTR 14 13 STARTUP AND SHUTDOWN MAX8537 toc08 V+ VIN 5V/div VDDQ 1V/div VDDQ VTT 8.5V/div VTTR 8.5V/div VTT 15 INPUT VOLTAGE (V) POWER-DOWN MAX8537 toc07 2.6 2.5 2.4 2.3 2.2 2.1 2.0 1.9 1.8 1.7 1.6 1.5 1.4 1.3 1.2 12 LOAD CURRENT (mA) POWER-UP 10 OUTPUT VOLTAGE vs. INPUT VOLTAGE VIN = 12V 1.29 5 LOAD CURRENT (A) VTTR vs. LOAD CURRENT VIN = 12V 1.29 0 LOAD CURRENT (A) VTT vs. LOAD CURRENT 1.30 100 10 MAX8537 toc05 1 VTT 2.53 85 65 50 MAX8537 toc03 90 VIN = 12V MAX8537 toc06 80 SENSE RESISTOR = 3m VIN = 12V VDDQ = 2.5V VDDQ vs. LOAD CURRENT 2.55 VDDQ 85 95 EFFICIENCY (%) EFFICIENCY (%) SENSE RESISTOR = 0 VIN = 12V VDDQ = 2.5V 90 MAX8537 toc01 95 VDDQ EFFICIENCY vs. LOAD CURRENT 100 MAX8537 toc02 VDDQ EFFICIENCY vs. LOAD CURRENT 100 MAX8537 toc09 5V/div EN1/EN2 5V/div 2V/div VDDQ 2V/div VTT 1V/div 2V/div 2V/div VVTTR 4ms/div 4ms/div MAX8537/MAX8538/MAX8539 Typical Operating Characteristics (Circuit of Figure 1, TA = +25C, 400kHz switching frequency, VIN = 12V, unless otherwise noted.) VTTR 1V/div 1ms/div _______________________________________________________________________________________ 5 Typical Operating Characteristics (continued) (Circuit of Figure 1, TA = +25C, 400kHz switching frequency, VIN = 12V, unless otherwise noted.) POWER-OK VDDQ LOAD TRANSIENT AND VTT TRACKING VTT STARTUP AND SHUTDOWN MAX8537 toc10 MAX8537 toc11 VDDQ 2V/div VTT 1V/div VTT 1V/div VTTR 1V/div POK2 5V/div VDDQ 2V/div IOUT_VDDQ = 20A IOUT_VTT = 8A VTT 50mV/div VTTR 50mV/div 20A di/dt = 5A/s 10A VDDQ_IOUT 10A/div VTT_IOUT = 12A VTTR = 15mA 2ms/div 2ms/div 200s/div VDDQ = 2.5V AT 20A BODE PLOT, VIN = 12V VTT LOAD-TRANSIENT RESPONSE MAX8537 toc13 di/dt = 1A/s -8A 160 140 120 dB (DEGREES) 8A MAX8537 toc14 180 VTT AC-COUPLED 50mV/div VTTR AC-COUPLED 50mV/div VDDQ AC-COUPLED 50mV/div 100 80 60 40 20 VTT_IOUT 10A/div IOUT_VDDQ = 20A IOUT_VTTR = 15mA 0 -20 -40 200s/div 100 1k 10k 1M 100k kHz VTT = 1.25V AT 12A BODE PLOT, VIN = 12V SHORT CIRCUIT AND RECOVERY MAX8537 toc16 MAX8537 toc15 150 125 VOUT1 1V/div VOUT2 1V/div 100 75 50 IL1 10V/div IIN 5A/div 25 0 -25 -50 102 103 104 105 106 10ms/div Hz 6 MAX8537 toc12 VDDQ 100mV/div EN2 5V/div POK1 5V/div dB (DEGREES) MAX8537/MAX8538/MAX8539 Dual-Synchronous Buck Controllers for Point-ofLoad, Tracking, and DDR Memory Power Supplies _______________________________________________________________________________________ Dual-Synchronous Buck Controllers for Point-ofLoad, Tracking, and DDR Memory Power Supplies PIN NAME (MAX8537/ MAX8539) NAME (MAX8538) 1 BST2 BST2 Bootstrap Input to Power Internal High-Side Driver for Step-Down 2. Connect to an external capacitor and diode according to Figure 1. 2 DH2 DH2 High-Side Gate-Driver Output for Step-Down 2. Swings from LX2 to BST2. 3 LX2 LX2 External Inductor Input for Step-Down 2. Connect to the switched side of the inductor. LX2 serves as the lower supply-voltage rail for the DH2 high-side gate driver and the current-limit circuitry. FUNCTION 4 ILIM2 ILIM2 Output Current-Limit Setting for Step-Down 2. Connect a resistor from ILIM2 to the drain of the step-down 2 high-side MOSFET, or to the junction of the source of the high-side MOSFET and the current-sense resistor to set the current-limit threshold. See the Current-Limit Setting section. 5 POK1 POK1 Open-Drain Output. High impedance when step-down 1 is within 12% of its regulation voltage. POK1 is pulled low in shutdown. 6 DL2 DL2 7 POK2 POK2 8 EN2 EN2 Enable Input for Step-Down 2 (also for VTTR for the MAX8537 and MAX8539) 9 EN1 EN1 Enable Input for Step-Down 1 10 FREQ FREQ 11 COMP2 COMP2 12 FB2 FB2 Feedback Input for Step-Down 2 with VREFIN as the Threshold. User must have impedance <40k. 13 SS2 SS2 Soft-Start for Step-Down 2. Connect a capacitor to GND to set the soft-start time. REFIN -- -- N.C. 15 GND GND Analog Ground for Internal Circuitry 16 SS1 SS1 Soft-Start for Step-Down 1. Connect a capacitor to GND to set the soft-start time. 17 FB1 FB1 Feedback Input for Step-Down 1 with 0.8V Threshold. User must have impedance <40k. 18 COMP1 COMP1 VTTR -- -- GND Analog Ground for Internal Circuitry 20 AVL AVL Analog VL Input Pin. Connect to VL through a 4.7 resistor. Bypass with a 0.1F (or larger) ceramic capacitor to GND. 21 V+ V+ Input Supply Voltage 14 19 Low-Side Gate-Driver Output for Step-Down 2. Swings from PGND to VL. Open-Drain Output. High impedance when step-down 2 is within 12% of its regulation voltage. POK2 is pulled low in shutdown or if REFIN is undervoltage. Frequency Adjust. Connect a resistor from this pin to ground to set the frequency. The range of the FREQ resistor is 163k, 20k, and 100k (corresponding to 1.4MHz, 1.0MHz, and 200kHz). Compensation Pin for Step-Down 2. Connect to compensation networks. Reference Input for VTT and VTTR. Connect it to a resistor-divider from VDDQ. REFIN common-mode voltage range is 0.5V to 2.5V. Current through the divider-resistors must be 100A. For the MAX8538, connect pin 14 to GND. Compensation Pin for Step-Down 1. Connect to compensation networks. VTTR Output Capable of Sourcing and Sinking Up to 15mA. Always bypass with a 1F ceramic capacitor (or larger) to GND. _______________________________________________________________________________________ 7 MAX8537/MAX8538/MAX8539 Pin Description Dual-Synchronous Buck Controllers for Point-ofLoad, Tracking, and DDR Memory Power Supplies MAX8537/MAX8538/MAX8539 Pin Description (continued) PIN 22 NAME (MAX8537/ MAX8539) VL NAME (MAX8538) FUNCTION VL Internal 5V Linear Regulator to Power the IC. VL is always on. Bypass with a ceramic capacitor with 1F/10mA of load current. The internal VL regulator can be disabled by connecting VL and V+ to an externally generated 5V. VL output current can be boosted with an external PNP transistor. 