LM3753
February 3, 2012
Scalable 2-Phase Synchronous Buck Controller with
Integrated FET Drivers and Linear Regulator Controller
General Description
Not Recommended for New Designs — See LM3754
The LM3753 is a full featured single-output dual-phase volt-
age-mode synchronous PWM buck controller. It can be con-
figured to control from 2 to 12 interleaved power stages
creating a single high power output. This controller utilizes
voltage-mode control with input voltage feed-forward for high
noise immunity. An internal average current loop forces real
time current sharing between phases during load transients.
The LM3753 supports a Tracking function, so the output is
controlled both up and down. This allows the output voltage
to follow a system supply through the use of the TRACK pin.
Available in the 5 mm x 5 mm thermally enhanced 32-lead
LLP package with a thermal pad.
Features
■Wide input voltage range of 4.5V to 18V
■Up to 12 channels for 300A load
■System accuracy better than 1%
■0.6V to 3.6V output voltage range
■Switching frequency from 200 kHz to 1 MHz
■Phase current sharing ±12% max over temperature
■Integrated 4.35V ±2.3% LDO
■Inductor DCR or sense resistor current sensing
■Interleaved switching for low I/O ripple current
■Integrated synchronous NFET drivers
■Dedicated Tracking function
■Output voltage differential remote sensing
■Minimum controllable on-time of only 50 ns
■Programmable Enable and input UVLO
■Power Good flag
■OVP, UVP and hiccup over-current protection
Applications
■CPUs, GPUs (graphic cards), ASICs, FPGAs, Large
Memory Arrays, DDR
■High Current POL Converters
■Networking Systems
■Power Distribution Systems
■Telecom/Datacom DC/DC Converters
■Desktops, Servers and Workstations
Simplified Application
30091901
© 2012 Texas Instruments Incorporated 300919 SNVS614A www.ti.com
LM3753 Scalable 2-Phase Synchronous Buck Controller with Integrated FET Drivers and Linear
Regulator Controller
Connection Diagram
30091903
Top View
32-Lead LLP
Ordering Information
Order Number Package Type NSC Package Drawing Supplied As
LM3753SQ LLP-32 SQA32A 1000 Units / Reel
LM3753SQX LLP-32 SQA32A 4500 Units / Reel
Not Recommended for New Designs — See LM3754
www.ti.com 2
LM3753
Pin Descriptions
Pin Number Pin Name Description
1 HG2 Gate drive of the high-side N-channel MOSFET for Phase 2.
2 SW2 Switching node of the power stage of Phase 2.
3 LG2 Gate drive of the low-side N-channel MOSFETs for Phase 2.
4 VDD Power supply for gate drivers. Decouple VDD to PGND with a ceramic capacitor. VDD can either
be supplied by an external 5V ±10% bus, or by the internal regulator, which uses an external
NPN pass device. If using the internal regulator, connect VDD to the emitter of the NPN pass
device.
5 PGND Power Ground. Tie PGND and SGND together on the board through the DAP.
6 LG1 Gate drive of the low-side N-channel MOSFETs for Phase 1.
7 SW1 Switching node of the power stage of Phase 1.
8 HG1 Gate drive of the high-side N-channel MOSFET for Phase 1.
9 BOOT1 Bootstrap of Phase 1 for the high-side gate drive power supply.
10 PGOOD Power Good open-drain output. Active HIGH.
11 SYNCOUT Synchronization Output. For multi-controller systems this pin should be connected to the SYNC
pin of the next controller in daisy-chain configuration
12 SYNC Synchronization Input. SYNCOUT of one controller is connected to SYNC of the next controller
in a daisy-chain fashion. To synchronize the whole chain of controllers to an external clock, wire
the external clock to the SYNC pin of the first controller of the chain (called the Master controller).
Otherwise, connect the SYNC input of the Master controller to ground and all of the controllers
will be controlled by the internal oscillator of the Master.
13 FAULT Input/Output. Wire the FAULT pin of all controllers together. FAULT gets pulled Low during
startup, an over-current fault, or an over-voltage fault. FAULT = Low signals all controllers to stop
switching and prepare for the next startup sequence. The first LM3753 in the system (the Master)
supplies the FAULT pin pull-up current for all of the controllers.
14 NBASE Connect to the base of external series-pass NPN if using the LM3753 internal LDO controller to
generate VDD. Otherwise leave unconnected.
15 VIN Input Voltage. Connect VIN to the input supply rail used to supply the power stages. This input
is used to provide the feed-forward for the voltage control of VOUT and for generating the internal
VCC voltage.
16 VCC Supply for internal control circuitry. Decouple VCC to PGND with a ceramic capacitor. When VIN
> 5.5V, the internal LDO will supply 4.35V to this pin. When 4.5V < VIN < 5.5V, connect VIN to
VCC. In this case the internal VCC LDO will turn off and VCC current will be supplied directly by
VIN.
17 SGND Signal Ground. Tie PGND and SGND together on the board through the DAP.
18 COMP Error Amplifier Output. For the Master, a compensation network is placed between the COMP
pin and the FB pin. The COMP pin of the Master should be connected to the SNSP pin of each
of the Slaves. The COMP pin of each of the Slaves must be connected to its VDIF pin
19 FB Feedback Input. This is the inverting input of the error amplifier. Connect the Master FB pin to
the output voltage divider and compensation network. Connect each Slave FB pin to its own VCC
pin. This will put that controller in Slave mode and disable its error amplifier.
20 VDIF Output of the remote-sense differential amplifier. Connect the Master VDIF pin to the output
voltage divider and compensation network. The Slave differential amplifier is used to buffer
COMP from the Master controller. Connect each Slave VDIF pin to its own COMP pin.
21 SNSM Inverting input of the remote-sense differential amplifier. Connect SNSM of the Master controller
to PGND at the load point. On Slave controllers, the differential amplifier is used to buffer COMP
from the Master controller. Connect SNSM of each Slave controller directly to the Master
controller SGND pin.
22 SNSP Non-inverting input of the remote-sense differential amplifier. Connect the SNSP of the Master
controller to VOUT at the load point. On Slave controllers, the differential amplifier is used to buffer
COMP of the Master controller. Connect SNSP of each Slave controller to the Master controller
COMP pin.
3 www.ti.com
LM3753
Pin Number Pin Name Description
23 TRACK Tracking Input. Connect the TRACK pins of all of the controllers in the system together. Wire the
TRACK pin to the external TRACK control signal. Tracking is always enabled on power-up,
shutdown and brownout.
24 FREQ Frequency Adjust. A frequency adjust resistor and decoupling capacitor are connected between
FREQ and SGND to program the switching frequency between 200 kHz to 1 MHz (each phase).
These components must be supplied on the Master and Slaves, even if the system is
synchronized to an external clock.
25 IAVE Current Averaging. Connect a 4.02 kΩ, 1%, resistor between each controller’s IAVE pin and
SGND. In the case where one phase is not used, connect an 8.06 kΩ resistor. Connect a filter
capacitor between IAVE and SGND at each controller,
26 EN Enable Input. Used for VIN UVLO function, connect EN to the midpoint of a voltage divider from
VIN to SGND. The EN pins of all controllers must be wired together. For an on/off EN function,
wire the EN pins of all controllers together and control with an open drain output.
27 CS2 Positive current-sense input of Phase 2. Connect to the DCR network or the current-sense
resistor of Phase 2. The negative current-sense input is the CSM pin.
28 ILIM Current Limit Set. Connect a resistor between ILIM and CSM. The resistance between ILIM and
CSM programs the current limit.
29 CSM Negative current-sense input of the internal current-sense amplifiers. Connect to VOUT.
30 CS1 Positive current-sense input of Phase 1. Connect to the DCR network or the current-sense
resistor of Phase 1. The negative current-sense input is the CSM pin.
31 PH Phase Select Input. Connect this pin to the middle of a resistor divider between VCC and SGND
to program the number of phases in the system.
32 BOOT2 Bootstrap pin of Phase 2 for the high-side gate drive power supply.
DAP Die Attach Pad. Must be connected to PGND and SGND but cannot be used as the primary
ground connection; do not place any traces or vias other than GND in the outer layer under the
DAP; see AN-1187 application note.
www.ti.com 4
LM3753
Absolute Maximum Ratings (Note 1)
If Military/Aerospace specified devices are required,
please contact the Texas Instruments Sales Office/
Distributors for availability and specifications.
VIN to SGND, PGND −0.3V to 24V
SGND to PGND −0.3V to 0.3V
VCC and VDD to VIN +0.3V
VDD to PGND −0.3V to 6V
PGOOD, FAULT to SGND −0.3V to 6V
VCC, EN, TRACK, SYNC,
CS1, CS2, CSM, ILIM,
SNSM, SNSP to SGND
−0.3V to 6V
FREQ, PH, FB to SGND −0.3 to VCC + 0.3V
BOOT1, BOOT2 to PGND
(Note 2)
−0.3V to 24V Peak
SW1, SW2 to PGND
(Note 2)
−0.3VDC to 24V Peak
−3V for less than 40 ns
BOOT1 to SW1,
BOOT2 to SW2 (Note 2)
−0.3V to 6.0V Peak
SYNCOUT ±20 mA
PGOOD, FAULT ±20 mA
VDIF ±5 mA
COMP ±4 mA
ESD Rating
HBM (Note 3)2 kV
Junction Temperature
(TJ-MAX)
+150°C
Storage Temperature Range −65°C to +150°C
Operating Ratings (Note 1)
VIN Low Range 4.5V to 5.5V
VIN High Range when using
integrated VCC LDO
5.5V to 18V
VIN High Range when using
integrated VDD linear
regulator controller
6V to 18V
VCC External Supply Voltage 4.5V to 5.5V
VDD External Supply Voltage 4.5V to 5.5V
Output Voltage Range 0.6V to 3.6V
TRACK 0V to 5V
SYNC, EN 0V to 5.5V
SNSM −0.25V to 1.0V
SNSP to SNSM 0V to 3.6V
IAVE 0V to 1.15V
CS1 and CS2 to CSM −15 mV to 45 mV
CS1, CS2, ILIM and CSM to
SGND
0V to 3.6V
ILIM to CSM 0V to 200 mV
Junction Temperature Range −5°C to +125°C
Thermal Data
Junction-to-Ambient Thermal
Resistance (θJA), LLP-32
Package (Note 4)
26.4°C/W
Electrical Characteristics Limits in standard type are for TJ = 25°C only; limits in boldface type apply over the
junction temperature (TJ) range of −5°C to +125°C. Minimum and Maximum limits are guaranteed through test, design, or statistical
correlation. Typical values represent the most likely parametric norm at TJ = 25°C, and are provided for reference purposes only.
Unless otherwise stated VVIN = 12V, VVDD = 5V, VVCC = internal LDO, VEN = 2V, RFRQ = 78.7 kΩ, VPH = 0V, VCS1 = VCS2 = VCSM =
VTRACK = VSNSP = 1.8V, VILIM − VCSM = 100 mV, VSNSM = VSYNC = 0V, VSYNCOUT floating.
