19-0666; Rev 0; 10/06

KIT
ATION
EVALU BL E
AVAILA

1.5MHz, 30A High-Efficiency, LED Driver

with Rapid LED Current Pulsing
General Description Features

MAX16818
The MAX16818 pulse-width modulation (PWM) LED dri- ♦ High-Current LED Driver Controller IC, Up to 30A
ver controller provides high-output-current capability in Output Current
a compact package with a minimum number of external ♦ Average-Current-Mode Control
components. The MAX16818 is suitable for use in syn- ♦ True-Differential Remote-Sense Input
chronous and nonsynchronous step-down (buck)
topologies, as well as in boost, buck-boost, SEPIC, and ♦ 4.75V to 5.5V or 7V to 28V Input Voltage Range
Cuk LED drivers. The MAX16818 is the first LED driver ♦ Programmable Switching Frequency or External
controller that enables Maxim’s patent-pending technol- Synchronization from 125kHz to 1.5MHz
ogy for fast LED current transients of up to 20A/µs and ♦ Clock Output for 180° Out-of-Phase Operation
30kHz dimming frequency.
♦ Integrated 4A Gate Drivers
This device utilizes average-current-mode control that ♦ Output Overvoltage and Hiccup Mode
enables optimal use of MOSFETs with optimal charge Overcurrent Protection
and on-resistance characteristics. This results in the
minimized need for external heatsinking even when ♦ Thermal Shutdown
delivering up to 30A of LED current. True differential ♦ Thermally Enhanced 28-Pin Thin QFN Package
sensing enables accurate control of the LED current. A ♦ -40°C to +125°C Operating Temperature Range
wide dimming range is easily implemented to accom-
modate an external PWM signal. An internal regulator
enables operation over a wide input voltage range:
4.75V to 5.5V or 7V to 28V and above with a simple Ordering Information
external biasing device. The wide switching frequency
range, up to 1.5MHz, allows for the use of small induc- PIN- PKG
PART TEMP RANGE
tors and capacitors. PACKAGE CODE
The MAX16818 features a clock output with 180° phase MAX16818ATI+ -40°C to +125°C 28 TQFN-EP* T2855-3
delay to control a second out-of-phase LED driver to MAX16818ETI+ -40°C to +85°C 28 TQFN-EP* T2855-3
reduce input and output filter capacitors size or to mini-
+Denotes lead-free package.
mize ripple currents. The MAX16818 offers programma-
ble hiccup, overvoltage, and overtemperature protection. *EP = Exposed paddle.

The MAX16818ETI+ is rated for the extended tempera-

ture range (-40°C to +85°C) and the MAX16818ATI+ is
rated for the automotive temperature range (-40°C to Simplified Diagram
+125°C). This LED driver controller is available in a
lead-free, 0.8mm high, 5mm x 5mm 28-pin TQFN pack- 7V TO 28V
age with exposed paddle.
C1
IN
Q1
Applications EN DH
VLED
Front Projectors/Rear Projection TVs ILIM L1
Portable and Pocket Projectors
MAX16818
Automotive, Bus/Truck Exterior Lighting Q2
DL
LCD TVs and Display Backlight C2
Q3
Automotive Emergency Lighting and Signage
OVI CSP
R1
CLP PGND

.
HIGH-FREQUENCY
PULSE TRAIN
Pin Configuration appears at end of data sheet. NOTE: MAXIM PATENT-PENDING TOPOLOGY

________________________________________________________________ Maxim Integrated Products 1

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.
 1.5MHz, 30A High-Efficiency, LED Driver
with Rapid LED Current Pulsing
ABSOLUTE MAXIMUM RATINGS
MAX16818

IN to SGND.............................................................-0.3V to +30V Continuous Power Dissipation (TA = +70°C)

BST to SGND..........................................................-0.3V to +35V 28-Pin TQFN (derate 34.5mW/°C above +70°C) .......2758mW
BST to LX..................................................................-0.3V to +6V Operating Temperature Range
DH to LX .......................................-0.3V to [(VBST - VLX_) + 0.3V] MAX16818ATI+..............................................-40°C to +125°C
DL to PGND................................................-0.3V to (VDD + 0.3V) MAX16818ETI+................................................-40°C to +85°C
VCC to SGND............................................................-0.3V to +6V Maximum Junction Temperature .....................................+150°C
VCC, VDD to PGND ...................................................-0.3V to +6V Storage Temperature Range .............................-60°C to +150°C
SGND to PGND .....................................................-0.3V to +0.3V Lead Temperature (soldering, 10s) .................................+300°C
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 = 5V, VDD = VCC, TA = TJ = TMIN to TMAX, unless otherwise noted. Typical specifications are at TA = +25°C.) (Note 1)