23 DL1 DL1 24 PGND PGND Low-Side Gate-Driver Output for Step-Down 1. Swings from PGND to VL. Power Ground for Gate-Driver Circuits Output Current-Limit Setting for Step-Down 1. Connect a resistor from ILIM1 to the drain of the step-down 1 high-side MOSFET, or to the junction of the source of the high-side MOSFET and the current-sense resistor to set the current-limit threshold. See the Current-Limit Setting section. 25 ILIM1 ILIM1 26 LX1 LX1 External Inductor Input for Step-Down 1. Connect to the switched side of the inductor. LX1 serves as the lower supply-voltage rail for the DH1 high-side gate driver and current-limit circuitry. 27 DH1 DH1 High-Side Gate-Driver Step-Down 1. Swings from LX1 to BST1. 28 BST1 BST1 Bootstrap Input to Power Internal High-Side Driver for Step-Down 1. Connect to an external capacitor and diode according to Figure 1. Detailed Description The MAX8537/MAX8539 controllers provide a complete power-management solution for both DDR and combiner supplies. The MAX8537 and MAX8539 are configured for out-of-phase and in-phase DDR power-supply operations, respectively. In addition to the dual-synchronous buck controllers, they also contain an additional amplifier to generate a total of three outputs: the main memory voltage (V DDQ ), the tracking sinking/sourcing termination voltage (VTT), and the termination reference voltage (V TTR). The MAX8538 is configured as a dual out-of-phase controller for pointof-load supplies. Each buck controller can source or sink up to 25A of current, while the termination reference can supply up to 15mA output. The MAX8537/MAX8539 have a 1% accurate reference. The first buck controller generates VDDQ using external resistor-dividers. The second synchronous buck controller and the amplifier generate 1/2 VDDQ voltage for VTT and VTTR. The VTT and VTTR voltages are maintained within 1% of 1/2 VDDQ. An internal high-bandwidth (25MHz) operational amplifier is used as an error amplifier to regulate the output voltage. This allows fast transient response, reducing the number of output capacitors. Synchronous rectification ensures high efficiency and balanced current sourcing and sinking capability for VTT. An all-N-FET design optimizes efficiency and cost. The two converters can be operated in-phase or out-of-phase to minimize capacitance and optimize performance for all VIN/VOUT combinations. Both channels have independent enable and powergood functions. They also have high-side current-sense architectures. ILIM pins allow the setting of an adjustable, lossless current limit for different combinations of load current and RDS(ON). Additionally, accurate overcurrent protection is achieved by using a sensing resistor in series with the high-side FET. The positive current-limit threshold is programmable through an external resistor. Overvoltage protection is achieved by latching off the high-side MOSFET and latching on the low-side MOSFET when the output voltage exceeds 17% of its set output. The MAX8537/MAX8538/MAX8539 use a constant-frequency voltage-mode architecture with operating frequencies of 200kHz to 1.4MHz to allow flexible design. 8 _______________________________________________________________________________________ Dual-Synchronous Buck Controllers for Point-ofLoad, Tracking, and DDR Memory Power Supplies On startup, the synchronous rectifier (low-side MOSFET) forces LX to ground and charges the boost capacitor to VL. On the second half-cycle, the switchmode power supply turns on the high-side MOSFET by closing an internal switch between BST and DH. This provides the necessary gate-to-source voltage to turn on the high-side switch, an action that boosts the 5V gate-drive signal above the input voltage. Internal 5V Linear Regulator All MAX8537/MAX8538/MAX8539 functions are powered from the on-chip low-dropout 5V regulator with the input connected to V+. Bypass the regulator's output (VL) with a 1F/10mA or greater ceramic capacitor. The V+ to VL dropout voltage is typically 500mV, so when V+ is less than 5.5V, VL is typically (V+ - 500mV). The internal linear regulator can source up to 70mA to supply the IC, power the low-side gate drivers, and charge the external boost capacitors. The current required to drive the external MOSFETs is calculated as the total gate charge of the MOSFETs at 5V multiplied by the switching frequency. At higher frequency, the MOSFET drive current may exceed the capability of the internal linear regulator. The output current at VL can be supplemented with an external PNP transistor as shown in Figures 4 and 5, which also moves most of the power dissipation off the IC. The external PNP can increase the output current at VL to over 200mA. The dropout voltage increases to 1V (typ). Undervoltage Lockout (UVLO) If VL drops below 3.75V, the MAX8537/MAX8538/ MAX8539 assume that the supply voltage is too low to make valid decisions, so UVLO circuitry inhibits switching and forces POK and DH low and DL high. After VL rises above 4.3V, the controller powers up the outputs (see the Startup section). Synchronous-Rectifier Driver (DL) Startup Synchronous rectification reduces the conduction loss in the rectifier by replacing the normal Schottky catch diode with a low-resistance MOSFET switch. The MAX8537/MAX8538/MAX8539 controllers also use the synchronous rectifier to ensure proper startup of the boost gate-drive circuit. Externally, the MAX8537/MAX8538/MAX8539 start switching when VL rises above the 4.3V UVLO threshold. However, the controller does not start unless all four of the following conditions are met: 1) EN_ is high, 2) VL > 4.3V, 3) the internal reference exceeds 80% of its nominal value (VREF > 0.64V), and 4) the thermal limit is not exceeded. Once the MAX8537/MAX8538/ MAX8539 assert the internal enable signal, the controller starts switching and enables