Symbol Parameter Conditions Min Typ Max Units
System Accuracy
VOUT Output Voltage Accuracy
Includes trimmed EA and diff
amplifier offset and gain errors; 0.5
mA load at VDIF
VOUT = 3.6V –0.65 –0.11 0.45 %
VOUT = 2.5V –0.75 –0.134 0.6 %
VOUT = 1.8V –0.9 –0.165 0.7 %
VOUT = 0.6V –2.25 –0.4 1.25 %
Phase Current Equalization
ΔIPH Current Equalization (from average
per phase current)
VCSM = 1.8V, VCS1 = VCS2 = VCSM + 30 mV,
VIAVE = 740 mV, VCOMP = 1.9V
–12 12 %
System Supplies and UVLO
VIN
IVIN VIN Operating Current 2-phase switching gate drivers unloaded 15 mA
IVIN-Q VIN Quiescent Current VFB = 650 mV, no PWM switching, NBASE
is floating (no NPN)
9 18 mA
IVIN-SD VIN Shutdown Current VEN = 0V 200 450 µA
5 www.ti.com
LM3753
Symbol Parameter Conditions Min Typ Max Units
VCC
VVCC VCC Linear Regulator Output
Voltage
0 to 3 mA sourced to an external load;
VVIN = 5.5V to 18V
4.25 4.35 4.45 V
IVCC VCC Input Current from External
Supply
VVIN = 5.5V, VVCC = 5.5V 10 20 mA
IVCC-SD VCC Input Shutdown Current from
External Supply
VEN = 0V, VVIN = 12V, VVCC = 5V 260 µA
IVCC-LIM VCC Output Current Limit VVCC = 2.5V 930 53 mA
VVCC = 0V 50
VVCC-EN VCC UVLO Thresholds VVCC Rising 4.04 4.14 4.24 V
VVCC Falling 3.9 44.1
VVCC-HYS VCC Threshold Hysteresis 140 mV
tD-VCC VCC UVLO/UVP Debounce Time 8 µs
VDD, NBASE, BOOT1, BOOT2, SW1, SW2
VVDD VDD Controller Regulation Voltage VVIN = 6V to 18V 4.6 4.85 5.1 V
VNBASE VIN-to-NBASE Dropout VVIN − 5.5V, 700 mV source connected
from VDD to NBASE, INBASE = 5 mA
330 mV
VVIN − 5.5V, 700 mV source connected
from VDD to NBASE, INBASE = 1 mA
130
VNBASE-REG NBASE Load Regulation VVIN = 18V, 700 mV source connected from
VDD to NBASE, INBASE steps 1 mA to 5 mA
4 mV
IVDD VDD Operating Current from
External Power Supply
VVDD = VVIN = VVCC = 5.5V, fSW = 300 kHz,
Gate Drivers unloaded
1 mA
IVDD-SD VDD Shutdown Current VEN = 0V, VVIN = 12V, VVDD = 5V 2 30 µA
INBASE-CL NBASE Current Limit VNBASE = VVDD + 0.7V, ΔVVDD = −100 mV 5.8 10 mA
VNBASE = VVDD = 0V 20
IBOOT-SD BOOT1, BOOT2 Shutdown Current VEN = 0V, VSW1(2) = 0V, VBOOT − VSW = 5V 4.5 15 µA
IBOOT BOOT1, BOOT2 Operating Current VBOOT1(2) = 17.0V, VSW1(2) = 12.0V, fSW =
300 kHz, Gate Drivers unloaded
650 µA
ISW SW1, SW2 Leakage Current with
Pre-Biased Output
VVCC = 0V, VEN = 0V, VSW1(2) = 3.6V 3 µA
VVDD-TH VDD UVLO Thresholds VVDD Rising 3.8 4.02 4.28 V
VVDD Falling 3.37 3.71 4.03 V
VVDD-HYS VDD UVLO/UVP Hysteresis 310 mV
tD-VDD VDD UVLO/UVP Debounce Time 11 µs
Thermal Shutdown
TJ-SD Thermal Shutdown Threshold Rising 160 °C
TJ-HYS Thermal Shutdown Threshold
Hysteresis
30 °C
www.ti.com 6
LM3753
Symbol Parameter Conditions Min Typ Max Units
EN
VEN-H HIGH Level Input Voltage 1.51 V
VEN-L LOW Level Input Voltage 1.14 V
VEN-TH EN Threshold VVIN = 4.5V to 18V, VVCC = 4.5V (Rising) 1.26 1.39 1.51 V
VVIN = 4.5V to 18V, VVCC = 4.5V (Falling) 1.14 1.25 1.35 V
VEN-HYS EN Threshold Hysteresis 140 mV
IEN EN Input Bias Current VEN = 1.5V 0.1 µA
VEN = 1.0V 0.4 1.7
Reference, Feedback & Error Amplifier: FB, COMP
VFB FB Voltage Under Regulation VCOMP = 1.8V 0.593 0.599 0.605 V
VFB-REG1 FB Voltage VIN Line Regulation VVIN = 5.5V to 18V ±0.01 %
VFB-REG2 FB Voltage VCC Line Regulation VVCC = VVIN = VVDD = 4.5V to 5.5V (same
source)
±0.15 %
IFB FB Input Bias Current 45 130 nA
VFB-PTH FB Pin Master/Slave Programming
Threshold
3.2 V
AOL DC Gain FB to COMP, VCOMP = VFB + 1.0V 70 dB
fBW Error Amplifier
Unity Gain Bandwidth
RCOMP-SGND = 1.5 kΩ, CCOMP-SGND = 50 pF 15 MHz
VCOMP-SLEW Error Amplifier Slew Rate 6 V/µS
VCOMP-REG COMP Load Regulation, Sourcing VCOMP = 2.7V, ΔICOMP = +1 mA, DC Gain
= 40
−3 mV
PWM Ramp and Input Voltage Feed-Forward
DMAX Maximum Duty Cycle Controlled by
Clock
VVIN = 6V, VCOMP = 3.5V 81 %
DFF Duty Cycle Controlled by VIN Feed-
Forward
VVIN = 9V, VCOMP = 2.2V 42 %
tON-MIN Minimum Controllable On-Time 50 ns
VRAMP-MIN PWM Ramp Range Ramp Minimum 1.3 V
VRAMP-MAX Ramp Maximum 2.8 V
VRAMP PWM Ramp Amplitude 1.5 V
Differential Amplifier: SNSP, SNSM, VDIF
VOS-INPUT Input Offset Voltage VSNSP = 1.8V 3 mV
RINPUT-SNSP Input Resistance of SNSP 30 kΩ
AV-DIF Gain VSNSP = 0.6V to 3.6V 0.996 11.004 V/V
fBW-DIF 3dB Bandwidth 2 MHz
VDIF-REG1 VDIF Load Regulation, Sourcing VVDIF = 3.6V, IVDIF = 0.5 mA −3 mV
VDIF-REG2 VDIF Load Regulation, Sourcing VVDIF = 0.6V, IVDIF = 0.5 mA −3 mV
Current-Sense, Current Limit and Hiccup Mode: CS1, CS2, CSM, ILIM
VCS-OS Current-Sense Input Offset Voltage
Range, VCS1(2) – VCSM
VOUT = 1.8V ±2 mV
ICS CS1, CS2 Input Bias Current VCSM = 3.6V, VCS1(2) − VCSM = −15 mV and
+40 mV
−200 200 nA
VCSM = 0.6V, VCS1(2) − VCSM = −15 mV and
+40 mV
−450 450 nA
ICSM CSM Input Source Bias Current VCSM = 0.6V and 3.6V, VCS1(2) − VCSM = 40
mV
150 240 µA
ICSL CS1+ CS2 + CSM + ILIM Leakage
Current with Pre-Biased Output
VVCC = 0V, VEN = 0V, VCSM = VCS1 = VCS2
= VILIM = 3.6V
0.1 µA
7 www.ti.com
LM3753
Symbol Parameter Conditions Min Typ Max Units
fBW-CS 3dB Bandwidth, CS1(2) to PWM
COMPARATOR Input
1.0 MHz
IILIM-SOURCE ILIM Source Current VILIM = 0.6V to 3.6V, VVIN = 5.5V 85 94 103 µA
VCL Current Limit Threshold Voltage
VILIM − VCS1(2)
VILIM = 0.6V to 3.6V, VVIN = 5.5V −2.5 04.6 mV
tD-CL Current Limit Comparator
Propagation Delay
VCS1 or VCS2 stepped from 0.9V to 1.1V,
VILIM = 1V
200 ns
tD-ILIM Master or Slave Fast Current Limit
Delay
VFB = 280 mV, 1-phase over-current:
VCS1 OR VCS2 > VILIM
7 Switch
cycles
VFB = 280 mV, 2-phase over-current:
VCS1 AND VCS2 > VILIM
3 Switch
cycles
tD-HICCUP Master or Slave Over-Current
Hiccup Mode Delay
1-phase over-current:
VCS1 OR VCS2 > VILIM
446 Switch
cycles
2-phase over-current:
VCS1 AND VCS2 > VILIM
223 Switch
cycles
tD-COOL-DOWN Hiccup Over-Current Cool-Down
Time
6 ms
Power Good: PGOOD, OVP, UVP
VOVP OVP Threshold VFB rising edge 125 130 135 %VFB
tD-RESTART OVP Restart Delay 2 ms
NOVP-LATCH Number of OVP Events Before
Latch-Off
7
VUVP UVP Threshold VFB falling edge 75 80 85 %VFB
VUVP-HYS UVP Threshold Hysteresis 25 mV
tD-OVP/UVP OVP/UVP Debounce Time 5 µs
VPG-LO PGOOD Low Level IPGOOD = −4 mA 0.14 0.25 V
IPG-LEAK PGOOD Leakage Current VPGOOD = 5.5V 5 300 nA
FAULT
IFAULT Internal Pullup Current in Master
Mode
325 µA
VOL-FAULT FAULT Output Low Level IFAULT sinking 500 µA 0.21 V
VOH-FAULT FAULT Output High Level IFAULT sourcing 50 µA VCC −
0.1
V
Oscillator and Synchronization (PLL): SYNC, SYNCOUT, FREQ
fSW-MIN Minimum Switching Frequency RFRQ = 121 kΩ 200 kHz
fSW-MAX Maximum Switching Frequency RFRQ = 21.3 kΩ 1000 kHz
fSW Switching Frequency Accuracy RFRQ = 78.7 kΩ282 300 318 kHz
fSYNC SYNC Frequency Capture Range 200 kHz to 1 MHz ±25 %
VSYNC-RISE SYNC Rising Threshold 1.46 1.68 V
VSYNC-FALL SYNC Falling Threshold 1.12 1.3 V
tSYNC-MIN SYNC Minimum Pulse Width 150 ns
ISYNC SYNC Bias Current
(internal or external VCC)
VSYNC = 0 to 5.5V −15 25 µA
VSYNCOUT-HI SYNCOUT Logic High Level Sourcing 10 mA, VVCC = 4.5V external VCC −
0.42
V
VSYNCOUT-LO SYNCOUT Logic Low Level Sinking 10 mA, VVCC = 4.5V external 0.48 V
www.ti.com 8
LM3753
Symbol Parameter Conditions Min Typ Max Units
PHRATIO VPH/VVCC Divider Ratio to Set
Phase Number
2 & 4 Phases 0 0.138
3 Phases 0.152 3/14 0.279
5 Phases 0.294 5/14 0.418
6 Phases 0.438 7/14 0.562
8 Phases 0.587 9/14 0.703
10 Phases 0.730 11/14 0.844
12 Phases 0.874 1
IPH PH Bias Current VVCC = 4.5V forced, VPH = 0 to VVCC −150 150 nA
ΦHG1-N2 HG1 to HG2 Phase Shift for 2, 4, 6,
8, 10 or 12-Phase Modes
180 °
ΦHG1-N3 HG1 to HG2 Phase Shift for 3-
Phase Mode
240 °
ΦHG1-N5 HG1 to HG2 Phase Shift for 5-
Phase Mode
216 °
ΦSYNC SYNC to SYNCOUT Phase Shift
for N-phase Operation
N > 2 360/N °
N = 2 90
tSYNC-ERR SYNC to SYNCOUT Phase Shift
Error
5 ns
tSYNC-HG SYNC to HG1(2) 165 ns
ΦHG-ERR HG1 and HG2 Controller-to-
Controller Phase Delay Error
300 kHz, 6-phase 5 °
Tracking: TRACK
VTRACK Tracking Range 0 VREF V
VHYS-TRACK TRACK Falling Voltage Hysteresis 50 mV
tSS-INT Internal Soft-Start Time during
Fault Recovery
After Fault 3.8 ms
ITRACK TRACK Input Bias Current VTRACK = 0.3V 5 200 nA
VTRACK = 5V 0.2 mA
tLG-PW1 First LG = High Pulse Width during
Fault Recovery
460 ns
tLG-GTF LG Asynchronous-to-Synchronous
Gradual Transition Time during
Fault Recovery
1.8 ms
tD-EN-SW EN-to-Switching Delay Delay from EN = High to FAULT = High; no
pre-bias; VTRACK = 0.6V
2 ms
9 www.ti.com
LM3753
Symbol Parameter Conditions Min Typ Max Units
Gate Drivers
IPK-HG-SOURCE HG1 and HG2 Peak Source
Current
Less than 100 ns 1.9 A
RHG-SOURCE HG1 and HG2 Source Resistance VBOOT − VSW = 5V 2.5 Ω
IPK-HG-SINK HG1 and HG2 Peak Sink Current Less than 100 ns 4 A
RHG-SINK HG1 and HG2 Sink Resistance VBOOT − VSW = 5V 1 Ω
IPK-LG-SOURCE LG1 and LG2 Peak Source Current Less than 100 ns 2.3 A
RLG-SOURCE LG1 and LG2 Source Resistance 2 Ω
IPK-LG-SINK LG1 and LG2 Peak Sink Current Less than 100 ns 4 A
RLG-SINK LG1 and LG2 Sink Resistance 1 Ω
RHG-PULLDOWN HG-SW Pull-Down Resistor 16 kΩ
RLG-PULLDOWN LG-PGND Pull-Down Resistor 16 kΩ