PARAMETER SYMBOL CONDITIONS MIN TYP MAX UNITS

SYSTEM SPECIFICATIONS
7 28
Input Voltage Range VIN Short IN and VCC together for 5V input V
4.75 5.50
operation
Quiescent Supply Current IQ EN = VCC or SGND, not switching 2.7 5.5 mA
LED CURRENT REGULATOR
SENSE+ to SENSE- Accuracy No load, VIN = 4.75V to 5.5V, fSW = 500kHz 0.594 0.6 0.606
V
(Note 2) No load, VIN = 7V to 28V, fSW = 500kHz 0.594 0.6 0.606
Clock
Soft-Start Time tSS 1024
Cycles
STARTUP/INTERNAL REGULATOR
VCC Undervoltage Lockout UVLO VCC rising 4.1 4.3 4.5 V
VCC Undervoltage Hysteresis 200 mV
VCC Output Voltage VIN = 7V to 28V, ISOURCE = 0 to 60mA 4.85 5.1 5.30 V
MOSFET DRIVERS
Output Driver Impedance RON Low or high output, ISOURCE/SINK = 20mA 1.1 3.0 Ω
Output Driver Source/Sink Current IDH,IDL 4 A
Nonoverlap Time tNO CDH/DL = 5nF 35 ns
OSCILLATOR
Switching Frequency Range 125 1500 kHz
Switching Frequency RT = 500kΩ 121 125 129
Switching Frequency fSW RT = 120kΩ 495 521 547 kHz
Switching Frequency RT = 39.9kΩ 1515 1620 1725
120kΩ ≤ RT ≤ 500kΩ -5 +5
Switching Frequency Accuracy %
40kΩ ≤ RT ≤ 120kΩ -8 +8

2 _______________________________________________________________________________________
 1.5MHz, 30A High-Efficiency, LED Driver
with Rapid LED Current Pulsing
ELECTRICAL CHARACTERISTICS (continued)

MAX16818
(VCC = 5V, VDD = VCC, TA = TJ = TMIN to TMAX, unless otherwise noted. Typical specifications are at TA = +25°C.) (Note 1)

PARAMETER SYMBOL CONDITIONS MIN TYP MAX UNITS

CLKOUT Phase Shift φ_CLKOUT With respect to DH, fSW = 125kHz 180 Degrees
CLKOUT Output Low Level VCLKOUTL ISINK = 2mA 0.4 V
CLKOUT Output High Level VCLKOUTH ISOURCE = 2mA 4.5 V
SYNC Input-High Pulse Width tSYNC 200 ns
SYNC Input Clock High Threshold VSYNCH 2.0 V
SYNC Input Clock Low Threshold VSYNCL 0.4 V
SYNC Pullup Current ISYNC_OUT VRT/SYNC = 0V 250 750 µA
SYNC Power-Off Level VSYNC_OFF 0.4 V
INDUCTOR CURRENT LIMIT
Average Current-Limit Threshold VCL CSP to CSN 24.0 26.9 28.2 mV
Reverse Current-Limit Threshold VCLR CSP to CSN -3.2 -2.3 -0.1 mV
Cycle-by-Cycle Current Limit CSP to CSN 60 mV
Cycle-by-Cycle Overload VCSP to VCSN = 75mV 260 ns
Hiccup Divider Ratio LIM to VCM, no switching 0.547 0.558 0.565 V/V
Hiccup Reset Delay 200 ms
LIM Input Impedance LIM to SGND 55.9 kΩ
CURRENT-SENSE AMPLIFIER
CSP or CSN Input Resistance RCS 4 kΩ
Common-Mode Range VCMR(CS) VIN = 7V to 28V 0 5.5 V
Input Offset Voltage VOS(CS) 0.1 mV
Amplifier Gain AV(CS) 34.5 V/V
3dB Bandwidth f3dB 4 MHz
CURRENT-ERROR AMPLIFIER (TRANSCONDUCTANCE AMPLIFIER)
Transconductance gm 550 µS
Open-Loop Gain AVOL(CE) No load 50 dB
DIFFERENTIAL VOLTAGE AMPLIFIER FOR LED CURRENT (DIFF)
Common-Mode Voltage Range VCMR(DIFF) 0 +1.0 V
DIFF Output Voltage VCM VSENSE+ = VSENSE- = 0V 0.6 V
Input Offset Voltage VOS(DIFF) -1 +1 mV
Amplifier Gain AV(DIFF) 0.994 1 1.006 V/V
3dB Bandwidth f3dB CDIFF = 20pF 3 MHz
Minimum Output-Current Drive IOUT(DIFF) 4 mA
SENSE+ to SENSE- Input RVS VSENSE- = 0V 50 100 kΩ
V_IOUT AMPLIFIER
Gain-Bandwidth Product VV_IOUT = 2.0V 4 MHz
3dB Bandwidth VV_IOUT = 2.0V 1 MHz
Output Sink Current 30 µA
Output Source Current 90 µA