soft-start. High-Side Gate-Drive Supply (BST) Gate-drive voltage for the high-side N-channel switch is generated by a flying-capacitor boost circuit (Figure 1). The capacitor between BST and LX is alternately charged from the VL supply and placed in parallel to the high-side MOSFET's gate-source terminals. _______________________________________________________________________________________ 9 MAX8537/MAX8538/MAX8539 DC-DC Controller The MAX8537/MAX8538/MAX8539 step-down DC-DC converters use a PWM voltage-mode control scheme. An internal high-bandwidth (25MHz) operational amplifier is used as an error amplifier to regulate the output voltage. The output voltage is sensed and compared with an internal 0.8V reference or REFIN to generate an error signal. The error signal is then compared with a fixed-frequency ramp by a PWM comparator to give the appropriate duty cycle to maintain output voltage regulation. At the rising edge of the internal clock, and with DL (the low-side MOSFET gate drive) at 0V, the high-side MOSFET turns on. When the ramp voltage reaches the error-amplifier output voltage, the high-side MOSFET latches off until the next clock pulse. During the highside MOSFET on-time, current flows from the input, through the inductor, and to the output capacitor and load. At the moment the high-side MOSFET turns off, the energy stored in the inductor during the on-time is released to support the load as the inductor current ramps down by commutation through the low-side MOSFET body diode. After a fixed delay, the low-side MOSFET turns on to shunt the current from its body diode for lower voltage drop and increased efficiency. The low-side MOSFET turns off at the rising edge of the next clock pulse, and when its gate voltage discharges to zero, the high-side MOSFET turns on and another cycle starts. The controllers sense peak inductor current and provide hiccup-mode overload and short-circuit protection (see the Current Limit section). The MAX8537/MAX8538/MAX8539 operate in forcedPWM mode where the inductor current is always continuous, so even under light load the controller maintains a constant switching frequency to minimize noise and possible interference with system circuitry. 10 EN1 EN2 POK2 POK1 VOUT2 1.25V/ 12A R19 100k C11, C30, C31, C32 680F N1 R5 100k R12 510 R10 10.0k C19 8.2nF R20 100k R1 0.005 L2 0.8uH C2 10F C27 1000F C26 1000F C1 1000F VL R11 51k R6 100k C41 47nF R24 3.3 C23 0.01F C18 15pF C17 3.9nF R9 51.1k C8 47pF Figure 1. Typical Application Circuit: MAX8537 DDR Memory Application (400kHz Switching) ______________________________________________________________________________________ C25 220pF R17 10.0k N5 R2 402 C5 0.47F POK1 DL2 ILIM2 LX2 DH2 BST2 COMP2 FREQ EN1 EN2 REFIN SS2 R18 10.0k 14 13 12 FB2 11 10 9 8 7 POK2 5 6 4 3 2 1 U1 VL MAX8537 D1 VTTR AVL V+ VL DL1 PGND ILIM1 LX1 DH1 BST1 16 17 18 19 20 21 22 23 24 25 26 27 28 GND 15 SS1 FB1 COMP1 D2 C24 0.01F C20 39pF C21 3.9nF R8 4.7 C42 0.15F R16 10.0k R14 22k R25 3.3 R3 402 C13 10F C15 1F VL C9 47pF C6 0.47F C4 10F C14 1F R15 21.5k N7 N3 C22 820pF C16 1F R4 0.003 N8 R13 1.2k N4 C29 1000F C28 1000F C3 1000F C12, C36 220F L1 0.9H VTTR R7 2.2 VOUT1 2.5V/20A VIN (10.8V TO 13.2V) MAX8537/MAX8538/MAX8539 Dual-Synchronous Buck Controllers for Point-ofLoad, Tracking, and DDR Memory Power Supplies EN1 EN2 POK2 POK1 VOUT2 0.9V/ 7A R19 100k C11, C30, C31, C32 680uF R2 750 C7 0.1F R5 100k R12 2.2k R10 10.0k C19 2.2nF R20 100k L2 0.5uH C39 1.0nF R21 N1 2.2 C2 10F R11 82k R6 100k VL C23 0.01F C18 10pF C17 1.8nF R9 51.1k C25 220pF R17 10.0k N5 R24 1 C5 0.47F POK1 DL2 ILIM2 LX2 COMP2 FREQ EN1 EN2 REFIN SS2 R18 10.0k 14 13 12 FB2 11 10 9 8 7 POK2 5 6 4 3 2 DH2 1 BST2 U1 VL MAX8539 D1 VTTR AVL V+ VL DL1 PGND ILIM1 LX1 DH1 BST1 16 17 18 19 20 21 22 23 24 25 26 27 28 GND 15 SS1 FB1 COMP1 D2 C24 0.01F C20 56pF C21 15nF R8 4.7 C13 10F C15 1F VL R25 1 C6 0.47F R16 10.0k R14 12k R3 1.0k C10 0.1F C4 10F C14 1F N7 R15 12.7k R13 2.2k N3 C12, C36, C37 680F N8 L1 0.8H N4 R22 1.5 C22 2.7nF C16 1F C29 1000F C28 1000F C3 1000F VTTR R7 2.2 C40 2.2nF VOUT1 1.8V/15A VOUT1 VIN (10.8V TO 13.2V) MAX8537/MAX8538/MAX8539 VOUT1 Dual-Synchronous Buck Controllers for Point-ofLoad, Tracking, and DDR Memory Power Supplies Figure 2. MAX8539 DDR Memory Application (400kHz Switching) ______________________________________________________________________________________ 11 EN1 EN2 POK2 POK1 R19 100k C11 470F VOUT2 2.5V/ 5A N1 C19 1.8nF R2 511 C7 0.1F R5 100k R23 10.0k R12 1.0k R10 21.5k L2 3.2H R20 100k C39 1.0nF R21 2.2 C2 10F C1 1000F R11 21.5k R6 100k VL N5 R24 1 Figure 3. MAX8538 PowerPCTM Application (400kHz Switching) PowerPC is a trademark of Motorola, Inc. 12 ______________________________________________________________________________________ POK1 DL2 ILIM2 LX2 DH2 COMP2 FREQ EN1 EN2 13 SS2 12 FB2 11 10 9 8 7 POK2 5 6 4 3 2 BST2 C23 0.01F 14 N.C. C18 10pF C17 6.8nF R9 51.1k C5 0.47F 1 U1 VL MAX8538 D1 GND AVL V+ VL DL1 PGND ILIM1 LX1 DH1 BST1 16 17 18 19 20 21 22 23 24 25 26 27 28 GND 15 SS1 FB1 COMP1 D2 C24 0.01F C20 47pF R16 10.0k C21 R14 0.010F 14k R8 4.7 C13 10F C15 1F VL R25 1 C6 0.47F R3 1.0k C10 0.1F C4 10F C14 1F R15 12.7k R13 2.2k N7 N3 C29 1000F C28 1000F C3 1000F C22 2.2nF N8 C12, C26 680F L1 1.0H N4 R22 1.5 R7 2.2 C40 2.2nF VOUT1 1.8V/15A VIN (10.8V TO 13.2V) MAX8537/MAX8538/MAX8539 Dual-Synchronous Buck Controllers for Point-ofLoad, Tracking, and DDR Memory Power Supplies ______________________________________________________________________________________ VOUT2 2.5V/ 10A EN1 EN2 POK2 POK1 C11 330F R19 100k C30 330F N1 C19 1nF C7 0.1F R5 100k R2 1.0k R23 10.0k R12 4.3k R10 21.5k L2 0.66H R20 100k C39 2.2nF R21 1.5 C2 10F C1 1000F 25V R11 21.5k R6 100k VL N5 R24 1 C23 0.01F C18 15pF C17 3.9nF R9 20.0k C5 0.47F POK1 DL2 ILIM2 LX2 DH2 BST2 COMP2 