tD-HG-LG HG Falling to LG Rising Cross-
Conduction Protection Delay
(Dead-Time)
SW node not switching 30 ns
tD-LG-HG LG Falling to HG Rising Delay 28 ns
tDS-HG-LG HG Falling to LG Rising Cross-
Conduction Protection Delay
(Dead-Time)
SW node switching 10 ns
Note 1: Absolute Maximum Ratings indicate limits beyond which damage to the device may occur, including inoperability and degradation of device reliability
and/or performance. Functional operation of the device and/or non-degradation at the Absolute Maximum Ratings or other conditions beyond those indicated in
the Recommended Operating Conditions is not implied. Operating Range conditions indicate the conditions at which the device is functional and the device should
not be operated beyond such conditions. For guaranteed specifications and conditions, see the Electrical Characteristics table.
Note 2: Peak is the dc plus transient voltage including switching spikes.
Note 3: Human Body Model (HBM) is a 100 pF capacitor discharged through a 1.5 kΩ resistor into each pin. Applicable standard is JESD22-A114C. All pins
pass 2 kV HBM except VDD, VIN and VCC which are rated for 1.5 kV.
Note 4: Tested on a four layer JEDEC board. Four vias provided under the exposed pad. See JEDEC standards JESD51-5 and JESD51-7.
www.ti.com 10
LM3753
Typical Performance Characteristics
System Accuracy vs VOUT
30091904
fSW vs Temperature
30091907
VREF Deviation
30091905
RFRQ vs fSW
30091908
VREF vs Temperature
30091906
Load Step (High Slew)
30091909
11 www.ti.com
LM3753
Over-Voltage Fault
30091913
Repeated Over-Voltage Conditions
30091914
Tracking Startup
30091912
Over-Current Fault (Soft Short)
30091915
www.ti.com 12
LM3753
Block Diagram
30091916
13 www.ti.com
LM3753
Functional Description
GENERAL
The LM3753 is a two-phase voltage-mode step-down (buck)
switching regulator controller. From one to six LM3753 con-
trollers can be connected together to control from two to
twelve phases (2, 3, 4, 5, 6, 8, 10, or 12 phases). Since ex-
ternal switching components can typically handle 25A per
phase, a 12 phase system can supply a total of 300A.
Multiple controllers in a system communicate with each other
and work together. They will startup and shut down together,
each phase on each controller will share current equally, and
all the phases will react in unison to fault conditions. In a multi-
controller system, all controllers are the same part. One con-
troller functions as the Master and all the others act as Slaves.
The Master and Slave are differentiated by how they are con-
nected in the system. The Master controller senses the sys-
tem output voltage and VIN (as well as TRACK) and sets the
target duty cycle for each phase on all of the controllers. The
Master and Slave controllers monitor the current-sense infor-
mation from each phase. Based on this current information,
the controllers adjust the duty cycle on each phase up or down
from the target level, in order to achieve optimal current shar-
ing.
Each controller incorporates a phase locked loop (PLL) that
communicates with the PLLs on the other controllers. By this
means, the switching edges of the different phases are
spread out equally within one switch period. For N phases
operating at any switching frequency, the angle in degrees
between one phase switching and the next is 360° / N. A
SYNC pin is available that can be used to lock the Master
switching frequency and phase to an external clock.
The LM3753 has a Tracking function. The output voltage will
follow the TRACK pin, both up and down, whenever it is less
than VREF. Synchronous switching is always enabled, ex-
cept during fault recovery.
CONTROL ALGORITHM
The control architecture is primarily voltage-mode. An error
amplifier amplifies the difference between the FB pin voltage
and the internal reference voltage to generate a COMP signal.
This signal is compared against a ramp that consists of a fixed
value plus a term proportional to VIN which controls the duty
cycle. In order to facilitate current sharing there is an inner
current-sense loop. Information for the current through the in-
ductor in each phase is sensed either with a sense resistor or
with a DCR arrangement which uses the DC resistance of the
inductor. This current-sense signal is connected to the CS pin
(CS1 or CS2). The negative reference for current-sense is
VOUT which is common for both phases and connected to the
controller’s CSM pin. The controller amplifies the (CS1(2) –
CSM) voltage difference for each phase, and compares it to
the voltage on the IAVE pin, which tracks the average current
of all phases. Any phase whose current is more than the av-
erage has its duty cycle decreased and vice versa. The IAVE
signal is common to all controllers in a system. Each controller
outputs a current onto the IAVE bus so that the total current
on the bus is the sum of the current signals from all of the
phases. An external resistor to ground translates this current
signal to a voltage, which all of the controllers read back.
The LM3753 includes an uncommitted differential amplifier.
On the Master controller this amplifier is used to remotely
sense the converter’s output voltage, typically at the load. On
the Slave controllers this amplifier is used to buffer the Master
controller’s COMP signal and level shift it to the Slave
controller’s local ground.
POWER CONNECTIONS
The LM3753 has three supply pins, which are VIN, VCC, and
VDD. It employs two ground pins, SGND and PGND. VDD
and PGND are the power and ground for the gate driver stage
that controls the HG and LG pins. The quiescent current
drawn by VDD is very small – around 1 mA. To predict the
VDD current requirement one can assume it is mostly switch-
ing current and use the standard formula:
IVDD = (1 or 2) x fSW x QTOTAL_PHASE
QTOTAL_PHASE is the sum of the high-side switch gate charge
and the low-side gate charge. The (1 or 2) factor corresponds
to one or two phases running. The low-side driver is powered
directly from VDD. The high-side driver draws its power from
VDD through the external bootstrap Schottky diode. The rest
of the controller is powered by VCC and SGND.
The LM3753 has two on-board regulators, one to generate
VCC and one to generate VDD. The VCC regulator is self-
contained and only needs a 4.7 μF ceramic capacitor to
SGND. The VDD regulator uses an external NPN pass de-
vice. This device should be sized to meet the VIN to VDD
dropout requirements for the calculated IVDD. The collector of
this device goes to VIN, the base goes to NBASE and the
emitter goes to VDD. VDD also needs a 4.7 µF bypass ca-
pacitor to PGND. The internal VIN to NBASE dropout is
approximately 300 mV. The minimum VIN is calculated as:
VINMIN = VDDMIN + VBE_NPN + 300 mV
VDDMIN = MAX(VDDUVLO, VGATE-MIN)
VDDUVLO is the controller’s maximum VDD under-voltage
lockout voltage, which is 4.06V. VGATE-MIN is the minimum re-
quired gate drive voltage for the power MOSFET switches.
VINMIN is typically 5.5V to 6.0V. For VIN less than 5.5V, the
regulators are omitted and the VCC and VDD pins are con-
nected as shown in Figure 3.
30091919
FIGURE 1. Power Connections Using the Internal
Regulator
www.ti.com 14
LM3753
30091921
FIGURE 2. Power Connections Using a System 5V Rail
30091923
FIGURE 3. Power Connections for VIN = 5V
UNDER-VOLTAGE LOCKOUTS and ENABLE
The LM3753 controller has internal under-voltage lockout
(UVLO) detection on the VCC and VDD supplies. The under-
voltage lockout on VIN is set using the EN pin threshold.
Connect a voltage divider between VIN and SGND with the
midpoint going to the EN pin. The division ratio and the EN
pin threshold determine the VIN level that enables the con-
troller. This divider should be used in all cases. If the system
does not have a particular VIN under-voltage lockout require-
ment, the level is set to be below the minimum VIN level at
the worst case combination of tolerances and operating con-
ditions.
To guarantee startup at the lowest input voltage, set the di-
vider to the VEN-TH rising max specification. For a higher
accuracy VIN UVLO operation, the resistor divider minimum
current should be 1 mA or higher. This will reduce the thresh-
old error contribution of the EN pin bias current, which is
guaranteed to be less than 1.7 µA over temperature. The en-
able pin can also be used as a digital on-off. To do this, the
enable signal should be used to pull down the midpoint of the
voltage divider using open-drain logic or a transistor. A cus-
tomary implementation uses an external MOSFET.
30091924
FIGURE 4. Input Voltage UVLO with External Enable
While the EN pin has a threshold hysteresis of 140 mV typical,
a small noise-filtering capacitor may be added between the
EN pin and SGND. This is particularly useful when the con-
troller is turning on via the resistor divider by a slowly rising
VIN rail.
