19-3033; Rev 1; 4/04 VRM 9.0/VRM 9.1, Dual-Phase, Parallelable, Average-Current-Mode Controller The MAX5037A dual-phase, PWM controller provides high-output-current capability in a compact package with a minimum number of external components. The MAX5037A utilizes a dual-phase, average-current-mode control that enables optimal use of low R DS(ON) MOSFETs, eliminating the need for external heatsinks even when delivering high output currents. Differential sensing enables accurate control of the output voltage, while adaptive voltage positioning provides optimum transient response. An internal regulator enables operation with either +5V or +12V input voltage without the need for additional voltage sources. The high switching frequency, up to 500kHz per phase, and dual-phase operation allow the use of low output inductor values and input capacitor values. This accommodates the use of PC board-embedded planar magnetics achieving superior reliability, current sharing, thermal management, compact size, and low system cost. The MAX5037A also features a clock input (CLKIN) for synchronization to an external clock, and a clock output (CLKOUT) with programmable phase delay (relative to CLKIN) for paralleling multiple phases. The MAX5037A also limits the reverse current in case the bus voltage becomes higher than the regulated output voltage. The MAX5037A operates over the extended temperature range (-40°C to +85°C) and is available in 44-pin MQFP or thin QFN packages. Refer to the MAX5038A/ MAX5041A and MAX5065/MAX5067 data sheets for either a fixed output voltage controller or an adjustable output voltage controller in an SSOP or thin QFN package. Applications Servers and Workstations Point-of-Load High-Current/High-Density Telecom DC-DC Regulators Features ♦ +4.75V to +5.5V or +8V to +28V Input Voltage Range ♦ Up to 60A Output Current ♦ Internal Voltage Regulator for a +12V or +24V Power Bus ♦ Internal 5-Bit DAC VID Control (VRM 9.0/VRM 9.1 Compliant, 0.8% Accuracy) ♦ Programmable Adaptive Output Voltage Positioning ♦ True Differential Remote Output Sensing ♦ Out-of-Phase Controllers Reduce Input Capacitance Requirement and Distribute Power Dissipation ♦ Average-Current-Mode Control Superior Current Sharing Between Individual Phases and Paralleled Modules Accurate Current Limit Eliminates MOSFET and Inductor Derating ♦ Limits Reverse-Current Sinking in Paralleled Modules ♦ Integrated High-Output-Current Gate Drivers ♦ Selectable Fixed Frequency 250kHz or 500kHz per Phase (Up to 1MHz for Two Phases) ♦ External Frequency Synchronization from 125kHz to 600kHz ♦ Internal PLL with Clock Output for Paralleling Multiple DC-DC Converters ♦ Power-Good Output ♦ Phase Failure Detector Networking Systems ♦ Overvoltage and Thermal Protection Large-Memory Arrays ♦ 44-Pin MQFP or Thin QFN Packages RAID Systems Ordering Information High-End Desktop Computers TEMP RANGE PIN-PACKAGE MAX5037AEMH PART -40°C to +85°C 44 MQFP MAX5037AETH -40°C to +85°C 44 Thin QFN Pin Configuration appears 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 MAX5037A General Description MAX5037A VRM 9.0/VRM 9.1, Dual-Phase, Parallelable, Average-Current-Mode Controller ABSOLUTE MAXIMUM RATINGS Continuous Power Dissipation (TA = +70°C) 44-Pin MQFP (derate 12.7mW/°C above +70°C).......1013mW 44-Pin Thin QFN (derate 27.0mW/°C above +70°C) ...........................................................2162.2mW Package Thermal Resistance, θJC (Thin QFN only) ........+2°C/W Operating Temperature Range ...........................-40°C to +85°C Maximum Junction Temperature .....................................+150°C Storage Temperature Range .............................-60°C to +150°C Lead Temperature (soldering, 10s) .................................+300°C IN to SGND.............................................................-0.3V to +30V BST_ to SGND…………………………………….… .-0.3V to +35V DH_ to LX_ .................................-0.3V to [(VBST_ - VLX_) + 0.3V] DL_ to PGND ..............................................-0.3V to (VDD + 0.3V) BST_ to LX_ ..............................................................-0.3V to +6V VCC to SGND............................................................-0.3V to +6V VDD to PGND............................................................-0.3V to +6V SGND to PGND .....................................................-0.3V to +0.3V All Other Pins to SGND...............................-0.3V to (VCC + 0.3V) 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 (VCC = VDD = +5V, circuit of Figure 1, TA = -40°C to +85°C, unless otherwise noted. Typical specifications are at TA = +25°C.) (Note 1) PARAMETER SYMBOL CONDITIONS MIN TYP MAX UNITS SYSTEM SPECIFICATIONS 8 28 4.75 5.5 Input Voltage Range VIN Quiescent Supply Current IQ EN = VCC or SGND, VID inputs unconnected 4 Efficiency η ILOAD = 52A (26A per phase) 90 Short IN and VCC together for 5V input operation 6 V mA % STARTUP/INTERNAL REGULATOR VCC Undervoltage Lockout UVLO VCC rising 4.0 VCC Undervoltage Lockout Hysteresis 4.15 4.5 200 VCC Output Accuracy VIN = 8V to 28V, ISOURCE = 0 to 80mA 4.85 5.1 V mV 5.30 V VOUT/ADAPTIVE VOLTAGE POSITIONING (AVP) RREG = RF = 100kΩ, RIN = 1kΩ, no load, Figure 3 Nominal Output Voltage Accuracy (VID Setting) VIN = VCC = 4.75V to 5.5V, or VIN = 8V to 28V, RREG = RF = 100kΩ, RIN = 1kΩ, no load, Figure 3 Maximum REG Loading IREG_MAX REG Accuracy (Voltage Positioning) d (∆VOUT) Maximum CNTR Loading ICNTR_MAX Center Voltage Set-Point Accuracy (Note 2) d (∆VCNTR) -0.8 +0.8 % -1 +1 50 µA TA = 0°C to +85°C -3 +3 TA = -40°C to +85°C -5 +5 50 % µA TA = 0°C to +85°C -3 +3 TA = -40°C to +85°C -5 +5 % MOSFET DRIVERS Output Driver Impedance Output Driver Peak Source/Sink Current 2 RON IDH_, IDL_ Low or high output 1 4 _______________________________________________________________________________________ 3 Ω A VRM 9.0/VRM 9.1, Dual-Phase, Parallelable, Average-Current-Mode Controller (VCC = VDD = +5V, circuit of Figure 1, TA = -40°C to +85°C, unless otherwise noted. Typical specifications are at TA = +25°C.) (Note 1) PARAMETER Nonoverlap Time SYMBOL tNO CONDITIONS MIN CDH_/DL_ = 5nF TYP MAX 60 UNITS ns OSCILLATOR AND PLL Switching Frequency fSW PLL Lock Range fPLL PLL Locking Time tPLL CLKOUT Phase Shift (at fSW = 125kHz) φCLKOUT CLKIN = SGND 238 250 262 CLKIN = VCC 475 500 525 125 600 200 kHz µs PHASE = VCC 115 120 125 PHASE = unconnected 85 90 95 PHASE = SGND 55 60 65 ICLKIN 3 5 7 CLKIN High Threshold VCLKINH 2.4 CLKIN Low Threshold VCLKINL CLKIN Input Pulldown Current kHz Degrees µA V 0.8 V CLKIN High Pulse Width tCLKIN 200 ns PHASE High Threshold VPHASEH 4 V PHASE Low Threshold VPHASEL PHASE Input Bias Current IPHASEBIAS -50 CLKOUT Output Low Level VCLKOUTL ISINK = 2mA (Note 3) CLKOUT Output High Level VCLKOUTH ISOURCE = 2mA (Note 3) 4.5 VCL CSP_ to CSN_ 45 Reverse Current-Limit Threshold VCLR CSP_ to CSN_ -3.9 Cycle-by-Cycle Current Limit VCLPK CSP_ to CSN_ (Note 3) 1 V +50 µA 100 mV V CURRENT LIMIT Average Current-Limit Threshold Cycle-by-Cycle Overload Response Time tR 90 VCSP_ to VCSN_ = +150mV 48 112 51 mV -0.2 mV 130 mV 260 ns CURRENT-SENSE AMPLIFIER CSP_ to CSN_ Input Resistance Common-Mode Range Input Offset Voltage RCS_ 4 kΩ VCMR(CS) -0.3 +3.6 V VOS(CS) -1 +1 mV Amplifier Gain AV(CS) 18 V/V 3dB Bandwidth f3dB 4 MHz CURRENT-ERROR AMPLIFIER (TRANSCONDUCTANCE AMPLIFIER) Transconductance gmca Open-Loop Gain AVOL(CE) No load 550 µS 50 dB DIFFERENTIAL VOLTAGE AMPLIFIER (DIFF) Common-Mode Voltage Range VCMR(DIFF) DIFF Output Voltage VCM Input Offset Voltage VOS(DIFF) Amplifier Gain AV(DIFF) 3dB Bandwidth f3dB Minimum Output Current Drive IOUT(DIFF) -0.3 VSENSE+ = VSENSE- = 0 +1.0 V +2 mV 0.6 -2 0.997 CDIFF = 20pF 1 3 1.0 V 1.003 V/V MHz mA _______________________________________________________________________________________ 3 MAX5037A ELECTRICAL CHARACTERISTICS (continued) MAX5037A VRM 9.0/VRM 9.1, Dual-Phase, Parallelable, Average-Current-Mode Controller ELECTRICAL CHARACTERISTICS (continued) (VCC = VDD = +5V, circuit of Figure 1, TA = -40°C to +85°C, unless otherwise noted. Typical specifications are at TA = +25°C.) (Note 1) PARAMETER SENSE+ to SENSE- Input Resistance SYMBOL CONDITIONS RVS_ MIN TYP MAX UNITS 50 100 kΩ 70 dB VOLTAGE-ERROR AMPLIFIER (EAOUT) Open-Loop Gain AVOL(EA) Unity-Gain Bandwidth fUGEA EAN Input Bias Current IB(EA) Error-Amplifier Output Clamping Voltage 3 CNTR and REG = open, VEAN = 2.0V VCLAMP(EA) With respect to VCM MHz -100 +100 nA 810 918 mV POWER-GOOD, PHASE FAILURE DETECTION, OVERVOLTAGE PROTECTION, AND THERMAL SHUTDOWN PGOOD Trip Level PGOOD Output Low Level VOV VUV VPGLO PGOOD goes low when VOUT is outside of this window IPG PGOOD = VCC Phase Failure Trip Threshold VPH PGOOD goes low when CLP_ is higher than VPH OVPTH OVPOUT Source/Sink Current IOVPOUT OVPIN Input Resistance ROVPIN Thermal Shutdown TSHDN +8 +10 % VO -12.5 -10 -8.5 (VID) ISINK = 4mA PGOOD Output Leakage Current OVPIN Trip Threshold +6 Above VID programmed output voltage 0.20 V 1 µA 2.0 +10 +13 15 20 V +16 % VO (VID) VOVPOUT = 2.5V 190 Thermal-Shutdown Hysteresis 280 mA 370 kΩ 150 °C 8 °C LOGIC INPUTS FOR VID Logic-Input Pullup Resistors Logic-Input Low Voltage RVID 8 12 VIL Logic-Input High Voltage VIH VID Internal Pullup Voltage VVID 20 kΩ 0.8 V 1.7 All VID_ inputs unconnected 2.8 V 2.9 3.2 V EN INPUT EN Input Low Voltage VENL EN Input High Voltage VENH 3 IEN 4.5 EN Pullup Current 1 5 5.5 Note 1: Specifications from -40°C to 0°C are guaranteed by characterization but not production tested. Note 2: CNTR voltage accuracy is defined as the center of the adaptive voltage-positioning window (see Adaptive Voltage Positioning section). Note 3: Guaranteed by design. Not production tested. Note 4: See Peak-Current Comparator section. 