FREQ EN1 EN2 SS2 14 N.C. 13 12 FB2 11 10 9 8 7 POK2 5 6 4 3 2 1 U1 VL MAX8538 D1 21 22 23 24 25 26 27 28 GND 19 AVL 20 V+ VL DL1 PGND ILIM1 LX1 DH1 BST1 16 GND 15 SS1 FB1 17 COMP1 18 D2 C20 10pF C24 0.01F R3 1.21k R16 10.0k R8 4.7 C21 R14 2.7nF 33k C13 10F C15 1F VL R25 1 C6 0.47F C10 0.1F C4 10F C14 1F R15 31.6k R13 6.2k N7 N3 R22 1.5 C29 1000F 25V C28 1000F 25V C3 1000F 25V C22 680pF Q1 C12, C36 330F L1 0.66H C40 2.2nF R7 68 VIN (10.8V TO 13.2V) VOUT1 3.3V/12A MAX8537/MAX8538/MAX8539 C26 1000F 25V Dual-Synchronous Buck Controllers for Point-ofLoad, Tracking, and DDR Memory Power Supplies Figure 4. MAX8538 Dual-Output Application (1MHz Switching) 13 MAX8537/MAX8538/MAX8539 Dual-Synchronous Buck Controllers for Point-ofLoad, Tracking, and DDR Memory Power Supplies Power-Good Signal (POK_) The power-good signal (POK_) is an open-drain output. The MOSFET turns on and POK_ is held low until FB_ is 12% from its nominal threshold (0.8V for FB1 and VREFIN for FB2). Then there is a 64 clock-cycle delay before POK_ goes high impedance. For 400kHz switching frequency, this delay is 160s. To obtain a logic voltage output, connect a pullup resistor from POK_ to VL. A 100k resistor works well for most applications. If unused, leave POK_ grounded or unconnected. Enable (EN_), Soft-Start, and Soft-Stop Outputs of the MAX8537/MAX8538/MAX8539 can be turned on with logic high and off with logic low independently at EN1 and EN2. EN1 controls step-down 1, and EN2 controls step-down 2 and VTTR (MAX8537/ MAX8539 only). On the rising edge of EN_, the controller enters softstart. Soft-start gradually ramps up the reference voltage seen by the error amplifier to control the output's rate of rise and reduce the input surge current during startup. The soft-start period is determined by a 5A pullup current, the external soft-start capacitor connected from SS_ to ground, and the reference voltage (0.8V for FB1 and VREFIN for FB2, on the MAX8537/MAX8539; 0.8V for FB2 on the MAX8538). The output reaches regulation when soft-start is completed. On the falling edge of EN_, the controller enters soft-stop, which reverses the soft-start ramp. However, there is a delay due to 1V overcharge on the soft-start capacitor. The delay time can be calculated as tDELAY = CSS x 1V / 5A. At the end of soft-stop, DH is low and DL is high. Current Limit The MAX8537/MAX8538/MAX8539 DC-DC step-down controllers sense the peak inductor current either through the on-resistance of the high-side MOSFET for lossless sensing, or with a series resistor for more accurate sensing. In either case, when peak voltage across the sensing circuit (which occurs at the peak of the inductor current) exceeds the current-limit threshold set by the ILIM pin, the controller turns off the high-side MOSFET and turns on the low-side MOSFET. The MAX8537/MAX8538/MAX8539 current-limit threshold can be set by an external resistor that works in conjunction with an internal 200A current sink. See the Design Procedure section for how to set the ILIM with an external resistor. As the output load current increases above the threshold required to trip the peak current limit, the output voltage sags because the truncated duty cycle is insufficient to support the load current. When FB_ is 30% below its nominal threshold, output undervoltage pro14 tection is triggered and the controller enters hiccup mode to limit the power dissipation in a fault condition. See the Output Undervoltage Protection (UVP) section for a description of hiccup operation. Output Undervoltage Protection (UVP) Output UVP begins when the controller is at its current limit, FB_ is 30% below its nominal threshold, and softstart is complete. This condition causes the controller to drive DH and DL low, and to discharge the soft-start capacitor with a 5A pulldown current until VSS reaches 50mV. Then the controller begins switching and enables soft-start. If the overload condition still exists when softstart is complete, UVP triggers again. The result is hiccup mode, where the controller attempts to restart periodically as long as the overload condition exists. In hiccup mode, the soft-start capacitor voltage ramps from the nominal FB_ threshold + 12% down to 50mV. For the MAX8537/MAX8539, the tracking step-down must also have VREFIN > 0.45V to trigger UVP. Then the soft-start capacitor voltage ramps from VREFIN + 12% down to 50mV. Additionally, in the MAX8537/MAX8539 if output 1 is shorted, output 2 latches off. Recycle the input power or enable to restart output 2. Output Overvoltage Protection (OVP) The output voltages are continuously monitored for overvoltage. If the output voltage is more than 17% above the reference of the error amplifier, OVP is triggered after a 10s delay and the controller turns off. The DL low-side gate driver is latched high until EN_ is toggled or V+ power is cycled below 3.75V. This action turns on the synchronous-rectifier MOSFET with 100% duty cycle and, in turn, rapidly discharges the output filter capacitor and forces the output to ground. Note that DL latching high causes the output voltage to go slightly negative due to energy stored in the output LC at the instant OVP activates. If the load cannot tolerate being forced to a negative voltage, it can be desirable to place a power Schottky diode across the output to act as a reverse-polarity clamp. For step-down 2 of the MAX8537/MAX8539, the OVP threshold is 560mV for VREFIN 0.45V, and the OVP threshold is VREFIN + 17% for VREFIN > 0.45V. Thermal-Overload Protection Thermal-overload protection limits total power dissipation in the MAX8537/MAX8538/MAX8539. When the junction temperature exceeds TJ = +160C, a thermal sensor shuts down the device, forcing DH and DL low and allowing the IC to cool. The thermal sensor turns the