STARTUP SEQUENCE
When EN is below its threshold, the internal regulators are off
and the controller is in a low power state. When EN crosses
above its threshold the VCC regulator turns on. When VCC
rises above its under-voltage lockout threshold the VDD reg-
ulator turns on. When VDD rises above its under-voltage
lockout threshold the controller is ready to start.
If VDD or VCC is supplied externally and already sitting above
its under-voltage lockout point, then the controller is ready for
startup as soon as EN crosses above its threshold. Anytime
VCC or VDD drops below its UV threshold, switching stops
and the controller goes into a standby state. It will go through
normal startup once the supplies recover.
When the controller is ready to start, it reads the voltage on
the PH pin and determines how many phases are running in
the system. By this means the phase delay from SYNC to
SYNCOUT through the PLL is configured. Following this the
oscillator and PLL turn on and pulses will be observed on
SYNCOUT.
A 2 ms timer is initiated so that all of the PLLs in the system
can synchronize up. As each controller times out, it stops
pulling its FAULT pin low. At the end of this sequence, the
FAULT bus rises and the controllers are ready to switch.
The error amplifier uses a different input stage when TRACK
is below VREF. During normal operation the error amplifier
employs a low offset bipolar input stage. At startup, a MOS
input stage is used during the track phase which has a lower
input bias current but a higher input offset voltage. A 40 mV
offset is introduced when TRACK is less than 70 mV. This
offset forces the error amplifier output to be low during startup.
The offset transitions progressively to zero as TRACK moves
from 0 to 70 mV.
TRACKING
The LM3753 implements a tracking function. The error am-
plifier amplifies the minimum of VREF or TRACK at the FB
pin. By means of the closed loop regulation through the
switching stage, FB will be regulated to TRACK. When
TRACK is below VREF, the LM3753 will control FB both up
and down to follow TRACK. When TRACK is above VREF,
15 www.ti.com
LM3753
FB will be regulated to VREF. A pre-biased output will be
pulled down by the LM3753. Full synchronous switching is
always employed on the LM3753, except for restart after a
fault condition.
When the LM3753 is ready to switch, normally TRACK will be
grounded and COMP will be low. LG will get pulled to VDD to
turn on the synchronous switch. As TRACK slews above FB,
COMP will slew up and LG will go high for 300 ns to charge
the HG bootstrap capacitor. Following this HG begins switch-
ing. COMP will set the duty cycle with normal PWM control of
HG and LG. The loop acts to have FB follow TRACK. If
VOUT is too high, it will get pulled down. An internal timer sets
a 2 ms delay from the time of the first HG pulse, which occurs
as soon as COMP slews above the PWM ramp bottom.
When the 2 ms times out, PGOOD goes high if FB is above
the output under-voltage threshold on the Master, TRACK is
above VREF, no fault conditions are present, and SYNC is
toggling on the Slaves.
PHASE NUMBER SELECTION
The voltage at the PH pin determines the phase shift between
the two phases of each controller and also the phase shift
between the SYNC and SYNCOUT pulses in a Master-Slave
configuration. This voltage is read at startup and the resulting
phase configuration saved. The PH pin should be connected
to the center of a resistor divider between VCC and SGND to
select and program the required number of phases and the
corresponding phase delays per Table 1. Each controller re-
quires the same resistor divider at the PH pin.
30091925
FIGURE 5. Phase Selection
TABLE 1. Phase Divider Resistors
Number Of
Phases
Divide Ratio
Target
RPH1
(± 1%)
RPH2
(± 1%)
2 & 4 Phases 0.000 Omit 0
3 Phases 0.214 7870Ω 2150Ω
5 Phases 0.357 6490Ω 3570Ω
6 Phases 0.5 4990Ω 4990Ω
8 Phases 0.643 3570Ω 6490Ω
10 Phases 0.786 2150Ω 7870Ω
12 Phases 1 0 Omit
OVER-CURRENT and OVER-VOLTAGE FAULTS
If any controller experiences a fault condition, it will pull the
FAULT bus low and all of the controllers will stop switching.
From the time when EN is low to the point where FAULT rises,
both HG and LG are low so that the SW node of each phase
is floating. The FAULT input may be pulled low externally
through an open drain MOSFET to disable the system.
The LM3753 employs cycle-by-cycle current limiting. This oc-
curs on each phase for both Master and Slave controllers. The
current (that is the CS1(2) − CSM voltage) is continuously
compared to the over-current set point (ILIM − CSM). Any
time that the current-sense signal exceeds current limit, the
cycle is ended.
In order to determine that a current fault has occurred, each
controller counts the number of over-current pulses. When
the sum of the counts for phase 1 and phase 2 reaches 446
an over-current fault is declared. The counter is reset after 16
consecutive switching cycles with no over-current on either
phase.
There is a second method for achieving an over-current fault,
which is meant to react to heavy shorts on VOUT. The Master
controller will determine that an over-current fault has oc-
curred after 7 over-current cycles if the voltage at the FB pin
is less than 50% of its target value. This feature is disabled
during startup. Since the Slave controllers do not see the FB
voltage, they cannot detect this type of fault.
Any controller which sees an over-current fault will respond
by pulling the FAULT bus low. All of the controllers will react
and stop switching. Both HG and LG on each phase will be
pulled low. The inductor current in each phase will decay
through the body diodes of the low-side switches. The con-
troller which recognized the over-current fault will hold
FAULT low for 6 ms, which determines the hiccup time. This
allows the energy stored in the inductors to dissipate. After
this, FAULT is released and all of the controllers will restart
together.
The restart after fault for the LM3753 is different from the initial
startup. When an over-current fault occurs, TRACK is usually
above VREF. In order to avoid VOUT slewing up precipitously,
a fixed time internal soft-start is connected to the error ampli-
fier to control the rise of VOUT. The low-side switch is not
turned on until the internal soft-start exceeds FB or VREF,
which allows VOUT to remain high. The error amp will use as
a reference the minimum of VREF, TRACK or the internal
soft-start. Once switching ensues a gradual transition to fully
synchronous operation occurs.
Over-voltage faults are only recognized by the Master con-
troller. About 5 µs after FB crosses above the OVP threshold,
which is 30% above VREF, the Master controller declares an
over-voltage fault. It pulls the FAULT bus low and all of the
controllers stop switching, with HG being low and LG being
high. The low-side MOSFETs pull VOUT down to remove the
over-voltage condition. As soon as FB crosses below the un-
der-voltage detect point, which is 20% below VREF, the LG
outputs go low to turn off the low-side MOSFETs. This pre-
vents the negative inductor current from ramping too high.
The Master controller then waits 2 ms to allow any negative
inductor current to transition into the high-side MOSFETs
body diodes.
The restart from an over-voltage fault is the same as the
restart from an over-current fault. In addition there is an over-
voltage fault counter. On the seventh over-voltage fault, the
system does not restart. It waits for power or EN to be cycled.
This counter is reset to zero when power goes low or EN
crosses below its threshold.
PGOOD and PGOOD DELAY
PGOOD is an open-drain logic output. It is asserted HIGH
when the output voltage level is within the PGOOD window,
which is typically −20% to +30%. In order to operate, the
PGOOD output requires a pull-up resistor to an appropriate
supply voltage. This voltage is typically the supply for an ex-
ternal monitoring circuit. The resistor is selected so that it
limits the PGOOD sink current to less than 4 mA.
www.ti.com 16
LM3753
PGOOD is delayed from either power-up or VIN under-volt-
age lockout, and has three primary factors:
1) A synchronization delay, set to 2 ms after the slowest
controller in the system recognizes a valid level on EN, VCC
and VDD. This delay is timed out internally and allows for the
phase lock loops to synchronize.
2) TRACK up, in non-fault conditions.
3) Transition period from diode emulation mode to fully
synchronous operation, set to 2 ms.
CURRENT SENSE and CURRENT LIMIT
The LM3753 senses current to enforce equal current sharing
and to protect against over-current faults. There are two sys-
tem options for sensing current; a current-sense resistor, or
a DCR configuration which uses the DC resistance of the in-
ductor. The current-sense resistor is more accurate but less
efficient than the DCR configuration.
The input range of the differential current-sense signal (CS1
(2) – CSM) is from −15 mV to +40 mV. The common mode
range is the same as the controller’s output range which is 0V
to 3.6V. Two considerations determine the value of the cur-
rent-sense resistor. If the resistor is too large there is an
efficiency loss. If it is too small the current-sense signal to the
controller will be too low. Choose a resistor that gives a full
load current-sense signal of at least 25 mV. This is typically
a resistor in the 1 mΩ to 2 mΩ range. The current-sense re-
sistor is inserted between the inductor and the load. The load
side of the resistor which is VOUT, is connected to CSM, the
negative current-sense input. This is the negative current-
sense reference for both phases. The positive side of the
current-sense resistor goes to CS1(2).
For the DCR configuration a series resistor-capacitor combi-
nation is substituted for the current-sense resistor. The resis-
tor connects to the switch node (SW) and the capacitor
connects to VOUT. CSM is connected to VOUT as with the
sense resistor. CS1(2) is connected to the center point of the
resistor and capacitor, so that the current-sense signal is de-
veloped across the capacitor. The voltage across the capac-
itor is a low pass filtered version of the voltage across the
resistor-capacitor combination, in the same way the current
through the inductor is a low pass filtered version of the volt-
age applied across the inductor and its intrinsic series resis-
tance. Choose the DCR time constant (RDCR x CDCR) to be
1.0 to 1.5 times the inductor time constant (L / RL). RDCR is
selected so that the CS pin input bias current times RDCR does
not cause a significant change in the CS voltage. The inductor
time constant and the DCR time constant will skew over tem-
perature since the components have different temperature
coefficients. Critical applications may employ a correction cir-
cuit based on a positive temperature coefficient thermistor
(PTC).
The over-current limit is set by placing a resistor between ILIM
and CSM. The value of the resistor times the ILIM current of
94 µA sets the over-current limit.
CURRENT SHARING and CURRENT AVERAGING
The current sharing works by adjusting the duty cycle of each
phase up or down to make the phase current equal to the
average current. The maximum duty cycle shift is ±20%.
To determine the average current, each phase sources a cur-
rent onto the IAVE bus proportional to its load current as
measured by the current sense amplifier connected to the
CS1(2) and CSM pins. The IAVE pins of all controllers are
connected together and a resistance of 8 kΩ per phase (par-
allel) to SGND provides the proper voltage level for the IAVE
bus. Each phase compares its current sense output to the
IAVE bus and sums the resultant voltage into the common
COMP signal to adjust the duty cycle for optimum current
sharing.
IAVE forms the current sharing bus for the entire power con-
verter. The IAVE pins of all controllers must be connected
together. Filter capacitors with a time constant of RAV x CAV =
1 / fSW are connected between IAVE and SGND of each con-
troller. The parallel combination of the filter capacitors times
the summing resistors (one set per controller) forms the time
constant of the current sharing bus.
ERROR AMPLIFIER and LOOP COMPENSATION
The LM3753 uses a voltage mode PWM control method. This
requires a TYPE III or 3 pole, 2 zero compensation for opti-
mum bandwidth and stability. The error amplifier is a voltage
type operational amplifier with 70 dB open loop gain and unity
gain bandwidth of 15 MHz. This allows for sufficient phase
boost at high control loop frequencies without degrading the
error amplifier performance.