_______________________________________________________________________________________ 3
 1.5MHz, 30A High-Efficiency, LED Driver
with Rapid LED Current Pulsing
ELECTRICAL CHARACTERISTICS (continued)
MAX16818

(VCC = 5V, VDD = VCC, TA = TJ = TMIN to TMAX, unless otherwise noted. Typical specifications are at TA = +25°C.) (Note 1)

PARAMETER SYMBOL CONDITIONS MIN TYP MAX UNITS

Maximum Load Capacitance 50 pF
V_IOUT Output to IOUT Transfer
RSENSE = 1mΩ, 100mV ≤ V_IOUT ≤ 5.5V 132.3 135 137.7 mV/A
Function
Offset Voltage 1 mV
VOLTAGE-ERROR AMPLIFIER (EAOUT)
Open-Loop Gain AVOLEA 70 dB
Unity-Gain Bandwidth fGBW 3 MHz
EAN Input Bias Current IB(EA) VEAN = 2.0V -0.2 +0.03 +0.2 µA
Error Amplifier Output Clamping
VCLAMP(EA) With respect to VCM 883 930 976 mV
Voltage
POWER-GOOD AND OVERVOLTAGE PROTECTION
PGOOD goes low when VOUT is below this
PGOOD Trip Level VUV 87.5 90 92.5 %VOUT
threshold
PGOOD Output Low Level VPGLO ISINK = 4mA 0.4 V
PGOOD Output Leakage Current IPG PGOOD = VCC 1 µA
OVI Trip Threshold OVPTH With respect to SGND 1.244 1.276 1.308 V
OVI Input Bias Current IOVI 0.2 µA
ENABLE INPUT
EN Input High Voltage VEN EN rising 2.437 2.5 2.562 V
EN Input Hysteresis 0.28 V
EN Pullup Current IEN 13.5 15 16.5 µA
THERMAL SHUTDOWN
Thermal Shutdown Temperature rising 150 °C
Thermal Shutdown Hysteresis 30 °C
Note 1: Specifications at TA = +25° are 100% tested. Specifications over the temperature range are guaranteed by design.
Note 2: Does not include an error due to finite error amplifier gain. See the Voltage-Error Amplifier (EAOUT) section.

4 _______________________________________________________________________________________
 1.5MHz, 30A High-Efficiency, LED Driver
with Rapid LED Current Pulsing
Typical Operating Characteristics

MAX16818
(TA = +25°C, using Figure 5, unless otherwise noted.)

CURRENT-SENSE THRESHOLD
SUPPLY CURRENT (IQ) vs. FREQUENCY SUPPLY CURRENT vs. TEMPERATURE vs. OUTPUT VOLTAGE
60 70 29.0

MAX16818 toc02

MAX16818 toc03
EXTERNAL CLOCK MAX16818 toc01

NO DRIVER LOAD
50 28.5
SUPPLY CURRENT (mA) 68
SUPPLY CURRENT (mA)

VIN = 24V

(VCSP - VCSN) (mV)