part on again after the junction temperature cools ______________________________________________________________________________________ Dual-Synchronous Buck Controllers for Point-ofLoad, Tracking, and DDR Memory Power Supplies During a thermal event, the switching converters are turned off, POK1 and POK2 are pulled low, and the soft-starts are reset. Design Procedure Output Voltage Setting The output voltage can be set by a resistive divider network. Select R2, the resistor from FB to GND, between 5k and 15k. Then calculate R1 by: R1 = R2 x [(VOUT / 0.8) -1] Inductor Selection There are several parameters that must be examined when determining which inductor to use: input voltage, output voltage, load current, switching frequency, and LIR. LIR is the ratio of inductor current ripple to DC load current. A higher LIR value allows for a smaller inductor, but results in higher losses and higher output ripple. A good compromise between size and efficiency is a 30% LIR. Once all the parameters are chosen, the inductor value is determined as follows: L= VOUT x (VIN - VOUT ) VIN x fS x ILOAD(MAX) x LIR where fS is the switching frequency. Choose a standard value close to the calculated value. The exact inductor value is not critical and can be adjusted in order to make trade-offs among size, cost, and efficiency. Lower inductor values minimize size and cost, but also increase the output ripple and reduce the efficiency due to higher peak currents. On the other hand, higher inductor values increase efficiency, but eventually resistive losses due to extra turns of wire exceed the benefit gained from lower AC current levels. Find a lowloss inductor with the lowest possible DC resistance that fits the allotted dimensions. Ferrite cores are often the best choice, although powdered iron is inexpensive and can work well up to 300kHz. The chosen inductor's saturation current rating must exceed the peak inductor current determined as: LIR IPEAK = ILOAD(MAX) + x ILOAD(MAX) 2 Input Capacitor The input filter capacitor reduces peak currents drawn from the power source and reduces noise and voltage ripple on the input caused by the circuit's switching. The input capacitor must meet the ripple current requirement (IRMS) imposed by the switching currents defined by the following equation: IRMS = [IOUT12 x VOUT1 x (VIN - VOUT1)] + [IOUT2 2 x VOUT2 x (VIN - VOUT2 )] VIN Combinations of large electrolytic and small ceramic capacitors in parallel are recommended. Almost all of the RMS current is supplied from the large electrolytic capacitor, while the smaller ceramic capacitor supplies the fast rise and fall switching edges. Choose the electrolytic capacitor that exhibits less than 10C temperature rise at the maximum operating RMS current for optimum long-term reliability. Output Capacitor The key selection parameters for the output capacitor are the actual capacitance value, the equivalent series resistance (ESR), the equivalent series inductance (ESL), and the voltage-rating requirements, which affect the overall stability, output ripple voltage, and transient response. The output ripple has three components: variations in the charge stored in the output capacitor, voltage drop across the capacitor's ESR, and voltage drop across the capacitor's ESL, caused by the current into and out of the capacitor. The following equations estimate the worst-case ripple: VRIPPLE = VRIPPLE(ESR) + VRIPPLE(C) + VRIPPLE(ESL) VRIPPLE(ESR) = IP-P ESR VRIPPLE(C) = IP-P / (8 x COUT x fSW ) VRIPPLE(ESL) = VIN x ESL / (L + ESL) V -V V IP-P = IN OUT OUT fSW L VIN where IP-P is the peak-to-peak inductor current (see the Inductor Selection section). Higher output current requires paralleling multiple capacitors to meet the output ripple voltage. The MAX8537/MAX8538/MAX8539s' response to a load transient depends on the selected output capacitor. After a load transient, the output instantly changes by (ESR x ILOAD) + (ESL x dI/dt). Before the controller can respond, the output deviates further depending on the inductor and output capacitor values. After a short period of time (see the Typical Operating Characteristics), the controller responds by regulating the output voltage back to its nominal state. The controller response time depends on the closed-loop bandwidth. With higher bandwidth, the response time is faster, pre- ______________________________________________________________________________________ 15 MAX8537/MAX8538/MAX8539 by 10C, resulting in a pulsed output during continuous thermal-overload conditions. MAX8537/MAX8538/MAX8539 Dual-Synchronous Buck Controllers for Point-ofLoad, Tracking, and DDR Memory Power Supplies venting the output capacitor voltage from further deviation from its regulating value. Do not exceed the capacitor's voltage or ripple current ratings. MOSFET Selection The MAX8537/MAX8538/MAX8539 controllers drive two external, logic-level, N-channel MOSFETs as the circuitswitch elements. The key selection parameters are: 1) On-resistance (RDS(ON)): the lower, the better. 2) Maximum drain-to-source voltage (VDSS): should be at least 20% higher than the input supply rail at the high-side MOSFET's drain. 