The error amplifier output COMP connections are different for
Master and Slave controllers. For the Master, a compensation
network is placed between the COMP pin and the FB pin. The
COMP pin of the Master is connected to the SNSP pin of each
Slave. The SNSM pin of each Slave is connected to the bot-
tom of the Master feedback divider at SGND. The COMP pin
of each Slave is connected to its corresponding VDIF pin. This
provides sufficient buffering of the master COMP signal for
the internal summing of the current averaging circuit.
OSCILLATOR and SYNCHRONIZATION
A resistor and decoupling capacitor are connected between
FREQ and SGND to program the switching frequency be-
tween 200 kHz to 1 MHz. These components must be sup-
plied on each controller, even if the system is synchronized
to an external clock.
The switching frequency and synchronization are controlled
by the Master. The Master can switch in a free-running mode
or be synchronized to an external clock. To synchronize the
Master apply the external clock to the SYNC pin of the Master,
otherwise ground this pin. The amplitude of the signal on the
SYNC pin must be limited to be between 0V and VCC.
The value of the frequency setting resistor is determined as:
A 1000 pF ceramic capacitor is used to provide sufficient de-
coupling. If the Master is synchronized set the resistor ac-
cording the nominal applied frequency. If the signal on the
SYNC pin is below 150 kHz the signal will be ignored and the
device will revert to free-running mode. The SYNCOUT signal
from the Master is applied to the first Slave’s SYNC pin. The
SYNCOUT pin of the first Slave is connected to the SYNC pin
of the second Slave, and so on, in a daisy chain configuration.
SYNCOUT of the last Slave (or the Master in a single con-
troller system) is left unconnected.
The configuration of the system, namely the number of con-
trollers and phases is programmed by the voltage on the PH
pin. For each controller connect the midpoint of a resistor di-
vider between VCC and SGND to the PH pin. The division
ratios are given in the Electrical Characteristics table and
nominal resistor values in Table 1. This sets the phase shift
between SYNC and the SYNCOUT pin. Where an even num-
ber of phases (N) are employed, the phase delay from SYNC
17 www.ti.com
LM3753
to SYNCOUT is 360°/N. The phase difference between the
two phases on the same controller is 180°. For systems with
an odd number of the phases, the HG2 and LG2 gate drivers
on the last Slave are unconnected and the phase arrange-
ment is set according to Table 1
DUTY CYCLE LIMITATION
The minimum controllable on-time is typically 50 ns. This lim-
its the maximum VIN , VOUT and fSW combination.
fSW < (VOUT / VIN) x 20 MHz
The maximum guaranteed duty cycle is 81%. This limits the
minimum VIN to VOUT ratio.
(VOUT / VIN) x 1.25 < 0.81
The 1.25 term allows margin for efficiency and transient re-
sponse.
THERMAL SHUTDOWN
The internal thermal shutdown circuit causes the PWM con-
trol circuitry to be reset and the NFET drivers to turn off all
external power MOSFETs. The controller remains enabled
and all bias circuitry remains on. After the die temperature
falls below the lower hysteresis point, the controller will
restart.
NFET SYNCHRONOUS DRIVERS
The LM3753 has two sets of gate drivers designed for driving
N-channel MOSFETs in a synchronous mode. Power to the
high-side driver is supplied through the BOOT pin. For the
high-side gate HG to turn on the high-side FET, the BOOT
voltage must be at least one VGS greater than VIN. This volt-
age is supplied from a local charge pump which consists of a
Schottky diode and bootstrap capacitor, shown in Figure 6.
For the Schottky, a rating of at least 250 mA and 30V is rec-
ommended. A dual package may be used to supply both
BOOT1 and BOOT2 for each controller.
Both the bootstrap and the low-side FET driver are fed from
VDD. The drive voltage for the top FET driver is about VDD
− 0.5V at light load condition and about VDD at normal to full
load condition.
30091926
FIGURE 6. Bootstrap Circuit
REMOTE SENSE DIFFERENTIAL AMPLIFIER
The differential amplifier connected internally to the SNSP,
SNSM and VDIF pins is a single stage unity gain Instrumen-
tation amplifier. The differential gain is tightly controlled to
within 0.4%.
30091927
FIGURE 7. Differential Amplifier
On the master controller, the differential amplifier is used to
provide Kelvin sensing of the output voltage at the load. This
provides the most accurate sampling for load regulation.
On the slave controllers, the differential amplifier is used to
sense the COMP signal of the master controller with respect
to its signal ground and drive the COMP pin of that slave con-
troller relative to its local signal ground. This allows the master
controller to accurately provide the target duty cycle of the
slave controllers.
The differential amplifier has a low output impedance to allow
it to drive the COMP pins of the Slave controllers. This is nec-
essary because the current sense signal is internally added
to COMP to provide the duty cycle adjustment for phase-to-
phase current sharing.
Application Information
NUMBER of PHASES
The number of phases can be calculated by dividing the max-
imum output load current by 25A. Therefore a 120A load
requirement will need at least 5 phases, or 3 controllers. It
may be better to use 6 phases which will still require 3 con-
trollers, but will reduce the maximum current/phase to 20A.
Increasing the number of phases will also reduce the output
voltage ripple and the input capacitor requirements. Note that
the 25A/phase is dictated by external components and not by
the LM3753. After the number of phases has been chosen,
the PH pin on each controller should be programmed as dis-
cussed in the Functional Description under PHASE NUMBER
SELECTION. The same number of phases must be selected
for each controller.
POWERING OPTIONS
The power connections will be determined by the VIN range
and the availability of an external 5V rail. This is discussed in
detail in the Functional Description under POWER CONNEC-
TIONS. For 12V input systems, the use of an external 5V rail
to power the VDD bus can improve overall system efficiency.
MULTI-CONTROLLER SYSTEMS
For systems with more than 2 phases, there will be one con-
troller configured as the Master and from 1 to 5 controllers
configured as Slave.
The Master controller uses the differential amplifier to sense
the output voltage at the load point. It also provides the com-
mon COMP signal used by all controllers, provides the loop
compensation and synchronizes the system clock to an ex-
ternal clock if one is provided.
The SYNCOUT of the Master is connected to the SYNC input
of the first Slave controller.
The Slave controllers are configured by tying the FB input to
the VCC pin of that controller. Each Slave uses the differential
www.ti.com 18
LM3753
amplifier to sense the COMP signal of the Master controller
and drive its own COMP input. The SYNCOUT of each Slave
controller is connected to the SYNC input of the next Slave
controller.
All controllers have the same parallel RC components con-
nected from the FREQ pin to local ground corresponding to
the desired system clock even if synchronizing to an external
clock.
Common connections for all controllers:
1) IAVE (each controller will have a parallel RC filter to local
ground).
2) FAULT
3) EN
4) TRACK
5) PGOOD
TRACKING
The LM3753 will track the output of an external power supply
by connecting a resistor divider to the TRACK pin as shown
in Figure 8. This allows the output voltage slew rate to be
controlled for loads that require precise sequencing.
A value of 10 kΩ 1% is recommended for RT1 as a good com-
promise between high precision and low quiescent current
through the divider. Note that the TRACK pin must finish at
least 100 mV higher than the 0.6V reference to achieve the
full accuracy of the LM3753 regulation. To meet this require-
ment the tracking voltage is offset by 150 mV. The output
voltage will reach its final value at 80% of the external supply
voltage. The tracking resistors are determined by:
30091929
FIGURE 8. Tracking an External Supply
30091930
FIGURE 9. Tracking an External Supply
For equal slew rates, the relationship for the tracking divider
is set by:
30091932
FIGURE 10. Tracking an External Supply with Equal Slew
Rates
In order to track properly, the external power supply voltage
must be higher than the LM3753 output voltage.
External Components Selection
The following is a design example selecting components for
the Typical Application Schematic of Figure 20. The circuit is
designed for two controller 4-phase operation with 1.2V out
at 100A from an input voltage of 6V to 18V. The expected load
is a microprocessor or ASIC with fast load transients, and the
type of MOSFETs used are in SO-8 or its equivalent packages
such as PowerPAK ®, PQFN and LFPAK (LFPAK-i).
SWITCHING FREQUENCY
The selection of switching frequency is based on the tradeoff
between size, cost, and efficiency. In general, a lower fre-
quency means larger, more expensive inductors and capac-
itors will be needed. A higher switching frequency generally
results in a smaller but less efficient solution. For this appli-
cation a frequency of 300 kHz was selected as a good com-
promise between the size of the inductor and MOSFETs,
transient response and efficiency. Following the equation giv-
en for RFRQ in the Functional Description under OSCILLATOR
and SYNCHRONIZATION, for 300 kHz operation a 78.7 kΩ
1% resistor is used for RFRQ. A 1000 pF capacitor is used for
CFRQ.
OUTPUT INDUCTORS
The first criterion for selecting an output inductor is the induc-
tance itself. In most buck converters, this value is based on
the desired peak-to-peak ripple current, ΔIL that flows in the
inductor along with the load current. As with switching fre-
quency, the selection of the inductor is a tradeoff between size
and cost. Higher inductance means lower ripple current and
hence lower output voltage ripple. Lower inductance results
in smaller, less expensive devices. An inductance that gives
a ripple current of 1/5 to 2/5 of the maximum output current is
a good starting point. (ΔIL = (1/5 to 2/5) x IOUT). Minimum in-
ductance is calculated from this value, using the maximum
input voltage as:
19 www.ti.com
LM3753
By calculating in terms of amperes, volts, and megahertz, the
inductance value will come out in micro henries. The inductor
ripple current is found from the minimum inductance equation:
The second criterion is inductor saturation current rating. The
LM3753 has an accurately programmed peak current limit.
During an output short circuit, the inductor should be chosen
so as not to exceed its saturation rating at elevated temper-
ature. For the design example, a standard value of 440 nH is
chosen to fall within the ΔIL = (1/5 to 2/5) x IOUT range.
The dc loss in the inductor is determined by its series resis-
tance RL. The dc power dissipation is found from:
PDC = IOUT2 x RL
The ac loss can be estimated from the inductor
manufacturer’s data, if available. The ac loss is set by the
peak-to-peak ripple current ΔIL and the switching frequency
fSW.
OUTPUT CAPACITORS
The output capacitors filter the inductor ripple current and
provide a source of charge for transient load conditions. A
wide range of output capacitors may be used with the LM3753
that provides excellent performance. The best performance
is typically obtained using aluminum electrolytic, tantalum,
polymer, solid aluminum, organic or niobium type chemistries
in parallel with ceramic capacitors. The ceramic capacitors
provide extremely low impedance to reduce the output ripple
voltage and noise spikes, while the aluminum or other capac-
itors provide a larger bulk capacitance for transient loading.
When selecting the value for the output capacitors the two
performance characteristics to consider are the output volt-
age ripple and transient response. The output voltage ripple
for a single phase can be approximated as:
With all values normalized to a single phase, ΔVO (V) is the
peak to peak output voltage ripple, ΔIL (A) is the peak to peak
inductor ripple current, RC (Ω) is the equivalent series resis-
tance or ESR of the output capacitors, fSW (Hz) is the switch-
ing frequency, and CO (F) is the output capacitance. The
amount of output ripple that can be tolerated is application
specific. A general recommendation is to keep the output rip-
ple less than 1% of the rated output voltage. Figure 11 shows
the output voltage ripple for multi-phase operation.