4 V V _______________________________________________________________________________________ µA VRM 9.0/VRM 9.1, Dual-Phase, Parallelable, Average-Current-Mode Controller f = 500kHz 60 VIN = +5V VOUT = +1.8V 30 20 VOUT = +1.8V fSW = 250kHz VIN = +24V VOUT = +1.8V fSW = 125kHz 10 0 IOUT (A) IOUT (A) EFFICIENCY vs. OUTPUT CURRENT AND OUTPUT VOLTAGE EFFICIENCY vs. OUTPUT CURRENT AND OUTPUT VOLTAGE SUPPLY CURRENT vs. FREQUENCY AND INPUT VOLTAGE 80 VOUT = +1.5V 70 60 50 40 40 30 30 20 20 VIN = +12V fSW = 250kHz 10 0 VOUT = +1.8V ICC (mA) VOUT = +1.1V 50 90 VOUT = +1.8V 12.0 11.5 11.0 10.5 MAX5037A toc05 MAX5037A toc04 VOUT = +1.5V 100 VOUT = +1.1V VIN = +24V 10.0 9.5 9.0 8.5 8.0 VIN = +12V 7.5 7.0 6.5 6.0 VIN = +5V fSW = 500kHz 0 MAX5037A toc06 0 4 8 12 16 20 24 28 32 36 40 44 48 52 IOUT (A) η (%) VIN = +5V EXTERNALCLOCK NO DRIVER LOAD 0 4 8 12 16 20 24 28 32 36 40 44 48 52 0 4 8 12 16 20 24 28 32 36 40 44 48 52 100 150 200 250 300 350 400 450 500 550 600 IOUT (A) IOUT (A) FREQUENCY (kHz) SUPPLY CURRENT vs. TEMPERATURE AND FREQUENCY SUPPLY CURRENT vs. TEMPERATURE AND FREQUENCY SUPPLY CURRENT vs. LOAD CAPACITANCE PER DRIVER 250kHz 600kHz 150 70 100 500kHz 90 80 70 125 ICC (mA) 60 125kHz 50 40 MAX5037A toc09 90 80 175 MAX5037A toc07 100 ICC (mA) η (%) 40 30 0 4 8 12 16 20 24 28 32 36 40 44 48 52 60 100 60 50 40 75 30 0 50 40 0 70 10 60 0 4 8 12 16 20 24 28 32 36 40 44 48 52 80 ICC (mA) 50 10 90 20 70 VIN = +5V 60 20 100 10 80 MAX5037A toc08 40 90 η (%) η (%) f = 250kHz 70 50 VIN = +12V 70 80 η (%) 90 80 EFFICIENCY vs. OUTPUT CURRENT 100 MAX5037A toc02 90 100 MAX5037A toc01 100 EFFICIENCY vs. OUTPUT CURRENT AND INPUT VOLTAGE MAX5037A toc03 EFFICIENCY vs. OUTPUT CURRENT AND INTERNAL OSCILLATOR FREQUENCY VIN = +12V CDL_ = 22nF CDH_ = 8.2nF 50 30 20 VIN = +5V CDL_ = 22nF CDH_ = 8.2nF 25 -40 -15 10 35 TEMPERATURE (°C) 60 85 VIN = +12V fSW = 250kHz 10 0 -40 -15 10 35 TEMPERATURE (°C) 60 85 1 3 5 7 9 11 13 15 CDRIVER (nF) _______________________________________________________________________________________ 5 MAX5037A Typical Operating Characteristics (Circuit of Figure 1, TA = +25°C, unless otherwise noted.) Typical Operating Characteristics (continued) (Circuit of Figure 1, TA = +25°C, unless otherwise noted.) 53 1.9 2.1 MAX5037A toc11 54 VOUT = +1.8V 1.8 2.0 1.8 1.6 1.7 1.5 49 PHASE 2 1.4 48 PHASE 1 1.3 47 1.2 46 1.1 45 OVPTH (V) 1.7 51 50 1.8 1.3 VOUT = +1.1V 1.2 1.1 VIN (V) VIN (V) UNDERVOLTAGE THRESHOLD (PGOOD) vs. INPUT VOLTAGE OUTPUT VOLTAGE vs. ILOAD AND RCNTR OUTPUT VOLTAGE vs. ILOAD AND RCNTR 1.4 1.5 1.6 1.7 MAX5037A toc13 1.90 VOUT = +1.8V RCNTR = 50kΩ 1.85 1.60 MAX5037A toc14 1.3 VIN = +12V VID SETTING = +1.4V 1.55 RCNTR = 50kΩ 1.50 1.45 VOUT (V) 1.25 VOUT (V) 1.80 1.35 RCNTR = 100kΩ 1.75 1.15 1.05 RCNTR = 200kΩ 0.95 RCNTR = 100kΩ RCNTR = 200kΩ RCNTR = ∞ RCNTR = ∞ 1.25 VIN = +12V VID SETTING = +1.75V 0.85 1.60 0 4.75 4.80 4.85 4.90 4.95 5.00 5.05 5.10 5.15 5.20 5.25 1.40 1.30 1.65 0.75 1.45 1.35 1.70 VOUT = +1.1V 1.20 5 10 15 20 25 30 35 40 45 50 55 0 5 10 15 20 25 30 35 40 45 50 55 ILOAD (A) ILOAD (A) OUTPUT VOLTAGE vs. OUTPUT CURRENT AND ERROR AMP GAIN (RF / RIN) DIFFERENTIAL AMPLIFIER BANDWIDTH DIFF OUTPUT ERROR vs. SENSE+ TO SENSE- VOLTAGE PHASE 2.5 Rf / RIN = 12.5 GAIN (V/V) 1.75 1.70 Rf / RIN = 10 0 5 10 15 20 25 30 35 40 45 50 55 ILOAD (A) 0.175 0 0.150 -45 -90 1.5 1.0 -135 GAIN 0.5 1.60 45 2.0 Rf / RIN = 7.5 1.65 0.200 0 0.01 0.1 1 10 ERROR (%) 3.0 90 PHASE (DEGREES) Rf / RIN = 15 1.80 MAX5037A toc17 3.5 MAX5037A toc16 VIN = +12V VOUT = +1.8V MAX5037A toc18 VIN (V) 1.85 MAX5037A toc15 4.75 4.80 4.85 4.90 4.95 5.00 5.05 5.10 5.15 5.20 5.25 VOUT (V) 1.2 1.55 6 1.5 4.75 4.80 4.85 4.90 4.95 5.00 5.05 5.10 5.15 5.20 5.25 1.1 1.75 1.65 1.6 1.4 VOUT = +1.1V 1.0 1.0 VOUT = +1.8V 1.9 52 VOV (V) (VCSP_ - VCSN_) (mV) 2.0 MAX5037A toc10 55 VUV (V) OVERVOLTAGE THRESHOLD (OVPOUT) vs. INPUT VOLTAGE OVERVOLTAGE THRESHOLD (PGOOD) vs. INPUT VOLTAGE MAX5037A toc12 CURRENT-SENSE THRESHOLD vs. OUTPUT VOLTAGE VOUT (V) MAX5037A VRM 9.0/VRM 9.1, Dual-Phase, Parallelable, Average-Current-Mode Controller VIN = +12V NO DRIVER 0.125 0.100 0.075 -180 0.050 -225 0.025 -270 0 1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2.0 FREQUENCY (MHz) _______________________________________________________________________________________ ∆VSENSE (V) VRM 9.0/VRM 9.1, Dual-Phase, Parallelable, Average-Current-Mode Controller VCC LOAD REGULATION vs. INPUT VOLTAGE VCC LINE REGULATION ICC = 0 5.20 5.15 5.10 VIN = +8V 4.95 4.90 4.85 DC LOAD 4.80 5.00 5.00 4.95 4.90 4.90 4.85 4.85 4.80 4.80 15 30 45 60 75 90 105 120 135 150 ICC = 80mA 4.75 8 10 12 14 16 18 20 22 24 26 28 8 9 10 11 12 13 ICC (mA) VIN (V) VIN (V) DRIVER RISE TIME vs. DRIVER LOAD CAPACITANCE DRIVER FALL TIME vs. DRIVER LOAD CAPACITANCE HIGH-SIDE DRIVER (DH_) SINK AND SOURCE CURRENT 80 70 60 50 40 DL_ DH_ 30 20 10 0 11 16 21 26 31 MAX5037A toc23 80 70 60 50 40 DL_ 36 DH_ 1.6A/div DH_ 30 20 10 0 VIN = +12V fSW = 250kHz 6 100 90 tR (ns) 100 90 MAX5037A toc24 120 110 MAX5037A toc22 120 110 1 5.05 4.95 4.75 0 tR (ns) 5.10 ICC = 40mA 5.05 VCC (V) 5.00 VCC (V) VCC (V) 5.05 MAX5037A toc21 VIN = +12V 5.10 5.20 5.15 5.25 MAX5037A toc20 VIN = +24V 5.15 VCC LINE REGULATION 5.25 MAX5037A toc19 5.20 VIN = +12V CDH_ = 22nF VIN = +12V fSW = 250kHz 1 6 11 CDRIVER (nF) 16 21 26 31 100ns/div 36 CDRIVER (nF) PLL LOCKING TIME 250kHz TO 350kHz AND 350kHz TO 250kHz MAX5037A toc26 LOW-SIDE DRIVER (DL_) SINK AND SOURCE CURRENT MAX5037A toc25 PLL LOCKING TIME 250kHz TO 500kHz AND 500kHz TO 250kHz MAX5037A toc27 CLKOUT 5V/div 350kHz PLLCMP DL_ 1.6A/div 200mV/div CLKOUT 5V/div 250kHz 500kHz PLLCMP 200mV/div 0 250kHz VIN = +12V CDL_ = 22nF VIN = +12V NO LOAD VIN = +12V NO LOAD 0 100ns/div MAX5037A Typical Operating Characteristics (continued) (Circuit of Figure 1, TA = +25°C, unless otherwise noted.) 100µs/div 100µs/div _______________________________________________________________________________________ 7 MAX5037A VRM 9.0/VRM 9.1, Dual-Phase, Parallelable, Average-Current-Mode Controller Typical Operating Characteristics (continued) (Circuit of Figure 1, TA = +25°C, unless otherwise noted.) PLL LOCKING TIME 250kHz TO 150kHz AND 150kHz TO 250kHz MAX5037A toc28 HIGH-SIDE DRIVER (DH_) RISE TIME HIGH-SIDE DRIVER (DH_) FALL TIME MAX5037A toc29 MAX5037A toc30 CLKOUT 5V/div DH_ 2V/div DH_ 2V/div 250kHz PLLCMP 200mV/div 150kHz 0 VIN = +12V CDH_ = 22nF VIN = +12V CDH_ = 22nF VIN = +12V NO LOAD 100µs/div 40ns/div 40ns/div LOW-SIDE DRIVER (DL_) RISE TIME LOW-SIDE DRIVER (DL_) FALL TIME OUTPUT RIPPLE MAX5037A toc32 MAX5037A toc31 MAX5037A toc33 DL_ 2V/div DL_ 2V/div VOUT (AC-COUPLED) 10mV/div VIN = +12V VOUT = +1.75V IOUT = 52A VIN = +12V CDL_ = 22nF VIN = +12V CDL_ = 22nF 40ns/div 40ns/div 500ns/div VPGOOD 1V/div VPGOOD 1V/div VOUT 1V/div VOUT 1V/div VIN 5V/div VIN = +12V VOUT = +1.75V IOUT = 52A VIN = +12V VOUT = +1.75V IOUT = 52A 8 MAX5037A toc36 MAX5037A toc35 MAX5037A toc34 2ms/div LOAD-TRANSIENT RESPONSE ENABLE STARTUP RESPONSE INPUT STARTUP RESPONSE 1ms/div VEN 2V/div VOUT 50mV/div VIN = +12V VOUT = +1.75V ISTEP = 8A TO 52A tRISE = 1µs 40µs/div _______________________________________________________________________________________ VRM 9.0/VRM 9.1, Dual-Phase, Parallelable, Average-Current-Mode Controller REVERSE-CURRENT SINK AT INPUT TURN-ON REVERSE-CURRENT SINK vs. TEMPERATURE MAX5037A toc38 MAX5037A toc37 2.8 R1 = R2 = 1.5mΩ 2.7 R1 = R2 = 1.5mΩ VIN = +12V VOUT = +1.5V VEXTERNAL = 2.5V IREVERSE (A) VEXTERNAL = +3.3V 2.6 REVERSE CURRENT 5A/div 0A 2.5 VEXTERNAL = +2V 2.4 2.3 VIN = +12V VOUT = +1.5V -40 -15 10 35 60 85 200µs/div TEMPERATURE (°C) REVERSE-CURRENT SINK AT INPUT TURN-ON REVERSE-CURRENT SINK AT ENABLE TURN-ON MAX5037A toc39 VIN = +12V VOUT = +1.5V VEXTERNAL = 3.3V R1 = R2 = 1.5mΩ MAX5037A toc40 VIN = +12V VOUT = +1.5V VEXTERNAL = 2.5V R1 = R2 = 1.5mΩ REVERSE CURRENT 10A/div 0A REVERSE CURRENT 5A/div 0A 200µs/div 200µs/div REVERSE-CURRENT SINK AT ENABLE TURN-ON MAX5037A toc41 VIN = +12V VOUT = +1.5V VEXTERNAL = 3.3V R1 = R2 = 1.5mΩ REVERSE CURRENT 10A/div 0A 200µs/div _______________________________________________________________________________________ 9 MAX5037A Typical Operating Characteristics (continued) (Circuit of Figure 1, TA = +25°C, unless otherwise noted.) VRM 9.0/VRM 9.1, Dual-Phase, Parallelable, Average-Current-Mode Controller MAX5037A Pin Description PIN NAME FUNCTION 1–4, 44 VID3–VID0, VID4 DAC Code Inputs. VID0 is the LSB and VID4 is the MSB for the internal 5-bit DAC (Table 1). Connect to SGND for logic low or leave open circuit for logic high. These inputs have 12kΩ internal pullup resistors to an internal 3V regulator. 