3) Gate charges (Qg, Qgd, Qgs): the lower, the better. Choose MOSFETs with RDS(ON) rated at VGS = 4.5V. For a good compromise between efficiency and cost, choose the high-side MOSFET that has conduction loss equal to the switching loss at the nominal input voltage and maximum output current. For the low-side MOSFET, make sure it does not spuriously turn on due to dV/dt caused by the high-side MOSFET turning on, as this results in shoot-through current degrading the efficiency. MOSFETs with a lower Qgd/Qgs ratio have higher immunity to dV/dt. For proper thermal-management design, the power dissipation must be calculated at the desired maximum operating junction temperature, maximum output current, and worst-case input voltage (for low-side MOSFET, worst case is at V IN(MAX) ; for high-side MOSFET, it could be either at VIN(MIN) or VIN(MAX)). High-side and low-side MOSFETs have different loss components due to the circuit operation. The low-side MOSFET, operates as a zero-voltage switch; therefore, the major losses are the channel conduction loss (PLSCC) and the body-diode conduction loss (PLSDC): PLSCC = [1 - (VOUT / VIN)] x (ILOAD)2 x RDS,ON Use RDS,ON at TJ(MAX): PLSDC = 2 x ILOAD x VF x tdt x fS where VF is the body-diode forward voltage drop, tdt is the dead-time between the high-side MOSFET and the low-side MOSFET switching transitions, and fS is the switching frequency. The high-side MOSFET operates as a duty-cycle control switch and has the following major losses: the channel conduction loss (PHSCC), the V I overlapping switching loss (PHSSW), and the drive loss (PHSDR). The high-side MOSFET does not have body-diode conduction loss because the diode never conducts current. 16 PHSCC = (VOUT / VIN) x I2LOAD x RDS(ON) Use RDS(ON) at TJ(MAX): PHSSW = VIN x ILOAD x fS x [(Qgs + Qgd) / IGATE] where I GATE is the average DH-high driver outputcurrent capability determined by: IGATE(ON) = 2.5 / (RDH + RGATE) where RDH is the high-side MOSFET driver's average on-resistance (1.1 typ) and RGATE is the internal gate resistance of the MOSFET (~2): PHSDR = Qgs x VGS x fS x RGATE / (RGATE + RDH) where VGS ~ VL = 5V. In addition to the losses above, approximately 20% more for additional losses due to MOSFET output capacitances and low-side MOSFET body-diode reverse-recovery charge dissipated in the high-side MOSFET that exists, but is not well defined in the MOSFET data sheet. Refer to the MOSFET data sheet for thermal-resistance specification to calculate the PC board area needed to maintain the desired maximum operating junction temperature with the above-calculated power dissipation. To reduce EMI caused by switching noise, add a 0.1F ceramic capacitor from the high-side switch drain to the low-side switch source or add resistors in series with DH and DL to slow down the switching transitions. However, adding series resistors increases the power dissipation of the MOSFETs, so be sure this does not overheat the MOSFETs. The minimum load current must exceed the high-side MOSFET's maximum leakage current over temperature if fault conditions are expected. Current-Limit Setting The MAX8537/MAX8538/MAX8539 controllers sense the peak inductor current to provide constant current and hiccup current limit. The peak current-limit threshold is set by an external resistor together with the internal current sink of 200A. The voltage drop across the resistor RILIM_with 200A current through it sets the maximum peak inductor current that can flow through the high-side MOSFET or the optional current-sense resistor by the equations below: IPEAK(MAX) = 200A x RILIM_ / RDSON(HSFET) or IPEAK(MAX) = 200A x RILIM_ / RSENSE RILIM_ should be less than 1.5k for optimum currentlimit accuracy. The actual corresponding maximum load current is lower than the IPEAK(MAX) above by half ______________________________________________________________________________________ Dual-Synchronous Buck Controllers for Point-ofLoad, Tracking, and DDR Memory Power Supplies The circuit parasitic capacitance (CPAR) at LX is then equal to 1/3rd the value of the added capacitance above. The circuit parasitic inductance (LPAR) is calculated by: LPAR = 1 (2fR ) 2 x CPAR The resistor for critical dampening (RSNUB) is equal to 2 x fR x LPAR. Adjust the resistor value up or down to tailor the desired damping and the peak voltage excursion. The capacitor (CSNUB) should be at least 2 to 4 times the value of the C PAR in order to be effective. The power loss of the snubber circuit is dissipated in the resistor (PRSNUB) and can be calculated as: PRSNUB = CSNUB x ( VIN ) 2 x fSW where VIN is the input voltage and fSW is the switching frequency. Choose an RSNUB power rating that meets the specific application's derating rule for the power dissipation calculated. Additionally, there is parasitic inductance of the current-sensing element, whether the high-side MOSFET R DS(ON) (L SENSE_FET ) or the actual current-sense resistor RSENSE (LRSENSE) are used, which is in series with the output filter inductor. This parasitic inductance, together with the output inductor, form an inductive divider and cause error in the current-sensing voltage. To compensate for this error, a series RC circuit can be added in parallel with the sensing element (see Figure 1). The RC time constant should equal L RSENSE / RSENSE, or LSENSE_FET / RDS(ON). First, set the value of R equal to or less than RILIM_ / 100. Then, the value of C can be calculated as: C = LRSENSE / (RSENSE x R) or C = LSENSE_FET / (RDS(ON) x R) Any PC board trace inductance in series with the sensing element and output inductor should be added to the specified FET or resistor inductance per the respective manufacturer's data sheet. For the case of the MOSFET, it is the inductance from the drain to the source lead. An additional switching noise filter may be needed at ILIM_ by connecting a capacitor in parallel with RILIM_ (in the case of RDS(ON) sensing) or from ILIM_ to LX (in the case of resistor sensing). For the case of RDS(ON) sensing, the value of the capacitor should be: C > 50 / (3.1412 x fS x RILIM_) For the case of resistor sensing: C < 25 x 10-9 / RILIM_ Soft-Start Capacitor Setting The two step-down converters have independent, adjustable soft-start. External capacitors from SS1/SS2 to ground are charged by an internal 5A current source to the corresponding feedback threshold. Therefore, the soft-start time