30091936
FIGURE 11. Multi-Phase Output Voltage Ripple
Based on the normalized single phase ripple, the worst case
multi-phase output voltage ripple can be approximated as:
ΔVO(N) = ΔVO / N
Where N is the number of phases.
The output capacitor selection will also affect the output volt-
age droop and overshoot during a load transient. The peak
transient of the output voltage during a load current step is
dependent on many factors. Given sufficient control loop
bandwidth an approximation of the transient voltage can be
obtained from:
With all values normalized to a single phase, VP (V) is the
output voltage transient and ΔIO (A) is the load current step
change. CO (F) is the output capacitance, L (H) is the value
of the inductor and RC
(Ω) is the series resistance of the output
capacitor. VL (V) is the minimum inductor voltage, which is
duty cycle dependent.
For D < 0.5, VL = VOUT
For D > 0.5, VL = VIN − VOUT
This shows that as the input voltage approaches VOUT, the
transient droop will get worse. The recovery overshoot re-
mains fairly constant.
The loss associated with the output capacitor series resis-
tance can be estimated as:
Output Capacitor Design Procedure
For the design example VIN = 12V, VOUT = 1.2V, D = VOUT /
VIN = 0.1, L = 440 nH, ΔIL = 9A, ΔIO = 20A and VP = 0.12V.
To meet the transient voltage specification, the maximum
RC is:
For the design example, the maximum RC is 6 mΩ. Choose
RC = 3 mΩ as the design limit.
www.ti.com 20
LM3753
From the equation for VP, the minimum value of CO is:
For D < 0.5, VL = VOUT
For D > 0.5, VL = VIN − VOUT
With RC = VP / ΔIO this reduces to:
With RC = 0 this reduces to:
Since D < 0.5, VL = VOUT. With RC = 3 mΩ, the minimum value
for CO is 476 μF.
The minimum control loop bandwidth fC is given by:
For the design example, the minimum value for fC is 44 kHz.
Two 220 μF, 5 mΩ polymer capacitors in parallel with two 22
μF, 3 mΩ ceramics per phase will meet the target output volt-
age ripple and transient specification.
INPUT CAPACITORS
The input capacitors for a buck regulator are used to smooth
the large current pulses drawn by the inductor and load when
the high-side MOSFET is on. Due to this large ac stress, input
capacitors are usually selected on the basis of their ac rms
current rating rather than bulk capacitance. Low ESR is ben-
eficial because it reduces the power dissipation in the capac-
itors. Although any of the capacitor types mentioned in the
OUTPUT CAPACITORS section can be used, ceramic ca-
pacitors are common because of their low series resistance.
In general the input to a buck converter does not require as
much bulk capacitance as the output.
The input capacitors should be selected for rms current rating
and minimum ripple voltage. The equation for the rms current
and power loss of the input capacitor in a single phase can
be estimated as:
Where IO (A) is the output load current and RCIN (Ω) is the
series resistance of the input capacitor. Since the maximum
values occur at D = 0.5, a good estimate of the input capacitor
rms current rating in a single phase is one-half of the maxi-
mum output current.
Neglecting the series inductance of the input capacitance, the
input voltage ripple for a single phase can be estimated as:
By defining the maximum input voltage ripple, the minimum
requirement for the input capacitance can be calculated as:
For multi-phase operation, the general equation for the input
capacitor rms current is approximated as:
This is valid for D < 1 / N and repeats for a total of N times.
IO represents the total output current and N is the number of
phases. Figure 12 shows the input capacitor rms current as
a function of the output current, duty cycle and number of
phases.
30091948
FIGURE 12. Input Capacitor RMS Current as a Function
of Output Current
For multi-phase operation the maximum rms current can be
approximated as:
ICIN(RMS)MAX ≈ 0.5 x IO / N
In most applications for point-of-load power supplies, the in-
put voltage is the output of another switching converter. This
output often has a lot of bulk capacitance, which may provide
adequate damping.
When the converter is connected to a remote input power
source through a wiring harness, a resonant circuit is formed
by the line impedance and the input capacitors. If step input
voltage transients are expected near the maximum rating of
the LM3753, a careful evaluation of the ringing and possible
overshoot at the device VIN pin should be completed. To
minimize overshoot make CIN > 10 x LIN. The characteristic
source impedance and resonant frequency are:
21 www.ti.com
LM3753
The converter exhibits a negative input impedance which is
lowest at the minimum input voltage:
The damping factor for the input filter is given by:
Where RLIN is the input wiring resistance and RCIN is the series
resistance of the input capacitors. The term ZS / ZIN will always
be negative due to ZIN.
When δ = 1, the input filter is critically damped. This may be
difficult to achieve with practical component values. With δ <
0.2, the input filter will exhibit significant ringing. If δ is zero or
negative, there is not enough resistance in the circuit and the
input filter will sustain an oscillation.
When operating near the minimum input voltage, an alu-
minum electrolytic capacitor across CIN may be needed to
damp the input for a typical bench test setup. Any parallel
capacitor should be evaluated for its rms current rating. The
current will split between the ceramic and aluminum capaci-
tors based on the relative impedance at the switching fre-
quency. Using a square wave approximation, the rms current
in each capacitor is found from:
Input Capacitor Design Procedure
Ceramic capacitors are sized to support the required rms cur-
rent. An aluminum electrolytic capacitor is used for damping.
Find the minimum value for the ceramic capacitors from:
Allowing ΔVIN = 0.6V for the design example, the minimum
value is CIN = 34.7 μF. Find the rms current rating from:
ICIN(RMS)MAX ≈ 0.5 x IO / N
Using the same criteria, the result is 12.5A rms. Manufacturer
data for 4.7 μF, 25V, X7R capacitors in a 1210 package allows
for 4A rms with a 20°C temperature rise. For the design ex-
ample, using two ceramic capacitors for each phase will meet
both the input voltage ripple and rms current target. Since the
series resistance is so low at about 4 mΩ per capacitor, a
parallel aluminum electrolytic is used for damping. A good
general rule is to make the damping capacitor at least five
times the value of the ceramic. By sizing the aluminum such
that it is primarily resistive at the switching frequency, the de-
sign is greatly simplified since the ceramic capacitors are
primarily reactive. In this case the approximation for the rms
current in the damping capacitor is:
Where CIN2 is the damping capacitance, RCIN2 is its series
resistance and CIN1 is the ceramic capacitance. A 470 μF,
25V, 0.06Ω, 1.19A rms aluminum electrolytic capacitor in a
10 mm x 10.2 mm package is chosen for the damping capac-
itor. Calculated rms current for the aluminum electrolytic is
0.67A.
MOSFETS
Selection of the power MOSFETs is governed by a tradeoff
between cost, size and efficiency.
Losses in the high-side FET can be broken down into con-
duction loss, gate charge loss and switching loss. Conduction
or I2R loss is approximately:
PCOND_HI = D x (IOUT2 x RDS(on)_HI x 1.3)
(High-side FET)
PCOND_LO = (1 − D) x (IOUT2 x RDS(on)_LO x 1.3)
(Low-side FET)
In the above equations the factor 1.3 accounts for the in-
crease in MOSFET RDS(on) due to self heating. Alternatively,
the 1.3 can be ignored and the RDS(on) of the MOSFET esti-
mated using the RDS(on) vs. Temperature curves in the MOS-
FET datasheets.
The gate charge loss results from the current driving the gate
capacitance of the power MOSFETs, and is approximated as:
PDR = VIN x (QG_HI + QG_LO) x fSW
Where QG_HI and QG_LO are the total gate charge of the high-
side and low-side FETs respectively at the typical 5V driver
voltage. Gate charge loss differs from conduction and switch-
ing losses in that the majority of dissipation occurs in the
LM3753 and VDD regulator.
The switching loss occurs during the brief transition period as
the FET turns on and off, during which both current and volt-
age is present in the channel of the FET. This can be approx-
imated as:
Where QGD is the high-side FET Miller charge with a VDS
swing between 0 to VIN; CISS is the input capacitance of the
high-side MOSFET in its off state with VDS = VIN. α and β are
fitting coefficient numbers, which are usually between 0.5 to
1, depending on the board level parasitic inductances and re-
verse recovery of the low-side power MOSFET body diode.
Under ideal condition, setting α = β = 0.5 is a good starting
point. Other variables are defined as:
www.ti.com 22
LM3753
IL_VL = IOUT − 0.5 x ΔIL
IL_PK = IOUT + 0.5 x ΔIL
RG_ON = 5 + RG_INT + RG_EXT
RG_OFF = 2 + RG_INT + RG_EXT
Switching loss is calculated for the high-side FET only. 5 and
2 represent the LM3753 high-side driver resistance in the
transient region. RG_INT is the gate resistance of the high-side
FET, and RG_EXT is the extra external gate resistance if ap-
plicable. RG_EXT may be used to damp out excessive parasitic
ringing at the switch node.
For this example, the maximum drain-to-source voltage ap-
plied to either MOSFET is 18V. The maximum drive voltage
at the gate of the high-side MOSFET is 5V, and the maximum
drive voltage for the low-side MOSFET is 5V. The selected
MOSFET must be able to withstand 18V plus any ringing from
drain to source, and be able to handle at least 5V plus ringing
from gate to source. If the duty cycle of the converter is small,
then the high-side MOSFET should be selected with a low
gate charge in order to minimize switching loss whereas the
bottom MOSFET should have a low RDS(on) to minimize con-
duction loss.
For a typical input voltage of 12V and output current of 25A
per phase, the MOSFET selections for the design example
are SIR850DP for the high-side MOSFET and 2 x SIR892DP
for the low-side MOSFET.
A 2.2Ω resistor for the high-side gate drive may be added in
series with the HG output. This helps to control the MOSFET
turn-on and ringing at the switch node. Additionally, 0.5A
Schottky diodes may be placed across the high-side MOS-
FETs. The external Schottky diodes have a much faster re-
covery characteristic than the MOSFET body diode, and help
to minimize switching spikes by clamping the SW pin to VIN.
Another technique to control ringing at the switch node is to
place an RC snubber from SW to PGND directly across the
low-side MOSFET. Typical values at 300 kHz are 1Ω and 680
pF.
To improve efficiency, 3A Schottky diodes may be placed
across the low-side MOSFETs. The external Schottky diodes
have a much lower forward voltage than the MOSFET body
diode, and help to minimize the loss due to the body diode
recovery characteristic.
EN and VIN UVLO
For operation at 6V minimum input, set the EN divider to en-
able the LM3753 at approximately 5.5V nominal. Values of
RUV1 = 1.37 kΩ and RUV2 = 4.02 kΩ will meet the target
threshold.
CURRENT SENSE
For resistor current sense, a 1 mΩ 1W resistor is used for a
full scale voltage of 25 mV at 25A out.
For DCR sensing, RS is equal to the inductor resistance of
RL = 0.32 mΩ plus an estimated trace resistance of 0.2 mΩ..
The full scale voltage is about 13 mV at 25A. For equal time
constants, the relationship of the integrating RC is determined
by:
Choosing CDCR = 0.15 μF:
RDCR = 440 nH / (0.15 μF x 0.52 mΩ) = 5.64 kΩ.