40 28.0
66
30 VIN = 12V 27.5
64
20 27.0
VIN = 5V
62 VIN = 12V
10 fSW = 250kHz 26.5
VIN = 12V
CDL/CDH = 22nF fSW = 250kHz
0 60 26.0
100 300 500 700 900 1100 1300 1500 -40 -15 10 35 60 85 0 1 2 3 4 5
FREQUENCY (kHz) TEMPERATURE (°C) VOUT (V)

VCC LOAD REGULATION DRIVER RISE TIME

HICCUP CURRENT LIMIT vs. REXT vs. INPUT VOLTAGE vs. DRIVER LOAD CAPACITANCE
26.0 5.25 100

MAX16818 toc06
MAX16818 toc04

MAX16818 toc05

VIN = 12V
fSW = 250kHz
25.5
5.15 80
VIN = 24V
CURRENT LIMIT (A)

25.0
5.05 VIN = 12V 60
VCC (V)

tR (ns)

24.5
DL
4.95 40
24.0 VIN = 5V DH
VIN = 12V
fSW = 250kHz 4.85 20
23.5
R1 = 1mΩ
VOUT = 1.5V
23.0 4.75 0
0 4 8 12 16 20 0 25 50 75 100 125 150 1 6 11 16 21
REXT (MΩ) VCC LOAD CURRENT (mA) CAPACITANCE (nF)

DRIVER FALL TIME HIGH-SIDE DRIVER (DH) SINK LOW-SIDE DRIVER (DL) SINK
vs. DRIVER LOAD CAPACITANCE AND SOURCE CURRENT AND SOURCE CURRENT
MAX16818 toc08 MAX16818 toc09
100
MAX16818 toc07

VIN = 12V CLOAD = 22nF CLOAD = 22nF

fSW = 250kHz VIN = 12V VIN = 12V
80

60
tF (ns)

DL 2A/div 3A/div
40
DH

20

0
1 6 11 16 21 100ns/div 100ns/div
CAPACITANCE (nF)

_______________________________________________________________________________________ 5
 1.5MHz, 30A High-Efficiency, LED Driver
with Rapid LED Current Pulsing
Typical Operating Characteristics (continued)
MAX16818

(TA = +25°C, using Figure 5, unless otherwise noted.)

HIGH-SIDE DRIVER (DH) RISE TIME HIGH-SIDE DRIVER (DH) FALL TIME LOW-SIDE DRIVER (DL) RISE TIME
MAX16818 toc10 MAX16818 toc11 MAX16818 toc12

CLOAD = 22nF CLOAD = 22nF CLOAD = 22nF

VIN = 12V VIN = 12V VIN = 12V

2V/div 2V/div 2V/div

40ns/div 40ns/div 40ns/div

LOW-SIDE DRIVER (DL) FALL TIME FREQUENCY vs. RT

MAX16818 toc13
10,000

MAX16818 toc14
CLOAD = 22nF VIN = 12V
VIN = 12V
fSW (kHz)

2V/div 1000

100
40ns/div 30 110 190 270 350 430 510
70 150 230 310 390 470
RT (kΩ)

FREQUENCY vs. TEMPERATURE SYNC, CLKOUT, AND LX WAVEFORM

MAX16818 toc16
260
MAX16818 toc15

VIN = 12V
258 SYNC
256 5V/div

254
252 CLKOUT
fSW (kHz)

5V/div
250
248 VIN = 12V
fSW = 250kHz
246
244 LX
10V/div
242
240
-40 -15 10 35 60 85 1μs/div
TEMPERATURE (°C)

6 _______________________________________________________________________________________
 1.5MHz, 30A High-Efficiency, LED Driver
with Rapid LED Current Pulsing
Pin Description

MAX16818
PIN NAME FUNCTION
1 PGND Power-Supply Ground
2, 7 N.C. No Connection. Not internally connected.
3 DL Low-Side Gate Driver Output
Boost Flying Capacitor Connection. Reservoir capacitor connection for the high-side MOSFET driver
4 BST
supply. Connect a ceramic capacitor between BST and LX.
5 LX Source connection for the high-side MOSFET.
6 DH High-Side Gate Driver Output. Drives the gate of the high-side MOSFET.
Signal Ground. Ground connection for the internal control circuitry. Connect SGND and PGND
8, 22, 25 SGND
together at one point near the IC.
9 CLKOUT Oscillator Output. Rising edge of CLKOUT is phase-shifted from the rising edge of DH by 180°.
10 PGOOD Power-Good Output
Output Enable. Drive high or leave unconnected for normal operation. Drive low to shut down the
11 EN power drivers. EN has an internal 15µA pullup current. Connect a capacitor from EN to SGND to
program the hiccup-mode duty cycle.