can be calculated as: TSS = CSS x VFB / 5A For example, 0.01F from SS1 to ground corresponds to approximately a 1.6ms soft-start period for stepdown 1. Compensation Design The MAX8537/MAX8538/MAX8539 use a voltage-mode control scheme that regulates the output voltage by comparing the error-amplifier output (COMP) with a fixed internal ramp to produce the required duty cycle. The error amplifier is an operational amplifier with 25MHz bandwidth to provide fast response. The output lowpass LC filter creates a double pole at the resonant frequency that introduces a gain drop of 40dB per decade and a phase shift of 180 degrees per decade. The error amplifier must compensate for this gain drop and phase shift to achieve a stable high-bandwidth closed-loop system. The basic regulator loop can be thought of as consisting of a power modulator and an error amplifier. The power modulator has DC gain set by VIN / VRAMP, with a double pole, fP_LC, and a single zero, fZ_ESR, set by the output inductor (L), the output capacitor (CO), and its equivalent series resistance (RESR). Below are the equations that define the power modulator: ______________________________________________________________________________________ 17 MAX8537/MAX8538/MAX8539 of the inductor ripple current (see the Inductor Selection section). If RDS(ON) of the high-side MOSFET is used for current sensing, make sure to use the maximum RDS(ON) at the highest operating junction temperature to avoid fault tripping of the current limit at elevated temperature. Consideration should also be given to the tolerance of the 200A current sink. When RDS(ON) of the high-side MOSFET is used for current sensing, ringing on the LX voltage waveform can interfere with the current limit. Below is the procedure for selecting the value of the series RC snubber circuit: 1) Connect a scope probe to measure VLX to GND, and observe the ringing frequency, fR. 2) Find the capacitor value (connected from LX to GND) that reduces the ringing frequency by half. MAX8537/MAX8538/MAX8539 Dual-Synchronous Buck Controllers for Point-ofLoad, Tracking, and DDR Memory Power Supplies GMOD(DC) = VIN VRAMP , where VRAMP = 1V (typ) 1 2 L C O 1 fZ _ ESR = 2 x RESR x CO fP _ LC = When the output capacitor is composed of paralleling n number of the same capacitors, then: CO = n x CEACH and RESR = RESR _ EACH n Thus, the resulting fZ_ESR is the same as that of a single capacitor. The total closed-loop gain must be equal to unity at the crossover frequency, where the crossover frequency is less than or equal to 1/5th the switching frequency (fS): fC fS / 5 So the loop-gain equation at the crossover frequency is: GEA(FC) x GMOD(FC) = 1 where GEA(FC) is the error-amplifier gain at f C, and GMOD(FC) is the power-modulator gain at fC. The loop compensation is affected by the choice of output filter capacitor due to the position of its ESR-zero frequency with respect to the desired closed-loop crossover frequency. Ceramic capacitors are used for higher switching frequencies (above 750kHz) and have low capacitance and low ESR; therefore, the ESR-zero frequency is higher than the closed-loop crossover frequency. Electrolytic capacitors (e.g., tantalum, solid polymer, and OS-CON) are needed for lower switching frequencies and have high capacitance and higher ESR; therefore, the ESR-zero frequency is lower than the closed-loop crossover frequency. Thus, the compensation design procedures are separated into two cases: Case 1: Crossover frequency is less than the outputcapacitor ESR-zero (fC < fZ_ESR). The modulator gain at fC is: GMOD(FC) = GMOD(DC) x (fP_LC / fC)2 Since the crossover frequency is lower than the output capacitor ESR-zero frequency and higher than the LC double-pole frequency, the error-amplifier gain must have a +1 slope at fC so that, together with the -2 slope of the LC double pole, the loop crosses over at the desired -1 slope. 18 The error amplifier has a dominant pole at a very low frequency (~0Hz), and two additional zeros and two additional poles as indicated by the equations below and illustrated in Figure 6: fZ1_EA = 1 / (2 x R4 x C2) fZ2_EA = 1 / (2 x (R1 + R3) x C1) fP2_EA = 1 / (2 x R3 x C1) fP3_EA = 1 / (2 x R4 x (C2 x C3 / (C2 + C3))) Note that fZ2_EA and fP2_EA are chosen to have the converter closed-loop crossover frequency, fC, occur when the error-amplifier gain has +1 slope, between fZ2_EA and fP2_EA. The error-amplifier gain at fC must meet the requirement below: GEA(FC) = 1 / GMOD(FC) The gain of the error amplifier between f Z1_EA and fZ2_EA is: GEA(fZ1_EA - fZ2_EA) = GEA(FC) x fZ2_EA / fC = fZ2_EA / (fC x GMOD(FC)) This gain is set by the ratio of R4/R1, where R1 is calculated in the Output Voltage Setting section. Thus: R4 = R1 x fZ2_EA / (fC x GMOD(FC)) where fZ2_EA = fP_LC. Due to the underdamped (Q > 1) nature of the output LC double pole, the first error-amplifier zero frequency must be set less than the LC double-pole frequency in order to provide adequate phase boost. Set the erroramplifier first zero, fZ1_EA, at 1/4th the LC double-pole frequency. Hence: C2 = 2 / ( x R4 x fP_LC) Set the error amplifier fP2_EA at fZ_ESR and fP3_EA equal to half the switching frequency. The error-amplifier gain between fP2_EA and fP3_EA is set by the ratio of R4/RI and is equal to: GEA(fZ1_EA - fZ2_EA) x (fZ_ESR / fP_LC) where RI = R1 x R3 / (R1 + R3). Then: RI = R4 x fP_LC / (GEA(fZ1_EA - fZ2_EA) x fZ_ESR) = R4 x fC x GMOD(FC) / fZ_ESR The value of R3 can then be calculated as: R3 = R1 x RI / (R1 - RI) Now we can calculate the value of C1 as: C1 = 1 / (2 x R3 x fZ_ESR) and C3 as: C3 = C2 / ((2 x C2 x R4 x fP3_EA) - 1) ______________________________________________________________________________________ Dual-Synchronous Buck Controllers for Point-ofLoad, Tracking, and