Using a standard value of 5.90 kΩ, the average current
through RDCR is calculated as 203 μA from:
IDCR = VOUT / RDCR
IDCR is sufficiently high enough to keep the CS input bias cur-
rent from being a significant error term.
CURRENT LIMIT
For the design example, the desired current limit set point is
chosen as 34.5A peak per phase, which is about 25% above
the full load peak value. Using DCR sense with RS = 0.52
mΩ:
RILIM = 34.5A x 0.52 mΩ / 94 μA = 191Ω
For resistor sense, the relatively low output inductor value
forms a voltage divider with the intrinsic inductance of the
sense resistor. When the MOSFETs switch, this adds a step
to the otherwise triangular current sense voltage. The step
voltage is simply the input voltage times the inductive divider.
With L = 440 nH and LS = 1 nH, the step voltage is:
VLS = 12V x 1 nH / 441 nH = 27.2 mV
Using the same method as DCR sense, an RC filter is added
to recover the actual resistive sense voltage. Choosing C = 1
nF the resistor is calculated as:
R = 1 nH / (1 nF x 1 mΩ) = 1 kΩ
The current limit resistor is then calculated as:
RILIM = 34.5A x 1 mΩ / 94 μA = 367Ω
The closest standard value of 365Ω 1% is selected for the
design example.
TRACK
For the design example, an external voltage of 3.3V is used
as the controlling voltage. The divider values are set so that
both voltages will rise together, with VEXT reaching its final
value just before VOUT. Following the method in the Applica-
tion Information under TRACKING and allowing for a 120 mV
offset between FB and TRACK, standard 1% values are se-
lected for RT1 = 10 kΩ and RT2 = 35.7 kΩ.
VCC, VDD and BOOT
VCC is used as the supply for the internal control and logic
circuitry. A 4.7 μF ceramic capacitor provides sufficient filter-
ing for VCC.
CVDD provides power for both the high-side and low-side
MOSGET gate drives, and is sized to meet the total gate drive
current. Allowing for ΔVVDD = 100 mV of ripple, the minimum
value for CVDD is found from:
Using QG_HI = 2 x 10 nC and QG_LO = 4 x 21 nC per controller
with a 5V gate drive, the minimum value for CVDD = 1.04 μF.
To use common component values, CVDD1 and CVDD2 are also
selected as 4.7 μF ceramic.
23 www.ti.com
LM3753
A general purpose NPN transistor is sized to meet the re-
quirements for the VDD supply. Based on the gate charge of
104 nC per controller, the required current is found from:
IGC = QG_TOTAL x fSW
At 300 kHz, IGC = 31.2 mA per controller. For a two controller
system, the minimum HFE for the transistor is determined by:
HFEMIN = IGC_TOTAL / 5 mA
The power dissipated by the transistor is:
PR = (VIN − VDD) x IGC_TOTAL
The transistor must support 62.4 mA with an HFE of at least
12.5 over the entire operating range. At 18V in the power dis-
sipated is 0.8W. A CJD44H11 in a DPAK case is chosen for
the design example. A 0.047 μF capacitor from base to PGND
will improve the transient performance of the VDD supply.
CBOOT provides power for the high-side gate drive, and is
sized to meet the required gate drive current. Allowing for
ΔVBOOT = 100 mV of ripple, the minimum value for CBOOT is
found from:
Using QG_HI = 10 nC per phase with a 5V gate drive, the min-
imum value for CBOOT = 0.1 μF. CBOOT is selected as 0.22 μF
ceramic per phase for the design example. A 0.5A Schottky
diode is used for DBOOT at each controller.
PRE-LOAD RESISTOR
For normal operation, a pre-load resistor is generally not re-
quired. During an abnormal fault condition with the output
completely disconnected from the load, the output voltage
may rise. This is primarily due to the high-side driver off-state
bias current, and reverse leakage current of the high-side
Schottky clamp diode.
At room temperature with 12V input, the reverse leakage of
each 0.5A Schottky diode is about 15 μA. With the EN pin high
and the FAULT pin low, the bias current in each high-side
driver is about 105 μA. Allowing for a 2 to 1 variation, the
maximum value of resistor to keep the output voltage from
rising above 5% of its nominal value is found from:
R = 0.05 x 1.2V / 330 µA = 182Ω
A value of 120Ω is selected for the design example. This rep-
resents a 10 mA pre-load at the rated output voltage, which
is 0.01% of the 100A full load current.
CONTROL LOOP COMPENSATION
The LM3753 uses voltage-mode PWM control to correct
changes in output voltage due to line and load transients. In-
put voltage feed-forward is used to adjust the amplitude of the
PWM ramp. This stabilizes the modulator gain from variations
due to input voltage, providing a robust design solution. A fast
inner current sharing circuit ensures good dynamic response
to changes in load current.
The control loop is comprised of two parts. The first is the
power stage, which consists of the duty cycle modulator, cur-
rent sharing circuit, output filter and load. The second part is
the error amplifier, which is a voltage type operational ampli-
fier with a typical dc gain of 70 dB and a unity gain frequency
of 15 MHz. Figure 13 shows the power stage, error amplifier
and current sharing components.
30091962
FIGURE 13. Power Stage, Error Amplifier and Current Sharing
www.ti.com 24
LM3753
The simplified power stage transfer function (also called the
control-to-output transfer function) for the LM3753 can be
written as:
Where:
With:
Km is the dc modulator gain and Ri is the current-sharing gain.
KFF is the input voltage feed-forward term, which is internally
set to a value of 0.232 V/V. The IAVE filter is accounted for
by Ha(s), which provides additional damping of the modulator
transfer function.
RAV sets the gain of the current averaging amplifier. A fixed
value of 8 kΩ/phase must be used for proper scaling. Since
the effective resistance is in parallel, each LM3753 should
have a 4.02 kΩ 1% resistor at IAVE for 2-phase/controller op-
eration. CAV sets the IAVE filter time constant of the current
sharing amplifier. For optimal performance of the current
sharing circuit, the IAVE filter is designed to settle to its final
value in five switching cycles. The optimal IAVE time constant
is defined as:
T = CAV x RAV
A value of CAV = 1/(RAV x fSW) per phase must be used for the
optimal time constant. Each LM3753 should have a value of
two times the normalized single phase value of CAV at IAVE
for 2-phase/controller operation. In this manner, the IAVE
time constant maintains a fixed value of T for any number of
phases.
Typical frequency response of the gain and the phase for the
power stage are shown in Figure 14 and Figure 15. It is de-
signed for VIN = 12V, VOUT = 1.2V, IOUT = 25A per phase and
a switching frequency of 300 kHz. For 2-phase operation
RAV = 4.02 kΩ and CAV = 1000 pF. The power stage compo-
nent values per phase are:
L = 0.44 μH, RL = 0.52 mΩ, CO1 = 440 μF, RC1 = 2.5 mΩ,
CO2 = 44 μF, RC2 = 1.5 mΩ, RS = RL = 0.52 mΩ and RO =
VOUT / IOUT = 48 mΩ.
30091966
FIGURE 14. Power Stage Gain
30091967
FIGURE 15. Power Stage Phase
Assuming a pole at the origin, the simplified equation for the
error amplifier transfer function can be written in terms of the
mid-band gain as:
Where:
25 www.ti.com
LM3753
In general, the goal of the compensation circuit is to give high
gain, a bandwidth that is between one-fifth and one-tenth of
the switching frequency, and at least 45° of phase margin.
Control Loop Design Procedure
Once the power stage design is complete, the power stage
components are used to determine the proper frequency
compensation. Knowing the dc modulator gain and assuming
an ideal single-pole system response, the mid-band error am-
plifier gain is set by the target crossover frequency. Based on
the ideal amplifier transfer function, the zero-pair is set to
cancel the complex conjugate pole of the output filter. One
pole is set to cancel the ESR of the output capacitor. The
second pole is set equal to the switching frequency. A cor-
rection factor is used to accommodate the modulator damping
when the output filter pole is within a decade of the target
crossover frequency.
The compensation components will scale from the feedback
divider ratio and selection of the bottom feedback divider re-
sistor. A maximum value for the divider current is typically set
at 1 mA. Using a divider current of 200 μA will allow for a
reasonable range of values. For the bottom feedback resistor
RFBB = VREF / 200 μA = 3 kΩ. Choosing a standard 1% value
of 3.01 kΩ, the top feedback resistor is found from:
For VOUT = 1.2V and VREF = 0.6V, RFBT = 3.01 kΩ.
Based on the previously defined power stage values, calcu-
late general terms:
For the design example D = 0.1, Ri = 0.026Ω, T = 3.33 μs and
Km = 3.22.
Calculate the output filter pole frequency and the ESR zero
frequency from:
For the output filter pole using CO = CO1 + CO2, ωP = 68.5 krad/
sec. Since CO1 >> CO2, the ESR zero is calculated using
CO1 and RC1 as ωZ = 909 krad/sec.
Choose a target crossover frequency fC greater than the min-
imum control loop bandwidth from the OUTPUT CAPACI-
TORS section. The optimum value of the crossover frequency
is usually between 5 and 10 times the filter pole frequency.
With fP = ωP / (2 x π) = 10.9 kHz, this places fC between 54.5
kHz and 109 kHz. The upper limit for fC is typically set at 1/5
of the switching frequency.
Choosing fC = 60 kHz for the design example ωC = 377 krad/
sec. The switching frequency is ωSW = 1.88 Mrad/sec.
For output capacitors with very low ESR, if the target
crossover frequency is more than 10 times the filter pole fre-
quency, bandwidth limiting of the error amplifier may occur.
See the Comprehensive Equations section to incorporate the
error amplifier bandwidth into the design procedure.
For reference, the parallel equivalent CO and RC at any fre-
quency can be calculated from:
At the target crossover frequency X1 = 0.00603, X2 = 0.0603,
Z = 0.00592 and A = 1.213. The parallel equivalent CO = 478
μF and RC = 2.1 mΩ.
Calculate the error amplifier gain coefficient and the compen-
sation component values. The (1 − ωP/ωC) term is the cor-
rection factor for the modulator damping.
For the design example, the calculated values are GC = 1.71,
CHF = 103 pF, CCOMP = 2236 pF, RCOMP = 6527Ω, RFF = 245
and CFF = 4483 pF.
Using standard values of CHF = 100 pF, CCOMP = 2200 pF,
RCOMP = 6.2 kΩ, RFF = 240Ω and CFF = 4700 pF, the error
amplifier plots of gain and phase are shown in Figure 16 and
Figure 17.
www.ti.com 26
LM3753
30091976
FIGURE 16. Error Amplifier Gain
30091977
FIGURE 17. Error Amplifier Phase
The complete control loop transfer function is equal to the
product of the power stage transfer function and error ampli-
fier transfer function. For the Bode plots, the overall loop gain
is the equal to the sum in dB and the overall phase is equal
to the sum in degrees. Results are shown in Figure 18 and
Figure 19. The crossover frequency is 57 kHz with a phase
margin of 73°.