5, 20, 35 SGND Signal Ground. Ground connection for the internal circuitry. QFN package exposed pad connected to SGND. 6 OVPIN Overvoltage Protection Circuit Input. Connect DIFF to OVPIN. When OVPIN exceeds +13% above the VID programmed output voltage, OVPOUT latches DH_ low and DL_ high. Toggle EN low to high or recycle the power to reset the latch. 7, 43 CLP1, CLP2 Current-Error Amplifier Output. Compensate the current loop by connecting an R-C network to ground. OVPOUT Overvoltage Protection Output. Use the OVPOUT active-high, push-pull output to trigger a safety device such as an SCR. 9 PGOOD Power-Good Output. The open-drain, active-low PGOOD output goes low when the VID programmed output voltage falls out of regulation or a phase failure is detected. The power-good window comparator thresholds are +8% and -10% of the VID programmed output voltage. Forcing EN low also forces PGOOD low. 10 SENSE+ Differential Output Voltage-Sensing Positive Input. Used to sense a remote load. Connect SENSE+ to VOUT+ at the load. The device regulates the difference between SENSE+ and SENSE- according to the programmed VID code and adaptive voltage positioning. 11 SENSE- Differential Output Voltage-Sensing Negative Input. Used to sense a remote load. Connect SENSE- to VOUT- or PGND at the load. 12 DIFF Differential Remote-Sense Amplifier Output. DIFF is the output of a precision unity-gain amplifier. 13 EAN Voltage-Error Amplifier Inverting Input. Receives the output of the differential remote-sense amplifier. Referenced to SGND. 14 EAOUT 8 REG Input. A resistor on REG applies the same voltage-positioning window at different VRM voltage settings. For a no-load output voltage (VCORE) equal to VID, set RREG = RF, where the RF is the feedback resistor of the voltage-error amplifier. VREG internally regulates to the programmed VID output voltage. 15 REG 16, 39 CSP1, CSP2 17, 40 CSN1, CSN2 Current-Sense Differential Amplifier Negative Input. Together with CSP_, senses the inductor current. 18 CNTR Adaptive Voltage Center Position Input. Connect a resistor between CNTR and SGND to program the center of the adaptive VOUT position. VCNTR regulates to +1.22V. 19 EN 21, 33, 37 N.C. 22, 34 BST1, BST2 23, 32 DH1, DH2 10 Voltage-Error Amplifier Output. Connect to an external, gain-setting feedback resistor. The error amplifier gain determines the output voltage load regulation for adaptive voltage positioning. Current-Sense Differential Amplifier Positive Input. Senses the inductor current. The differential voltage between CSP_ and CSN_ is amplified internally by the current-sense amplifier gain of 18. Output Enable. A logic low shuts down the power drivers. EN has an internal 5µA pullup current. No Connection. Not internally connected. Boost Flying-Capacitor Connection. Reservoir capacitor connection for the high-side FET driver supply. Connect a 0.47µF ceramic capacitor between BST_ and LX_. High-Side Gate-Driver Output. Drives the gate of the high-side MOSFET. ______________________________________________________________________________________ VRM 9.0/VRM 9.1, Dual-Phase, Parallelable, Average-Current-Mode Controller PIN NAME FUNCTION 24, 31 LX1, LX2 Inductor Connection. Source connection for the high-side MOSFETs. Also serves as the return terminal for the high-side driver. 25, 30 DL1, DL2 Low-Side Gate-Driver Output. Synchronous MOSFET gate drivers for the two phases. 26 VDD Supply Voltage for Low-Side and High-Side Drivers. VCC powers VDD. Connect a parallel combination of 0.1µF and 1µF ceramic capacitors to PGND and a 1Ω resistor to VCC to filter out the high peak currents of the driver from the internal circuitry. 27 VCC Internal 5V Regulator Output. VCC is derived internally from the IN voltage. Bypass to SGND with 4.7µF and 0.1µF ceramic capacitors. 28 IN 29 PGND 36 CLKOUT 38 CLKIN CMOS Logic Clock Input. Drive the internal oscillator with a frequency range between 125kHz and 600kHz. Connect to VCC or SGND. Connect CLKIN to SGND to set the internal oscillator to 250kHz or connect to VCC to set the internal oscillator to 500kHz. CLKIN has an internal 5µA pulldown current. 41 PHASE Phase Shift Setting Input. Drive PHASE high for 120°, leave PHASE unconnected for 90°, and force PHASE low for 60° of phase shift between the rising edges of CLKOUT and CLKIN/DH1. 42 PLLCMP External Loop-Compensation Input. Connect compensation network for the phase-locked loop (see Phase-Locked Loop section). Supply Voltage Connection. Connect IN to VCC for a 5V system. Power Ground. Connect PGND, low-side synchronous MOSFET’s source, and VDD bypass capacitor returns together. Oscillator Output. CLKOUT is phase shifted from CLKIN by the amount specified by PHASE. Use CLKOUT to parallel additional MAX5037s. ______________________________________________________________________________________ 11 MAX5037A Pin Description (continued) VRM 9.0/VRM 9.1, Dual-Phase, Parallelable, Average-Current-Mode Controller MAX5037A Functional Diagram EN IN +5V LDO REGULATOR UVLO POR TEMP SENSOR VCC TO INTERNAL CIRCUITS VDD CSP1 DRV_VCC CSP1 CSN1 SHDN BST1 CSN1 CLP1 DH1 CLP1 LX1 PHASE 1 CLK SGND DL1 MAX5037A GMIN PGND PHASELOCKED LOOP CLKIN PHASE RAMP1 CLKOUT PLLCMP RAMP GENERATOR DIFF PGOOD CLP1 SENSE- +0.6V DIFF AMP SENSE+ CLP2 CNTR ADAPTIVE VOLTAGE POSITIONING REG N POWERGOOD GENERATOR DIFF DAC_OUT PGND 13% OF DAC_OUT EAOUT EAN OVPOUT OVP COMP ERROR AMP VID0 VID1 VID2 VID3 ROM VOLTAGEPOSITIONING DAC DAC_OUT DRV_VCC VID4 PGND RAMP2 OVPIN CLP2 CSN2 CSP2 12 SHDN CLK GMIN PHASE 2 DH2 LX2 CLP2 CSN2 CSP2 ______________________________________________________________________________________ DL2 BST2 VRM 9.0/VRM 9.1, Dual-Phase, Parallelable, Average-Current-Mode Controller The MAX5037A (Figures 1 and 2) average-currentmode PWM controller drives two out-of-phase buck converter channels. Average-current-mode control improves current sharing between the channels while minimizing component derating and size. Parallel multiple MAX5037A regulators to increase the output current VIN = +5V C1, C2 VCC VCC R13 C31 IN C32 C42 R4 19 EN VIN 42 38 PLLCMP 28 CLKIN IN 11 10 17 C3–C7 5 x 22µF 16 SENSE- SENSE+ CSN1 CSP1 DH1 44 LX1 VID3 DL1 VIN 2 BST1 VID0 VCC VDD 6 VOUT = +1.1V TO +1.85V AT 52A C14, C15 LOAD D4 C40 C26–C30, C37 VIN 12 18 C38 C16–C25 28 C39 DIFF EAN DH2 EAOUT LX2 R10 IN OVPIN 13 15 D3 27 R3 MAX5037A R9 22 OVPOUT R12 14 C12 Q2 C41 8 R8 25 R1 D1 VID1 4 R7 L1 24 VID2 3 C43 Q1 VID4 1 DAC INPUTS 23 DL2 32 C8–C11 Q3 L2 31 30 REG Q4 R2 C13 D2 CNTR CLP1 7 R6 CLP2 43 PGND 29 SGND PHASE 5, 20, 35 41 PGOOD 9 BST2 34 R11 R5 C36 C34 C35 CSN2 CSP2 40 39 VCC PGOOD C33 *SEE TABLE 2 FOR COMPONENT VALUES. Figure 1. Typical VRM Application Circuit, VIN = +5V ______________________________________________________________________________________ 13 MAX5037A capacity. For maximum ripple rejection at the input, set the phase shift between phases to 90° for two paralleled converters, or 60° for three paralleled converters. The paralleling capability of the MAX5037A improves design flexibility in applications requiring upgrades (higher load). The programmable output voltage utilizes VID codes compliant with Intel’s VRM 9.0/VRM 9.1 specifications. Detailed Description MAX5037A VRM 9.0/VRM 9.1, Dual-Phase, Parallelable, Average-Current-Mode Controller Dual-phase converters with an out-of-phase locking arrangement reduce the input and output capacitor ripple current, effectively multiplying the switching frequency by the number of phases. Each phase of the MAX5037A consists of an inner average current loop controlled by a common outer-loop voltage-error amplifier (VEA). The combined action of the two inner current loops and the output voltage loop corrects the output voltage errors and forces the phase currents to be equal. VIN = +8V TO +28V C1, C2 2 x 47µF VCC VCC R13 2.2Ω C31 C32 C42 0.1µF R4 19 EN 42 PLLCMP 38 CLKIN 28 IN VIN C3–C7 5 x 22µF 11 10 17 16 SENSE- SENSE+ CSN1 CSP1 DH1 44 1 VID3 DL1 VIN 2 4 8 VID1 BST1 VID0 VCC 6 18 28 VOUT = +1.1V TO +1.85V AT 52A C14, C15 2 x 100µF LOAD D4 C39 1µF C40 0.1µF C16–C25 2 x 270µF DIFF EAN DH2 EAOUT LX2 R10 C38 4.7µF VIN 12 15 C41 0.1µF OVPIN 13 R9 D3 27 R3 VDD 14 C12 0.47µF Q2 22 OVPOUT MAX5037A R8 25 R1 1.35mΩ D1 R12 R7 L1 0.6µH 24 VID2 3 C43 Q1 VID4 LX1 DAC INPUTS 23 DL2 32 Q3 C8–C11 4 x 22µF L2 0.6µH 31 30 REG Q4 R2 1.35mΩ C13 0.47µF D2 CNTR CLP1 7 R6 CLP2 43 PGND 29 SGND PHASE 5, 20, 35 41 PGOOD 9 CSN2 CSP2 40 39 34 R11 R5 C36 C34 C35 BST2 VCC PGOOD C33 NOTE: SEE TABLE 2 FOR COMPONENT VALUES. Figure 2. Typical VRM Application Circuit, VIN = +8V to +28V 14 ______________________________________________________________________________________ C26–C30, C37 6 x 10µF VRM 9.0/VRM 9.1, Dual-Phase, Parallelable, Average-Current-Mode Controller ICC = IQ + fSW x (QG1 + QG2 + QG3 + QG4) (2) where, Q G1, Q G2, Q G3, and Q G4 are the total gate charge of the low-side and high-side external MOSFETs, IQ is 4mA (typ), and fSW is the switching frequency of each individual phase. For applications utilizing a +5V input voltage, disable the VCC regulator by connecting IN and VCC together. Undervoltage Lockout (UVLO)/Soft-Start The MAX5037A includes an undervoltage lockout with hysteresis and a power-on reset circuit for converter turn-on and monotonic rise of the