DDR Memory Power Supplies MAX8537/MAX8538/MAX8539 C3 C2 R4 C1 VOUT R3 R1 FB EA R2 GAIN (dB) COMP REF CLOSED-LOOP GAIN EA GAIN 0 fz1 fZ2 fP2 FREQUENCY fP3 fC Figure 6. Error-Amplifier Compensation Circuit; Closed-Loop and Error-Amplifier Gain Plot for Case 1 CLOSED-LOOP GAIN GAIN (dB) EA GAIN 0 fZ1 fZ2 fP3 fP2 FREQUENCY fC Figure 7. Closed-Loop and Error-Amplifier Gain Plot for Case 2 ______________________________________________________________________________________ 19 Dual-Synchronous Buck Controllers for Point-ofLoad, Tracking, and DDR Memory Power Supplies MAX8537/MAX8538/MAX8539 MAX8537/MAX8539 Functional Diagram ILIM1 IMAX SENSE 200A BST1 FREQ DH1 OSC CONTROL LOGIC 1V LX1 VL PWM DL1 PGND EN1 REF V+ 0.80V VL SOFT-START VL EAMP COMP1 0.936V SS1 OVP1 0.560V UVP1 FB1 SOFT-START REFIN REFIN EAMP 0.896V 0.704V POK1 SS2 VTTR ILIM2 IMAX SENSE AVL 200A BIAS BST2 GND DH2 CONTROL LOGIC LX2 VL PWM DL2 EN2 PGND REFIN COMP2 1.12REFIN 0.88REFIN POK2 EAMP FB2 UVP2 OVP2 0.7REFIN 20 1.17REFIN ______________________________________________________________________________________ Dual-Synchronous Buck Controllers for Point-ofLoad, Tracking, and DDR Memory Power Supplies TOP VIEW BST2 1 28 BST1 BST2 1 28 BST1 DH2 2 27 DH1 DH2 2 27 DH1 LX2 3 26 LX1 LX2 3 26 LX1 ILIM2 4 25 ILIM1 ILIM2 4 POK1 5 24 PGND POK1 5 MAX8537 MAX8539 25 ILIM1 24 PGND MAX8538 23 DL1 DL2 6 POK2 7 22 VL POK2 7 22 VL EN2 8 21 V+ EN2 8 21 V+ EN1 9 20 AVL EN1 9 20 AVL 19 VTTR FREQ 10 19 GND DL2 6 FREQ 10 COMP2 11 18 COMP1 COMP2 11 23 DL1 18 COMP1 FB2 12 17 FB1 FB2 12 17 FB1 SS2 13 16 SS1 SS2 13 16 SS1 REFIN 14 15 GND N.C. 14 15 GND QSOP Case 2: Crossover frequency is greater than the output-capacitor ESR zero (fC > fZ_ESR). The modulator gain at fC is: GMOD(FC) = GMOD(DC) x (fP_LC)2 / (fZ_ESR x fC) Since the output-capacitor ESR-zero frequency is higher than the LC double-pole frequency but lower than the closed-loop crossover frequency, where the modulator already has -1 slope, the error-amplifier gain must have zero slope at fC so the loop crosses over at the desired -1 slope. The error-amplifier circuit configuration is the same as case 1 above; however, the closed-loop crossover frequency is now between fP2 and fP3 as illustrated in Figure 7. The equations that define the error amplifier's zeros (fZ1_EA, fZ2_EA) and poles (fP2_EA, fP3_EA) are the same as case 1; however, f P2_EA is now lower than the closed-loop crossover frequency. Therefore, the erroramplifier gain between fZ1_EA and fZ2_EA is now calculated as: GEA(fZ1_EA - fZ2_EA) = GEA(FC) x fZ2_EA / fP2_EA = fZ2_EA / (fP2_EA x GMOD(FC)) This gain is set by the ratio of R4/R1, where R1 is calculated in the Output Voltage Setting section. Thus: QSOP R4 = R1 x fZ2_EA / (fP2_EA x GMOD(FC)) where fZ2_EA = fP_LC and fP2_EA = fZ_ESR. Similar to case 1, C2 can be calculated as: C2 = 2 / ( x R4 x fP_LC) Set the error-amplifier third pole, f P3_EA, at half the switching frequency, and let RI = (R1 x R3) / (R1 + R3). The gain of the error amplifier between f P2_EA and f P3_EA is set by the ratio of R4/R I and is equal to GEA(FC) = 1 / GMOD(FC). Then: RI = R4 x GMOD(FC) Similar to case 1, R3, C1, and C3 can be calculated as: R3 = R1 x Ri / (R1 - RI) C1 = 1 / (2 x R3 x fZ_ESR) C3 = C2 / ((2 x C2 x R4 x fP3_EA) - 1) Applications Information PC Board Layout Guidelines Careful PC board layout is critical to achieve low switching losses and clean, stable operation. The switching-power stage requires particular attention. Follow these guidelines for good PC board layout: 1) Place the decoupling capacitors as close to the IC pins as possible. ______________________________________________________________________________________ 21 MAX8537/MAX8538/MAX8539 Pin Configurations MAX8537/MAX8538/MAX8539 Dual-Synchronous Buck Controllers for Point-ofLoad, Tracking, and DDR Memory Power Supplies 2) Keep separate the power ground plane (connected to the sources of the low-side MOSFETs, pin 24, input capacitor ground, output capacitor ground, and VL decoupling capacitor ground) and the signal ground plane (connected to GND pin and the rest of the circuit ground returns). Place the input decoupling ceramic capacitor as directly and close to the high-side MOSFET drain and the low-side MOSFET source as possible. Place the RC snubber circuit as close to the low-side MOSFET as possible. 3) Keep the high-current paths as short as possible. 4) Connect the drains of the MOSFETs to a large land area to help cool the devices and further improve efficiency and long-term reliability. 5) Ensure all feedback connections are short and direct. Place the feedback resistors as close to the IC as possible. 6) Route high-speed switching nodes away from sensitive analog areas (FB, COMP). 7) Refer to the evaluation kit for a sample board layout. Chip Information TRANSISTOR COUNT: 5504 PROCESS: BiCMOS 22 ______________________________________________________________________________________ Dual-Synchronous Buck Controllers for Point-ofLoad, Tracking, and DDR Memory Power Supplies QSOP.EPS PACKAGE OUTLINE, QSOP .150", .025" LEAD PITCH 21-0055 E 1 1 Maxim cannot assume responsibility for use of any circuitry other than circuitry entirely embodied in a Maxim product. No circuit patent licenses are implied. Maxim reserves the right to change the circuitry and specifications without notice at any time. Maxim Integrated Products, 120 San Gabriel Drive, Sunnyvale, CA 94086 408-737-7600 ____________________ 23 (c) 2005 Maxim Integrated Products Printed USA is a registered trademark of Maxim Integrated Products, Inc. MAX8537/MAX8538/MAX8539 Package Information (The package drawing(s) in this data sheet may not reflect the most current specifications. For the latest package outline information, go to www.maxim-ic.com/packages.)