30091978
FIGURE 18. Control Loop Gain
30091979
FIGURE 19. Control Loop Phase
For the small-signal analysis, it is assumed that the control
voltage at the COMP pin is dc. In practice, the output ripple
voltage is amplified by the error amplifier gain at the switching
frequency, which appears at the COMP pin adding to the
control ramp. This tends to reduce the modulator gain, which
may lower the actual control loop crossover frequency. This
effect is greatly reduced as the number of phases is in-
creased.
27 www.ti.com
LM3753
Efficiency and Thermal
Considerations
The buck regulator steps down the input voltage and has a
duty ratio D of:
Where η is the estimated converter efficiency. The efficiency
is defined as:
The total power dissipated in the power components can be
obtained by adding together the loss as mentioned in the
OUTPUT INDUCTORS, OUTPUT CAPACITORS, INPUT
CAPACITORS and MOSFETS sections.
The highest power dissipating components are the power
MOSFETs. The easiest way to determine the power dissipat-
ed in the MOSFETs is to measure the total conversion loss
(PIN − POUT), then subtract the power loss in the capacitors,
inductors, LM3753 and VDD regulator. The resulting power
loss is primarily in the switching MOSFETs. Selecting MOS-
FETs with exposed pads will aid the power dissipation of
these devices. Careful attention to RDS(on) at high temperature
should be observed.
If a snubber is used, the power loss can be estimated with an
oscilloscope by observation of the resistor voltage drop at
both the turn-on and turn-off transitions. Assuming that the
RC time constant is << 1 / fSW:
P = ½ x C x (VP2 + VN2) x fSW
VP and VN represent the positive and negative peak voltage
across the snubber resistor, which is ideally equal to VIN.
LM3753 and VDD REGULATOR OPERATING LOSS
These terms accounts for the currents drawn at the VIN and
VDD pins, used for driving the logic circuitry and the power
MOSFETs. For the LM3753, the VIN current is equal to the
steady state operating current IVIN. The VDD current is pri-
marily determined by the MOSFET gate charge current IGC,
which is defined as:
IGC = QG_TOTAL x fSW
PD = (VIN x IVIN) + (VDD x IGC)
QG_TOTAL is the total gate charge of the MOSFETs connected
to each LM3753. PD represents the total power dissipated in
each LM3753. IVIN is about 15 mA from the Electrical Char-
acteristics table. The LM3753 has an exposed thermal pad to
aid power dissipation.
The power dissipated in the VDD regulator is determined by:
PR = (VIN − VDD) x IGC_TOTAL
IGC_TOTAL is the sum of the MOSFET gate charge currents for
all of the controllers.
Layout Considerations
To produce an optimal power solution with a switching con-
verter, as much care must be taken with the layout and design
of the printed circuit board as with the component selection.
The following are several guidelines to aid in creating a good
layout.
KELVIN TRACES for GATE DRIVE and SENSE LINES
The HG and SW pins provide the gate drive and return for the
high-side MOSFETs. These lines should run as parallel pairs
to each MOSFET, being connected as close as possible to
the respective MOSFET gate and source. Likewise the LG
and PGND pins provide the gate drive and return for the low-
side MOSFETs. A good ground plane between the PGND pin
and the low-side MOSFETs source connections is needed to
carry the return current for the low-side gates.
The SNSP and SNSM pins of the Master should be connected
as a parallel pair, running from the output power and ground
sense points. Keep these lines away from the switch node
and output inductor to avoid stray coupling. If possible, the
SNSP and SNSM traces should be shielded from the switch
node by ground planes.
SGND and PGND CONNECTIONS
Good layout techniques include a dedicated ground plane,
usually on an internal layer adjacent to the LM3753 and signal
component side of the board. Signal level components con-
nected to FB, TRACK, FREQ, IAVE, EN and PH along with
the VCC and VIN bypass capacitors should be tied directly to
the SGND pin. Connect the SGND and PGND pins directly to
the DAP, with vias from the DAP to the ground plane. The
ground plane is then connected to the input capacitors and
low-side MOSFET source at each phase.
MINIMIZE the SWITCH NODE
The copper area that connects the power MOSFETs and out-
put inductor together radiates more EMI as it gets larger. Use
just enough copper to give low impedance for the switching
currents and provide adequate heat spreading for the MOS-
FETs.
LOW IMPEDANCE POWER PATH
In a buck regulator the primary switching loop consists of the
input capacitor connection to the MOSFETs. Minimizing the
area of this loop reduces the stray inductance, which mini-
mizes noise and possible erratic operation. The ceramic input
capacitors at each phase should be placed as close as pos-
sible to the MOSFETs, with the VIN side of the capacitors
connected directly to the high-side MOSFET drain, and the
PGND side of the capacitors connected as close as possible
to the low-side source. The complete power path includes the
input capacitors, power MOSFETs, output inductor, and out-
put capacitors. Keep these components on the same side of
the board and connect them with thick traces or copper
planes. Avoid connecting these components through vias
whenever possible, as vias add inductance and resistance. In
general, the power components should be kept close togeth-
er, minimizing the circuit board losses.
www.ti.com 28
LM3753
Comprehensive Equations
POWER STAGE TRANSFER FUNCTION
To include all terms, it is easiest to use the impedance form
of the equation:
Where:
With:
ERROR AMPLIFIER TRANSFER FUNCTION
Using a single-pole operational amplifier model, the complete
error amplifier transfer function is given by:
Where the open loop gain AOL = 3162 (70 dB) and the unity
gain bandwidth ωBW = 2 x π x fBW.
The ideal transfer function is expressed in terms of the mid-
band gain as:
The feedback gain is then:
Where:
ERROR AMPLIFIER BANDWIDTH LIMIT
When the ideal error amplifier gain reaches the open loop
gain-bandwidth limit, the phase goes to zero. To incorporate
the amplifier bandwidth into the design procedure, determine
the boundary limit with respect to the ESR zero frequency:
Based on the relative ESR zero, the crossover frequency is
set at 1/3 of the bandwidth limiting frequency.
If ωZ > ωZB, calculate the optimal crossover frequency from:
If ωZ < ωZB, calculate the optimal crossover frequency from:
Using this method, the maximum phase boost is achieved at
the optimal crossover frequency.
In either case, the upper limit for fC is typically set at 1/5 of the
switching frequency.
29 www.ti.com
LM3753
Typical Application
30091902
All controllers in the system are the same part. The Master and Slave are differentiated by how they are connected in the system.
FIGURE 20. Typical Application
www.ti.com 30
LM3753
Design Examples
30091989
FIGURE 21. Master with DCR Sense
31 www.ti.com
LM3753
30091990
FIGURE 22. Slave with DCR Sense
www.ti.com 32
LM3753
30091991
FIGURE 23. Master with Resistor Sense
33 www.ti.com
LM3753
30091992
FIGURE 24. Slave with Resistor Sense
www.ti.com 34
LM3753
Physical Dimensions inches (millimeters) unless otherwise noted
32-Lead LLP Package
NS Package Number SQA32A
35 www.ti.com
LM3753
Notes
LM3753 Scalable 2-Phase Synchronous Buck Controller with Integrated FET Drivers and Linear
Regulator Controller
www.ti.com
IMPORTANT NOTICE
Texas Instruments Incorporated and its subsidiaries (TI) reserve the right to make corrections, modifications, enhancements, improvements,
and other changes to its products and services at any time and to discontinue any product or service without notice. Customers should
obtain the latest relevant information before placing orders and should verify that such information is current and complete. All products are
sold subject to TI’s terms and conditions of sale supplied at the time of order acknowledgment.
TI warrants performance of its hardware products to the specifications applicable at the time of sale in accordance with TI’s standard
warranty. Testing and other quality control techniques are used to the extent TI deems necessary to support this warranty. Except where
mandated by government requirements, testing of all parameters of each product is not necessarily performed.
TI assumes no liability for applications assistance or customer product design. Customers are responsible for their products and
applications using TI components. To minimize the risks associated with customer products and applications, customers should provide
adequate design and operating safeguards.
TI does not warrant or represent that any license, either express or implied, is granted under any TI patent right, copyright, mask work right,
or other TI intellectual property right relating to any combination, machine, or process in which TI products or services are used. Information
published by TI regarding third-party products or services does not constitute a license from TI to use such products or services or a
warranty or endorsement thereof. Use of such information may require a license from a third party under the patents or other intellectual
property of the third party, or a license from TI under the patents or other intellectual property of TI.
Reproduction of TI information in TI data books or data sheets is permissible only if reproduction is without alteration and is accompanied
by all associated warranties, conditions, limitations, and notices. Reproduction of this information with alteration is an unfair and deceptive
business practice. TI is not responsible or liable for such altered documentation. Information of third parties may be subject to additional
restrictions.
Resale of TI products or services with statements different from or beyond the parameters stated by TI for that product or service voids all
express and any implied warranties for the associated TI product or service and is an unfair and deceptive business practice. TI is not
responsible or liable for any such statements.
TI products are not authorized for use in safety-critical applications (such as life support) where a failure of the TI product would reasonably
be expected to cause severe personal injury or death, unless officers of the parties have executed an agreement specifically governing
such use. Buyers represent that they have all necessary expertise in the safety and regulatory ramifications of their applications, and
acknowledge and agree that they are solely responsible for all legal, regulatory and safety-related requirements concerning their products
and any use of TI products in such safety-critical applications, notwithstanding any applications-related information or support that may be
provided by TI. Further, Buyers must fully indemnify TI and its representatives against any damages arising out of the use of TI products in
such safety-critical applications.
TI products are neither designed nor intended for use in military/aerospace applications or environments unless the TI products are
specifically designated by TI as military-grade or "enhanced plastic."Only products designated by TI as military-grade meet military
specifications. Buyers acknowledge and agree that any such use of TI products which TI has not designated as military-grade is solely at
the Buyer's risk, and that they are solely responsible for compliance with all legal and regulatory requirements in connection with such use.
TI products are neither designed nor intended for use in automotive applications or environments unless the specific TI products are
designated by TI as compliant with ISO/TS 16949 requirements. Buyers acknowledge and agree that, if they use any non-designated
products in automotive applications, TI will not be responsible for any failure to meet such requirements.
Following are URLs where you can obtain information on other Texas Instruments products and application solutions:
Products Applications
Audio www.ti.com/audio Automotive and Transportation www.ti.com/automotive
Amplifiers amplifier.ti.com Communications and Telecom www.ti.com/communications
Data Converters dataconverter.ti.com Computers and Peripherals www.ti.com/computers
DLP®Products www.dlp.com Consumer Electronics www.ti.com/consumer-apps
DSP dsp.ti.com Energy and Lighting www.ti.com/energy
Clocks and Timers www.ti.com/clocks Industrial www.ti.com/industrial
Interface interface.ti.com Medical www.ti.com/medical
Logic logic.ti.com Security www.ti.com/security
Power Mgmt power.ti.com Space, Avionics and Defense www.ti.com/space-avionics-defense
Microcontrollers microcontroller.ti.com Video and Imaging www.ti.com/video
RFID www.ti-rfid.com
OMAP Mobile Processors www.ti.com/omap
Wireless Connectivity www.ti.com/wirelessconnectivity
TI E2E Community Home Page e2e.ti.com
Mailing Address: Texas Instruments, Post Office Box 655303, Dallas, Texas 75265
Copyright ©2012, Texas Instruments Incorporated