output voltage. The UVLO circuit monitors the VCC regulator output while actively holding down the power-good (PGOOD) output. The UVLO threshold is internally set between +4.0V and +4.5V with a 200mV hysteresis. Hysteresis at UVLO eliminates “chattering” during startup. The compensation network at the current-error amplifier, CLP1 and CLP2, provides an inherent soft-start to the VRM power supply. It includes a parallel combination of capacitors (C34, C36) and resistors (R5, R6) in series with other capacitors (C33, C35) (see Figure 1). The voltage at CLP_ limits the maximum current available to charge output capacitors. The capacitor on CLP_ in conjunction with the finite output-drive current of the current-error amplifier yields a finite rise time for the output current and thus the output voltage. Internal Oscillator The internal oscillator generates the 180° out-of-phase clock signals required by the pulse-width modulation (PWM) circuits. The oscillator also generates the 2VP-P voltage ramp signals necessary for the PWM comparators. Connect CLKIN to SGND to set the internal oscillator frequency to 250kHz or connect CLKIN to VCC to set the internal oscillator to 500kHz. CLKIN is a CMOS logic clock for the phase-locked loop (PLL). When driven externally, the internal oscillator locks to the signal at CLKIN. A rising edge at CLKIN starts the ON cycle of the PWM. Ensure that the external clock pulse width is at least 200ns. CLKOUT provides a phase-shifted output with respect to the rising edge of the signal at CLKIN. PHASE sets the amount of phase shift at CLKOUT. Connect PHASE to VCC for 120° of phase shift, leave PHASE unconnected for 90° of phase shift, or connect PHASE to SGND for 60° of phase shift with respect to CLKIN. The MAX5037A requires compensation on PLLCMP even when operating from the internal oscillator. The device requires an active PLL in order to generate the proper clock signal required for PWM operation. Control Loop The MAX5037A uses an average-current-mode control scheme to regulate the output voltage (Figure 3). The main control loop consists of an inner current loop and an outer voltage loop. The inner loop controls the output currents (IPHASE1 and IPHASE2), while the outer loop controls the output voltage. The inner current loop absorbs the inductor pole reducing the order of the outer voltage loop to that of a single-pole system. Most of the internal circuitry, including the oscillator, turns on when the input voltage reaches +4V. The MAX5037A draws up to 4mA of current before the input voltage reaches the UVLO threshold. ______________________________________________________________________________________ 15 MAX5037A VIN, VCC, and VDD The MAX5037A accepts an input voltage range of +4.75V to +5.5V or +8V to +28V. All internal control circuitry operates from an internally regulated nominal voltage of 5V. For input voltages of +8V or greater, the internal VCC regulator steps the voltage down to +5V. The VCC output voltage is a regulated 5V output capable of sourcing up to 80mA. Bypass VCC to SGND with 4.7µF and 0.1µF lowESR ceramic capacitors in parallel for high-frequency noise rejection and stable operation (Figure 1). VCC powers all internal circuitry. VDD is derived externally from VCC and provides power to the high-side and low-side MOSFET drivers. VDD is internally connected to the power source of the low-side MOSFET drivers. Use VDD to charge the boost capacitors that provide power to the high-side MOSFET drivers. Connect the VCC regulator output to VDD through an R-C lowpass filter. Use a 1Ω (R3) resistor and a parallel combination of 1µF and 0.1µF ceramic capacitors to filter out the high peak currents of the MOSFET drivers from the sensitive internal circuitry. Calculate power dissipation in the MAX5037A as a product of the input voltage and the total VCC regulator output current (ICC). ICC includes quiescent current (IQ) and gate drive current (IDD): (1) PD = VIN x ICC The current loop consists of a current-sense resistor, RS (an RC lowpass filter in the case of lossless inductor current sensing), a current-sense amplifier (CA_), a current-error amplifier (CEA_), an oscillator providing the carrier ramp, and a PWM comparator (CPWM_). The precision CA_ amplifies the sense voltage across RS by a factor of 18. The inverting input to the CEA_ senses the output of the CA_. The output of the CEA_ is the difference between the voltage-error amplifier output (EAOUT) and the amplified voltage from the CA_. The RC compensation network connected to CLP1 and CLP2 provides external frequency compensation for the respective CEA_. The start of every clock cycle enables the high-side drivers and initiates a PWM ON cycle. Comparator CPWM_ compares the output volt- age from the CEA_ with a 0 to 2V ramp from the oscillator. The PWM ON cycle terminates when the ramp voltage exceeds the error voltage. The outer voltage control loop consists of the differential amplifier (DIFF AMP), adaptive voltage-positioning (AVP) block, digital-to-analog converter (DAC), and voltage-error amplifier (VEA). The unity-gain differential amplifier provides true differential remote sensing of the output voltage. The differential amplifier output and the AVP connect to the inverting input (EAN) of the VEA. The noninverting input of VEA is internally connected to the DAC output. The VEA controls the two inner current loops (Figure 3). Use a resistive feedback network to set the gain of the VEA as required by the adaptive voltage-positioning circuit. CCF MAX5037A CLP1 CSP1 CSN1 RCF CCFF CA1 VIN AVP RF* IPHASE1 CEA1 SENSE+ CPWM1 DIFF AMP DRIVE 1 RS RIN* VEA SENSE- VOUT VIN CEA2 CPWM2 DAC DRIVE 2 IPHASE2 RS COUT CLP2 CSP2 CA2 CSN2 MAX5037A VRM 9.0/VRM 9.1, Dual-Phase, Parallelable, Average-Current-Mode Controller CCF RCF *RF AND RIN ARE EXTERNAL TO MAX5037A (RF = R8, RIN = R7, FIGURES 1 AND 2). CCCF Figure 3. MAX5037A Control Loop 16 ______________________________________________________________________________________ LOAD VRM 9.0/VRM 9.1, Dual-Phase, Parallelable, Average-Current-Mode Controller Peak-Current Comparator The peak-current comparator provides a path for fast cycle-by-cycle current limit during extreme fault conditions such as an output inductor malfunction (Figure 4). Note that the average current-limit threshold of 48mV still limits the output current during short-circuit conditions. So to prevent inductor saturation, select an output inductor with a saturation current specification greater than the average current limit. Proper inductor selection ensures that only extreme conditions trip the peak-current comparator, such as a broken output inductor. The 112mV voltage threshold for triggering the peak-current limit is twice the full-scale average current-limit voltage threshold. The peak-current comparator has a delay of only 260ns. Current-Error Amplifier Each phase of the MAX5037A has a dedicated transconductance current-error amplifier (CEA_) with a typical gm of 550µS and 320µA output sink and source current capability. The CEA_ outputs, CLP1 and CLP2, serve as the inverting input to the PWM comparator. CLP1 and CLP2 are externally accessible to provide frequency compensation for the inner current loops (Figure 3). Compensate CEA_ such that the inductor current down slope, which becomes the up slope to the inverting input of the PWM comparator, is less than the slope of the internally generated voltage ramp (see the Compensation section). PWM Comparator and R-S Flip-Flop The PWM comparator (CPWM) sets the duty cycle for each cycle by comparing the current-error amplifier output to a 2VP-P ramp. At the start of each clock cycle, an R-S flip-flop resets and the high-side driver (DH_) turns on. The comparator sets the flip-flop as soon as the ramp voltage exceeds the CLP_ voltage, thus terminating the ON cycle (Figure 4). Differential Amplifier The unity-gain differential amplifier (DIFF AMP) facilitates the output voltage remote sensing at the load (Figure 3). It provides true differential output voltage sensing while rejecting the common-mode voltage errors due to high-current ground paths. Sensing the output voltage directly at the load provides accurate load voltage sensing in high-current environments. The VEA provides the difference between the differential amplifier output (DIFF) and the desired VID programmed output voltage. The differential amplifier has a unity-gain bandwidth of 3MHz. The difference between SENSE+ and SENSE- regulates to the programmed VID output voltage. Connect SENSE+ to an external resistor-divider network at the output voltage to use the MAX5037A for output voltages higher than those allowed by the VID codes. DRV_VCC PEAK CURRENT COMPARATOR 112mV CLP_ CSP_ AV = 18 Gm = 550µS CSN_ BST_ PWM COMPARATOR GMIN S Q DH_ RAMP LX_ 2 x fS (V/s) CLK R Q DL_ PGND SHDN Figure 4. Phase Circuit (Phase 1/Phase 2) ______________________________________________________________________________________ 17 MAX5037A Current-Sense Amplifier The differential current-sense amplifier (CA_) provides a DC gain of 18. The maximum input offset voltage of the current-sense amplifier is 1mV and the common-mode voltage range is -0.3V to +3.6V. The current-sense amplifier senses the voltage across a current-sense resistor. MAX5037A VRM 9.0/VRM 9.1, Dual-Phase, Parallelable, Average-Current-Mode Controller Voltage-Error Amplifier The VEA sets the gain of the voltage control loop. The VEA determines the error between the differential amplifier output and the reference voltage generated from the DAC. The VEA output clamps to 0.9V relative to VCM (0.6V), thus limiting the average maximum current from individual phases. The maximum average current-limit threshold for each phase is equal to the maximum clamp voltage of the VEA divided by the gain (18) of the current-sense amplifier. This results in accurate settings for the average maximum current for each phase. Set the VEA gain using RF and RIN for the amount of output voltage positioning required within the rated current range as discussed in the Adaptive Voltage Positioning section (Figure 3). Adaptive Voltage Positioning Powering new generation processors requires new techniques to reduce cost, size, and power dissipation. Voltage positioning reduces the total number of output capacitors to meet a given transient response requirement. Setting the no-load output voltage slightly higher than the output voltage during nominally loaded conditions allows a larger downward voltage excursion when the output current suddenly increases. Regulating at a lower output voltage under a heavy load allows a larger upward voltage excursion when the output current suddenly decreases. A larger allowed voltage step excursion reduces the required number of output capacitors or allows for the use of higher ESR capacitors. Set the voltage-positioning window (∆VOUT) using the resistive feedback of the VEA. See the Adaptive Voltage-Positioning Design Procedure section and use the following equation to calculate the voltage-positioning window: (3) ∆VOUT = IOUT x RIN / (2 x GC x RF ) GC = 0.05 RS (4) where RIN and RF are the input and feedback resistors of the VEA, GC is the current-loop transconductance, and RS is the current-sense resistor or, if using lossless inductor current sensing, the DC resistance of the inductor. The voltage at CNTR (VCNTR) regulates to 1.2V (Figure 6). The current set by the resistor RCNTR is mirrored at the inverting input of the VEA, centering the output voltage-positioning window on the VID programmed output voltage. Set the center of the output voltage with a resistor from CNTR to SGND in the following manner: RCNTR = VCNTR × RIN ⎛ RIN ⎞ IOUT ⎜ ⎟ + (VOUT − VID) ⎝ 2RFGC ⎠ (5) where VOUT is a required value of output voltage at the corresponding IOUT. IOUT can be any value from no load to full load. Voltage positioning and the ability to operate with the multiple reference voltages may require the output to regulate away from a center value. Define the center value as the voltage where the output equals the VID reference voltage at one half the maximum output current (Figure 5). VCC 1X VCC 1X +1.2V VOLTAGE-POSITIONING WINDOW VCNTR + ∆VOUT/2 VCNTR VCNTR - ∆VOUT/2 NO LOAD 1/2 LOAD FULL LOAD EAN CNTR DAC_OUT 1X REG LOAD (A) Figure 5. Defining the Voltage-Positioning Window 18 Figure 6. Adaptive Voltage-Positioning Circuit ______________________________________________________________________________________ 1X VRM 9.0/VRM 9.1, Dual-Phase, Parallelable, Average-Current-Mode Controller RIN × RF RREG = ⎛ ⎞ V RIN + RF ⎜1 − COREMAX ⎟ VID ⎝ ⎠ (6) DAC Inputs (VID0–VID4) The DAC programs the output voltage. The DAC typically receives a digital code, alternatively, the VID inputs are hardwired to SGND or left open circuit. VID0–VID4 logic can be changed while the MAX5037A is active, initiating a transition to a new output voltage level. Change VID0–VID4 together, avoiding greater than 1µs skew between bits. Otherwise, incorrect DAC readings may cause a partial transition to the wrong voltage level followed by the intended transition to the correct voltage level, lengthening the overall transition time. For any low-going VID step of 100mV or more, the OVP can trip because the OVP trip reference changes instantaneously with the VID code, but the converter output does not follow immediately. The converter output drops at a rate depending on the output capacitor, inductor load, and the closed-loop bandwidth of the converter. Do not exceed a maximum VID step size of 75mV. VOLTAGE-POSITIONING WINDOW VCOREMAX ≤ VID VCOREMAX - ∆VOUT/2 VCOREMAX - ∆VOUT/2 NO LOAD 1/2 LOAD LOAD (A) Figure 7. Limiting the Voltage-Positioning Window FULL LOAD The available DAC codes and resulting output voltages (Table 1) comply with Intel’s VRM 9.0 specification. Internal pullup resistors connect the VID inputs to a nominal internal 3V supply. Force the VID inputs below 0.8V for logic low or leave unconnected for logic high. Output voltage accuracy with respect to the programmed VID voltage is ±0.8% over the -40°C to +85°C temperature range. Table 1. Output Voltage vs. DAC Codes VID INPUTS (0 = CONNECTED TO SGND, 1 = OPEN CIRCUIT) OUTPUT VOLTAGE (V) VID4 VID3 VID2 VID1 VID0 VOUT 1 1 1 1 1 Output off 1 1 1 1 0 1.100 1 1 1 0 1 1.125 1 1 1 0 0 1.150 1 1 0 1 1 1.175 1 1 0 1 0 1.200 1 1 0 0 1 1.225 1 1 0 0 0 1.250 1 0 1 1 1 1.275 1 0 1 1 0 1.300 1 0 1 0 1 1.325 1 0 1 0 0 1.350 1 0 0 1 1 1.375 1 0 0 1 0 1.400 1 0 0 0 1 1.425 1 0 0 0 0 1.450 0 1 1 1 1 1.475 0 1 1 1 0 1.500 0 1 1 0 1 1.525 0 1 1 0 0 1.550 0 1 0 1 1 1.575 0 1 0 1 0 1.600 0 1 0 0 1 1.625 0 1 0 0 0 1.650 0 0 1 1 1 1.675 0 0 1 1 0 1.700 0 0 1 0 1 1.725 0 0 1 0 0 1.750 0 0 0 1 1 1.775 0 0 0 1 0 1.800 0 0 0 0 1 1.825 0 0 0 0 0 1.850 ______________________________________________________________________________________ 19 MAX5037A Applying the voltage-positioning window at different VRM voltage settings requires that RREG = RF. The voltage on REG internally regulates to the programmed VID output voltage. Choose RREG to limit the current at REG to 50µA. For example, for a VID setting of 1.85V, calculate the minimum allowed R REG as R REG = 1.85V/50µA = 37kΩ. To use larger values of RREG while maintaining the required gain of the VEA, use larger values for RIN. In the case of a VID voltage setting equal to VCOREMAX at IOUT = 0 (no load), RCNTR = ∞ from the above equation (Figure 7). For systems requiring VCOREMAX as an absolute maximum voltage when IOUT = 0 (no load), calculate RREG using following the equation: MAX5037A VRM 9.0/VRM 9.1, Dual-Phase, Parallelable, Average-Current-Mode Controller Phase-Locked Loop: Operation and Compensation The phase-locked loop (PLL) synchronizes the internal oscillator to the external frequency source when driving CLKIN. Connecting CLKIN to VCC or SGND forces the PWM frequency to default to the internal oscillator frequency of 500kHz or 250kHz, respectively. The PLL uses a conventional architecture consisting of a phase detector and a charge pump capable of providing 20µA of output current. Connect an external series combination capacitor (C31) and resistor (R4) and a parallel capacitor (C32) from PLLCMP to SGND to provide frequency compensation for the PLL (Figure 1). The pole-zero pair compensation provides a zero at fZ defined by 1 / [R4 x (C31 + C32)] and a pole at fP defined by 1 / (R4 x C32). Use the following typical values for compensating the PLL: R4 = 7.5kΩ, C31 = 4.7nF, C32 = 470pF. When changing the PLL frequency, expect a finite locking time of approximately 200µs. The MAX5037A requires compensation on PLLCMP even when operating from the internal oscillator. The device requires an active-phase-locked loop in order to generate the proper internally shifted clock available at CLKOUT. MOSFET Gate Drivers (DH_, DL_) The high-side (DH_) and low-side (DL_) drivers drive the gates of external N-channel MOSFETs (Figure 1). The drivers’ high-peak sink and source current capability provides ample drive for the fast rise and fall times of the switching MOSFETs. Faster rise and fall times result in reduced cross-conduction losses. For modern CPU applications where the duty cycle is less than 50%, choose high-side MOSFETs (Q1 and Q3) with a moderate RDS(ON) and very low gate charge. Choose low-side MOSFETs (Q2 and Q4) with very low RDS(ON) and moderate gate charge. The driver block also includes a logic circuit that provides an adaptive nonoverlap time to prevent shootthrough currents during transition. The typical nonoverlap time is 60ns between the high-side and low-side MOSFETs. Protection The MAX5037A includes output overvoltage protection (OVP), undervoltage protection (UVP), phase failure, and overload protection to prevent damage to the powered electronic circuits. Overvoltage Protection (OVP) The OVP comparator compares the OVPIN input to the overvoltage threshold. The overvoltage threshold is typically +13% above the programmed VID output voltage. A detected overvoltage event latches the comparator output forcing the power stage into the OVP state. In the OVP state, the high-side MOSFETs turn off and the low-side MOSFETs latch on. Use the OVPOUT highcurrent-output driver to turn on an external crowbar SCR. When the crowbar SCR turns on, a fuse must blow or the source current for the MAX5037A regulator must be limited to prevent further damage to the external circuitry. Connect the SCR close to the input source and after the fuse. Use an SCR large enough to handle the peak I2t energy due to the input and output capacitors discharging and the current sourced by the power source output. Connect DIFF to OVPIN for differential output sensing and overvoltage protection. Add an RC delay to reduce the sensitivity of overvoltage circuit and avoid nuisance tripping of the converter (Figure 8). For any low-going VID step of 75mV or more, the OVP can trip because the OVP trip reference changes instantaneously with the VID code, but the converter output does not follow immediately. The converter output drops at a rate depending on the output capacitor, inductor load, and the closed-loop bandwidth of the converter. 0.1µF OVPIN 1kΩ BST_ VDD powers the low- and high-side MOSFET drivers. The high-side drivers derive their power through a bootstrap capacitor and VDD supplies power internally to the low-side drivers. Connect a 0.47µF low-ESR ceramic capacitor between BST_ and LX_. Bypass VDD to PGND with 1µF and 0.1µF low-ESR ceramic capacitors. Reduce the PC board area formed by these capacitors, the rectifier diodes between VDD and the boost capacitor, the MAX5037A, and the switching MOSFETs. DIFF RIN MAX5037A EAN RF EAOUT Figure 8. OVP Input Delay 20 ______________________________________________________________________________________ VRM 9.0/VRM 9.1, Dual-Phase, Parallelable, Average-Current-Mode Controller 2) Both phases are providing current. 3) EN is HIGH. A window comparator compares the differential amplifier output (DIFF) against 1.08 times the programmed VID output voltage for overvoltage and 0.90 times the programmed VID output voltage for undervoltage monitoring. The phase failure comparator detects a phase failure by comparing the current-error amplifier output (CLP_) with a 2.0V reference. Use a 10kΩ pullup resistor from PGOOD to a voltage source less than or equal to VCC. An output voltage outside the comparator window or a phase failure condition forces the open-drain output low. The open-drain MOSFET sinks 4mA of current while maintaining less than 0.2V at the PGOOD output. Phase Failure Detector Output current contributions from the two phases are within ±10% of each other. Proper current sharing reduces the necessity to overcompensate the external components. However, an undetected failure of one phase driver causes the other phase driver to run continuously as it tries to provide the entire current requirement to the load. Eventually, the stressed operational phase driver fails. During normal operating conditions, the voltage level on CLP_ is within the peak-to-peak voltage levels of the PWM ramp. If one of the phases fails, the control loop raises the CLP_ voltage above its operating range. To determine a phase failure, the phase failure detection circuit (Figure 9) monitors the output of the current amplifiers (CLP1 and CLP2) and compares them to a 2.0V reference. If the voltage levels on CLP1 or CLP2 are above the reference level for more than 1250 clock cycles, the phase failure circuit forces PGOOD low. Overload Conditions Average current-mode control has the ability to limit the average current sourced by the converter during a fault condition. When a fault condition occurs, the VEA output clamps to 0.9V with respect to the common-mode voltage (VCM = 0.6V) and is compared with the output of the current-sense amplifiers (CA1 and CA2) (see Figure 3). The current-sense amplifier’s gain of 18 limits the maximum current in the inductor or sense resistor to ILIMIT = 50mV/RS. Parallel Operation For applications requiring large output current, parallel up to three MAX5037As (six phases) to triple the available output current. The paralleled converters operating at the same switching frequency but different phases keep the capacitor ripple RMS currents to a minimum. Three parallel MAX5037A converters deliver up to 180A of output current. To set the phase shift of the on-board PLL, leave PHASE unconnected for 90° of phase shift (two paralleled converters), or connect PHASE to SGND for 60° of phase shift (three converters in parallel). Designate one converter as master and the remaining converters as slaves. Connect the master and slave controllers in a daisy-chain configuration as shown in Figure 10. Connect CLKOUT from the master controller to CLKIN of the first slaved controller, and CLKOUT from the first slaved controller to CLKIN of the second slaved controller. Choose the appropriate phase shift for minimum ripple currents at the input and output capacitors. The master controller senses the output differential voltage through SENSE+ and SENSE- and generates the DIFF voltage. Disable the voltage sensing of the slaved controllers by leaving DIFF unconnected (floating). Figure 11 shows a detailed typical parallel application circuit using two MAX5037As. This circuit provides four phases at an input voltage of 12V and an output voltage range of 1.1V to 1.85V at 104A. PGOOD DIFF DAC_OUT 8% OF DAC 10% OF DAC CLP1 +2.0V CLP2 PHASE FAILURE DETECTION Figure 9. Power-Good Generator ______________________________________________________________________________________ 21 MAX5037A Power-Good Generator (PGOOD) The PGOOD output is high if all of the following conditions are met (Figure 9): 1) The output is within 90% to 108% of the programmed output voltage. MAX5037A VRM 9.0/VRM 9.1, Dual-Phase, Parallelable, Average-Current-Mode Controller Applications Information Each MAX5037A circuit drives two 180° out-of-phase channels. Parallel two or three MAX5037A circuits to achieve four- or six-phase operation, respectively. Figure 1 shows the typical application circuit for twophase operation. The design criteria for a two-phase converter includes frequency selection, inductor value, input/output capacitance, switching MOSFETs, sense resistors, and the compensation network. Follow the same procedure for the four- and six-phase converter design, except for the input and output capacitance. The input and output capacitance requirement varies depending on the operating duty cycle. The examples discussed in this data sheet pertain to a typical VRM application with the following specifications: VIN = +12V VOUT = +1.1V to +1.85V IOUT(MAX) = 52A VCOREMAX = VID Programmed Output Voltage at No Load AVP (∆VOUT) = 120mV fSW = 250kHz Peak-to-Peak Inductor Current (∆IL) = 10A Table 2 shows a list of recommended external components (Figure 1) and Table 3 provides component supplier information. Table 2. Component List 22 DESIGNATION QTY C1, C2 2 47µF, 16V X5R input-filter capacitors, TDK C5750X5R1C476M DESCRIPTION C3–C11 9 22µF, 16V input-filter capacitors, TDK C4532X5R1C226M C12, C13 2 0.47µF, 16V capacitors, TDK C1608X5R1A474K C14, C15 2 100µF, 6.3V output-filter capacitors, Murata GRM44-1X5R107K6.3 C16–C25 10 270µF, 2V output-filter capacitors, Panasonic EEFUE0D271R C26–C30, C37 6 10µF, 6.3V output-filter capacitors, TDK C2012X5R0J106M C31 1 4700pF, 16V X7R capacitor, Vishay-Siliconix VJ0603Y471JXJ C32, C34, C36 3 470pF, 16V capacitors, Murata GRM1885C1H471JAB01 C33, C35, C43 3 0.01µF, 50V X7R capacitors, Murata GRM188R71H103KA01 C38 1 4.7µF, 16V X5R capacitor, Murata GRM40-034X5R475k6.3 C39 1 1.0µF, 10V Y5V capacitor, Murata GRM188F51A105 C40, C41, C42 3 0.1µF, 16V X7R capacitors, Murata GRM188R71C104KA01 D1, D2 2 Schottky diodes, ON-Semiconductor MBRS340T3 D3, D4 2 Schottky diodes, ON-Semiconductor MBR0520LT1 L1, L2 2 0.6µH, 27A inductors, Panasonic ETQP1H0R6BFX Q1, Q3 2 Upper power MOSFETs, Vishay-Siliconix Si7860DP Q2, Q4 2 Lower power MOSFETs, Vishay-Siliconix Si7886DP R1, R2 4 Current-sense resistors, use two 2.70mΩ resistors in parallel, Panasonic ERJM1WSF2M7U R3, R13 1 2.2Ω ±1% resistor 7.5kΩ ±1% resistor R4 1 R5, R6 2 1kΩ ±1% resistors R7 1 4.99kΩ ±1% resistor R8, R9 2 37.4kΩ ±1% resistors R11 1 10kΩ ±1% resistor R12 1 10Ω ±1% resistor ______________________________________________________________________________________ VRM 9.0/VRM 9.1, Dual-Phase, Parallelable, Average-Current-Mode Controller MAX5037A CSN1 CSP1 SENSE+ VIN SENSEDH1 VCC LX1 PHASE DL1 MAX5037A VCC CLKIN VIN DH2 LX2 VIN IN DL2 DIFF EAN CSP2 EAOUT CSN2 PGND SGND CLKOUT CSN1 CSP1 CLKIN VIN DH1 VCC LX1 PHASE DL1 MAX5037A IN VIN DH2 DIFF LX2 LOAD DL2 EAN CSP2 EAOUT CSN2 PGND SGND CLKOUT CSN1 CSP1 CLKIN VIN DH1 VCC LX1 PHASE DL1 MAX5037A IN VIN DH2 DIFF LX2 DL2 EAN CSP2 EAOUT CSN2 PGND SGND CLKOUT TO OTHER MAX5037s Figure 10. Parallel Configuration of Multiple MAX5037As ______________________________________________________________________________________ 23 MAX5037A VRM 9.0/VRM 9.1, Dual-Phase, Parallelable, Average-Current-Mode Controller VIN = +12V VIN C1, C2 2 x 47µF VCC C31 R13 2.2Ω C32 C43 R12 8 OVPOUT 44 C42 0.1µF R4 42 PLLCMP 38 CLKIN 28 IN VIN C3–C7 5 x 22µF 11 10 17 16 SENSE- SENSE+ CSN1 CSP1 DH1 VID3 DL1 2 3 4 VCC 19 R7 12 13 R8 14 15 R10 18 25 R1 1.35mΩ C12 0.47µF Q2 D1 VID1 BST1 VID0 VCC 22 D3 27 MAX5037A (MASTER) C38 4.7µF C41 0.1µF R3 EN 28 D4 C39 1µF C40 0.1µF OVPIN VIN DIFF EAN DH2 EAOUT LX2 R9 L1 0.6µH 24 VID2 VDD 6 Q1 VID4 LX1 1 23 DL2 32 C8–C11 4 x 22µF Q3 30 C13 0.47µF Q4 REG R2 1.35mΩ L2 0.6µH 31 D2 CNTR CLP1 7 CLP2 43 R6 PGND 29 SGND 5, 20, 35 CLKOUT 36 PHASE PGOOD 41 9 CSN2 CSP2 40 39 BST2 34 R11 R5 C36 C34 C35 PGOOD VCC C33 R24 C14, C15, C44, C45 2 x 100µF C70 VID0 R24 2.2Ω C71 C16–C25, C57–C60 2 x 270µF C26–C30, LOAD C37 6 x 10µF VID1 DAC INPUTS R25 C61 0.1µF R17 VID2 VID3 19 EN VID4 4 42 PLLCMP 28 IN VIN 38 CLKIN 5 x 22µF C46–C50 11 10 17 16 SENSE- SENSE+ CSN1 CSP1 DH1 VID1 DL1 2 1 44 23 Q5 24 25 6 R20 12 13 R21 14 D5 VID3 BST1 VID4 VCC OVPOUT VDD MAX5037A (SLAVE) 22 D7 27 R22 15 18 C65 4.7µF C64 0.1µF 28 D8 C62 1µF C63 0.1µF OVPIN VIN DIFF C51–C54 4 x 22µF EAN DH2 EAOUT LX2 R23 R14 1.35mΩ C55 0.47µF Q6 VID2 R16 8 L3 0.6µH VID0 LX1 3 DL2 32 L4 0.6µH 30 R15 1.35mΩ C56 0.47µF Q8 D6 CNTR CLP1 7 CLP2 43 R19 R18 PGND 29 C69 C67 Q7 31 REG C66 SGND PHASE PGOOD CSN2 CSP2 5, 20, 35 41 9 40 39 BST2 34 VCC C68 Figure 11. Four-Phase Parallel Application Circuit (VIN = +12V, VOUT = +1.1V to +1.85V at 104A) 24 VOUT = +1.1V TO +1.85V AT 104A ______________________________________________________________________________________ VRM 9.0/VRM 9.1, Dual-Phase, Parallelable, Average-Current-Mode Controller PHONE FAX Murata SUPPLIER 770-436-1300 770-436-3030 www.murata.com ON Semiconductor 602-244-6600 602-244-3345 www.on-semi.com Panasonic 714-373-7939 714-373-7183 www.panasonic.com TDK 847-803-6100 847-390-4405 www.tcs.tdk.com 1-800-551-6933 619-474-8920 www.vishay.com Vishay-Siliconix Number of Phases Selecting the number of phases for a voltage regulator depends mainly on the ratio of input-to-output voltage (operating duty cycle). Optimum output-ripple cancellation depends on the right combination of operating duty cycle and the number of phases. Use the following equation as a starting point to choose the number of phases: (7) NPH ≈ K/D where K = 1, 2, or 3 and the duty cycle D = VOUT/VIN. Choose K to make NPH an integer number. For example, converting VIN = +12V to VOUT = +1.75V yields better ripple cancellation in the six-phase converter than in the four-phase converter. Ensure that the output load justifies the greater number of components for multiphase conversion. Generally, limiting the maximum output current to 25A per phase yields the most costeffective solution. The maximum ripple cancellation occurs when NPH = K/D. Single-phase conversion requires greater size and power dissipation for external components such as the switching MOSFETs and the inductor. Multiphase conversion eliminates the heatsink by distributing the power dissipation in the external components. The multiple phases operating at given phase shifts effectively increase the switching frequency seen by the input/output capacitors, reducing the input/output capacitance requirement for the same ripple performance. The lower inductance value improves the large-signal response of the converter during a transient load at the output. Consider all these issues when determining the number of phases necessary for the voltage regulator application. WEBSITE Adaptive Voltage-Positioning Design Procedure The following steps outline the procedure for setting the adaptive voltage positioning: 1) Choose the voltage-error amplifier input (EAN) resistor RIN > 5kΩ. 2) Determine a reasonable amount of excursion from the desired output voltage that the system can tolerate and use as an estimate for the voltage-positioning window, ∆VOUT (see Figures 5 and 7). 3) Calculate RF from equations 22 and 23. Use equation 3 to verify that ∆VOUT remains within tolerable limits. 4) Calculate the centering resistor, RCNTR, from equation 5. RCNTR sets the center of the adaptive voltage positioning such that at 1/2 full-load current, the output voltage is the desired VID programmed output voltage (Figure 5). Do not use values less than 24kΩ for RCNTR. 5) Choose the regulation resistor, RREG, to have the same value as the feedback resistor, RF (RREG = RF). RREG maintains the adaptive voltage-positioning window at all VID output voltage settings. Do not use values less than 37kΩ for RREG. Inductor Selection The switching frequency per phase, peak-to-peak ripple current in each phase, and allowable ripple at the output determine the inductance value. Selecting higher switching frequencies reduces the inductance requirement, but at the cost of lower efficiency. The charge/discharge cycle of the gate and drain capacitances in the switching MOSFETs create switching losses. The situation worsens at higher input voltages, since switching losses are proportional to the square of input voltage. Use 500kHz per phase for VIN = +5V, 250kHz or less per phase for VIN > +12V. ______________________________________________________________________________________ 25 MAX5037A Table 3. Component Suppliers MAX5037A VRM 9.0/VRM 9.1, Dual-Phase, Parallelable, Average-Current-Mode Controller Although lower switching frequencies per phase increase the peak-to-peak inductor ripple current (∆IL), the ripple cancellation in the multiphase topology reduces the RMS ripple current of the input and output capacitor. Use the following equation to determine the minimum inductance value: LMIN = (VINMAX − VOUT ) × VOUT VIN × fSW × ∆IL (8) Choose ∆IL equal to about 40% of the output current per phase. Since ∆IL affects the output ripple voltage, the inductance value may need minor adjustment after choosing the output capacitors for full-rated efficiency. Choose inductors from the standard high-current, surface-mount inductor series available from various manufacturers. Particular applications may require custom-made inductors. Use high-frequency core material for custom inductors. High ∆IL causes large peak-topeak flux excursion increasing the core losses at higher frequencies. The high-frequency operation coupled with high ∆IL, reduces the required minimum inductance making possible even the use of planar inductors. The advantages of using planar magnetics include low-profile design, excellent current sharing between phases due to the tight control of parasitics, and low cost. For example, calculate the minimum inductance at VIN(MAX) = +13.2V, VOUT = +1.75V, ∆IL = 10A, and fSW = 250kHz: LMIN = (13.2 − 1.75) × 1.75 = 0.6µH 13.2 × 250k × 10 (9) The MAX5037A average current-mode control feature limits the maximum peak-inductor current and prevents the inductor from saturating. Choose an inductor with a saturating current greater than the worst-case peak inductor current. Use the following equation to determine the worst-case inductor current for each phase: IL _ PEAK = 0.051 ∆I + L RSENSE 2 where RSENSE is the sense resistor in each phase. (10) Switching MOSFETs When choosing a MOSFET for voltage regulators, consider the total gate charge, RDS(ON), power dissipation, and package thermal impedance. The product of the gate charge and on-resistance of the MOSFET is a figure of merit, with a lower number signifying better performance. Choose MOSFETs optimized for high-frequency switching applications. The average current from the MAX5037A gate-drive output is proportional to the total capacitance it drives from DH1, DH2, DL1, and DL2. The power dissipated in the MAX5037A is proportional to the input voltage and the average drive current. See the VIN, VCC, and VDD section to determine the maximum total gate charge allowed from all the driver outputs combined. The gate charge and drain capacitance (CV2) loss, the cross-conduction loss in the upper MOSFET due to finite rise/fall time, and the I2R loss due to RMS current in the MOSFET RDS(ON) account for the total losses in the MOSFET. Estimate the power loss (PDMOS_) in the high-side and low-side MOSFETs using the following equations: (11) PDMOS−HI = (QG × VDD × fSW ) + ⎛ VIN × IOUT × (tR + tF ) × fSW ⎞ 2 ⎜ ⎟ + 1.4RDS(ON) × I RMS−HI 4 ⎝ ⎠ where QG, RDS(ON), tR, and tF are the upper switching MOSFET’s total gate charge, on-resistance at +25°C, rise time, and fall time, respectively: D 2 2 DC + I PK + IDC × IPK × (I IRMS−HI = ) (12) 3 where D = V OUT /V IN , I DC = (I OUT - ∆I L )/2 and I PK = (IOUT + ∆IL)/2. PDMOS−LO = (QG × VDD × fSW ) + (13) 2 ⎛ 2×C ⎞ 2 OSS × VIN × fSW + 1.4R ⎜ ⎟ DS(ON) × I RMS−LO 3 ⎝ ⎠ IRMS−LO = ( ) I2DC + I2PK + IDC × IPK × (1− D) (14) 3 where COSS is the MOSFET drain-to-source capacitance. 26 ______________________________________________________________________________________ VRM 9.0/VRM 9.1, Dual-Phase, Parallelable, Average-Current-Mode Controller TJ = PDMOS x θJ-A + TA (15) Input Capacitors The discontinuous input-current waveform of the buck converter causes large ripple currents in the input capacitor. The switching frequency, peak inductor current, and the allowable peak-to-peak voltage ripple reflected back to the source dictate the capacitance requirement. Increasing the number of phases increases the effective switching frequency and lowers the peak-to-average current ratio, yielding a lower input capacitance requirement. The input ripple comprises ∆VQ (caused by the capacitor discharge) and ∆VESR (caused by the ESR of the capacitor). Use low-ESR ceramic capacitors with high ripplecurrent capability at the input. Assume the contributions from the ESR and capacitor discharge are equal to 30% and 70%, respectively. Calculate the input capacitance and ESR required for a specified ripple using the following equations: ESRIN = (∆VESR ) ⎛ IOUT ∆IL ⎞ + ⎜ ⎟ ⎝ N 2 ⎠ IOUT × D(1 − D) CIN = N ∆VQ × fSW (16) (17) where IOUT is the total output current of the multiphase converter and N is the number of phases. For example, at V OUT = 1.75V, the ESR and input capacitance are calculated for the input peak-to-peak ripple of 100mV or less yielding an ESR and capacitance value of 1mΩ and 200µF. Output Capacitors The worst-case peak-to-peak and capacitor RMS ripple current, the allowable peak-to-peak output ripple voltage, and the maximum deviation of the output voltage during step loads determine the capacitance and the ESR requirements for the output capacitors. In multiphase converter design, the ripple currents from the individual phases cancel each other and lower the ripple current. The degree of ripple cancellation depends on the operating duty cycle and the number of phases. Choose the right equation from Table 4 to calculate the peak-to-peak output ripple for a given duty cycle of two-, four-, and six-phase converters. The maximum ripple cancellation occurs when NPH = K / D. The allowable deviation of the output voltage during the fast-transient load dictates the output capacitance and ESR. The output capacitors supply the load step until the controller responds with a greater duty cycle. The response time (tRESPONSE) depends on the closed-loop bandwidth of the converter. The resistive drop across the capacitor ESR and capacitor discharge causes a voltage drop during a step load. Use a combination of SP polymer and ceramic capacitors for better transient load and ripple/noise performance. Table 4. Peak-to-Peak Output Ripple Current Calculations NO. OF DUTY PHASES (N) CYCLE (D) (%) EQUATION FOR ∆IP-P V (1 − 2D) ∆I = O L × fSW 2 < 50 2 > 50 4 0 to 25 V (1− 4D) ∆I = O L × fSW 4 25 to 50 V (1 − 2D)(4D − 1) ∆I = O 2 × D × L × fSW 4 > 50 V (2D − 1)(3 − 4D) ∆I = O D × L × fSW 6 < 17 V (1− 6D) ∆I = O L × fSW ∆I = (VIN − VO )(2D − 1) L × fSW ______________________________________________________________________________________ 27 MAX5037A For example, from the typical VRM specifications in the Applications Information section with VOUT = +1.75V, the high-side and low-side MOSFET RMS currents are 9.9A and 24.1A, respectively. Ensure that the thermal impedance of the MOSFET package keeps the junction temperature at least +25°C below the absolute maximum rating. Use the following equation to calculate maximum junction temperature: MAX5037A VRM 9.0/VRM 9.1, Dual-Phase, Parallelable, Average-Current-Mode Controller Keep the maximum output voltage deviation less than or equal to the adaptive voltage-positioning window (∆VOUT). Assume 50% contribution each from the output capacitance discharge and the ESR drop. Use the following equations to calculate the required ESR and capacitance value: ESROUT = ∆VESR ISTEP (18) I ×t COUT = STEP RESPONSE ∆VQ (19) where I STEP is the load step and t RESPONSE is the response time of the controller. Controller response time depends on the control-loop bandwidth. Compensation The main control loop consists of an inner current loop and an outer voltage loop. The MAX5037A uses an average current-mode control scheme to regulate the output voltage (Figure 3). IPHASE1 and IPHASE2 are the inner average current loops. The VEA output provides the controlling voltage for these current sources. The inner current loop absorbs the inductor pole reducing the order of the outer voltage loop to that of a singlepole system. A resistive feedback network around the VEA provides the best possible response, since there are no capacitors to charge and discharge during large-signal excursions. The required amount of adaptive voltage positioning (∆VOUT) determines the VEA gain. Use the following equation to calculate the value for RF when using adaptive voltage positioning: RF = Current Limit The average current-mode control technique of the MAX5037A accurately limits the maximum output current per phase. The MAX5037A senses the voltage across the sense resistor and limits the peak inductor current (IL-PK) accordingly. The ON cycle terminates when the current-sense voltage reaches 45mV (min). Use the following equation to calculate maximum current-sense resistor value: RSENSE = PDR = 0.045 IOUT N 2.5 × 10−3 RSENSE (20) (22) 0.05 RS (23) GC = where GC is the current-source transconductance and N is the number of phases. When designing the current-control loop ensure that the inductor downslope (when it becomes an upslope at the CEA output) does not exceed the ramp slope. This is a necessary condition to avoid subharmonic oscillations similar to those in peak current-mode control with insufficient slope compensation. Use the following equation to calculate the resistor RCF: (21) where PDR is the power dissipation in sense resistors. Select 5% lower value of RSENSE to compensate for any parasitics associated with the PC board. Also, select a noninductive resistor with the appropriate wattage rating. IOUT × RIN N × GC × ∆VOUT RCF ≤ 2 × fSW × L × 102 VOUT × RSENSE For example, the maximum RCF is 12kΩ for RSENSE = 1.35mΩ. Reverse Current Limit The MAX5037A limits the reverse current in the case that VBUS is higher than the preset output voltage setting. Calculate the maximum reverse current based on VCLR, the reverse current-limit threshold, and the currentsense resistor: IREVERSE = 28 (24) 2 x VCLR RSENSE ______________________________________________________________________________________ VRM 9.0/VRM 9.1, Dual-Phase, Parallelable, Average-Current-Mode Controller 1 CCF = 2 × π × fZ × RCF (25) BST2 SGND CLKOUT N.C. CLKIN CSP2 CSN2 PHASE PLLCMP CLP2 VID4 Pin Configuration 44 43 42 41 40 39 38 37 36 35 34 VID3 1 33 N.C VID2 2 32 DH2 VID1 3 31 LX2 VID0 4 30 DL2 SGND 5 29 PGND OVPIN 6 1 CCFF = 2 × π × fP × RCF 28 IN MAX5037A CLP1 7 27 VCC OVPOUT 8 26 VDD PGOOD 9 25 DL1 PC Board Layout SENSE+ 10 24 LX1 Use the following guidelines to lay out the switching voltage regulator. 1) Place the V IN , V CC, and V DD bypass capacitors close to the MAX5037A. 2) Minimize the area and length of the high-current loops from the input capacitor, upper switching MOSFET, inductor, and output capacitor back to the input capacitor negative terminal. 3) Keep short the current loop formed by the lower switching MOSFET, inductor, and output capacitor. 4) Place the Schottky diodes close to the lower MOSFETs and on the same side of the PC board. 5) Keep the SGND and PGND isolated and connect them at one single point close to the negative terminal of the input filter capacitor. 6) Run the current-sense lines CS+ and CS- very close to each other to minimize the loop area. Similarly, run the remote voltage sense lines SENSE+ and SENSE- close to each other. Do not cross these critical signal lines through power circuitry. Sense the current right at the pads of the current-sense resistors. SENSE- 11 7) Avoid long traces between the VDD bypass capacitors, driver output of the MAX5037A, MOSFET gates, and PGND. Minimize the loop formed by the VDD bypass capacitors, bootstrap diode, bootstrap capacitor, MAX5037A, and upper MOSFET gate. 23 DH1 BST1 N.C. SGND EN CNTR CSN1 CSP1 REG EAEOUT EAN 12 13 14 15 16 17 18 19 20 21 22 DIFF (26) MQFP/THIN QFN* *CONNECT THE QFN EXPOSED PAD TO SGND GROUND PLANE. 8) Place the bank of output capacitors close to the load. 9) Distribute the power components evenly across the board for proper heat dissipation. 10) Provide enough copper area at and around the switching MOSFETs, inductor, and sense resistors to aid in thermal dissipation. 11) Use 4oz copper to keep the trace inductance and resistance to a minimum. Thin copper PC boards can compromise efficiency since high currents are involved in the application. Also, thicker copper conducts heat more effectively, thereby reducing thermal impedance. Chip Information TRANSISTOR COUNT: 5431 PROCESS: BiCMOS ______________________________________________________________________________________ 29 MAX5037A CCF provides a low-frequency pole while RCF provides a midband zero. Place a zero at fZ to obtain a phase bump at the crossover frequency. Place a high-frequency pole (fP) at least a decade away from the crossover frequency to reduce the influence of the switching noise and achieve maximum phase margin. Use the following equations to calculate CCF and CCFF: 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.) MQFP44.EPS MAX5037A VRM 9.0/VRM 9.1, Dual-Phase, Parallelable, Average-Current-Mode Controller PACKAGE OUTLINE 44L MQFP, 1.60 LEAD FORM 21-0826 30 ______________________________________________________________________________________ D 1 1 VRM 9.0/VRM 9.1, Dual-Phase, Parallelable, Average-Current-Mode Controller DETAIL A 32, 44, 48L QFN.EPS E (NE-1) X e E/2 k e D/2 CL (ND-1) X e D D2 D2/2 b L E2/2 e E2 CL L L1 CL k DETAIL B CL L L e A1 A2 e A PACKAGE OUTLINE 32, 44, 48, 56L THIN QFN, 7x7x0.8mm 21-0144 E 1 2 PACKAGE OUTLINE 32, 44, 48, 56L THIN QFN, 7x7x0.8mm 21-0144 E 2 2 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 ____________________ 31 © 2004 Maxim Integrated Products Printed USA is a registered trademark of Maxim Integrated Products. MAX5037A Package Information (continued) (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.)