Switching Frequency Programming and Chip-Enable Input. Connect a resistor from RT/SYNC to
12 RT/SYNC SGND to set the internal oscillator frequency. Drive RT/SYNC to synchronize the switching frequency
with external clock.
13 V_IOUT Voltage Source Output Proportional to the Inductor Current. The voltage at V_IOUT = 135 x ILED x RS.
Current-Limit Setting Input. Connect a resistor from LIM to SGND to set the hiccup current-limit
14 LIM
threshold. Connect a capacitor from LIM to SGND to ignore short output overcurrent pulses.

Overvoltage Protection. Connect OVI to DIFF. When OVI exceeds 12.7% above the programmed
15 OVI output voltage, DH is latched low and DL is latched high. Toggle EN or recycle the input power to
reset the latch.
16 CLP Current-Error Amplifier Output. Compensate the current loop by connecting an RC network to ground.
17 EAOUT Voltage-Error Amplifier Output. Connect to the external compensation network.
18 EAN Voltage-Error Amplifier Inverting Input
Differential Remote-Sense Amplifier Output. DIFF is the output of a precision unity-gain amplifier
19 DIFF
whose inputs are SENSE+ and SENSE-.
Current-Sense Differential Amplifier Negative Input. The differential voltage between CSN and CSP is
20 CSN
amplified internally by the current-sense amplifier (gain = 34.5) to measure the inductor current.

_______________________________________________________________________________________ 7
 1.5MHz, 30A High-Efficiency, LED Driver
with Rapid LED Current Pulsing
Pin Description (continued)
MAX16818

PIN NAME FUNCTION

Current-Sense Differential Amplifier Positive Input. The differential voltage between CSN and CSP is
21 CSP
amplified internally by the current-sense amplifier (gain = 34.5) to measure the inductor current.
Differential LED Current-Sensing Negative Input. SENSE- is used to sense the LED current. Connect
23 SENSE-
SENSE- to the negative side of the LED current-sense resistor.
Differential LED Current-Sensing Positive Input. SENSE+ is used to sense the LED current. Connect
24 SENSE+
SENSE+ to the positive side of the LED current-sense resistor.
26 IN Supply Voltage Connection. Connect IN to VCC for a +5V system.
Internal +5V Regulator Output. VCC is derived from the IN voltage. Bypass VCC to SGND with 4.7µF
27 VCC
and 0.1µF ceramic capacitors.

Supply Voltage for Low-Side and High-Side Drivers. Connect a parallel combination of 0.1µF and 1µF
28 VDD ceramic capacitors to PGND and a 1Ω resistor to VCC to filter out the high peak currents of the driver
from internal circuitry.

Exposed Paddle. Connect the exposed paddle to a copper pad (SGND) to improve power
— EP
dissipation.

8 _______________________________________________________________________________________
 1.5MHz, 30A High-Efficiency, LED Driver
with Rapid LED Current Pulsing
Typical Application Circuits

MAX16818
ON/OFF

C3 R6
VIN
VCC 7V TO 28V
R3 R4 R5
VLED

C2
L1
14 13 12 11 10 9 8
LIM V_IOUT RT/SYNC EN PGOOD CLKOUT SGND VLED
C10 D1
15 OVI N.C. 7

C9 R12 Q1
16 CLP DH 6

R11
17 EAOUT LX 5 LED
C1
C8 STRING
R7
C7 18 EAN BST 4
MAX16818
R10
19 DIFF DL 3 R2

20 CSN N.C. 2
R1

21 CSP PGND 1
SGND SENSE- SENSE+ SGND IN VCC VDD
22 23 24 25 26 27 28
VCC

VIN R8

C6 C5 C4

Figure 1. Typical Application Circuit for a Boost LED Driver (Nonsynchronous)

_______________________________________________________________________________________ 9
 1.5MHz, 30A High-Efficiency, LED Driver
with Rapid LED Current Pulsing
Typical Application Circuits (continued)
MAX16818

ON/OFF

C3 R6
VIN
VCC 7V TO 28V LED
R3 R4 R5 R2
VLED STRING
1 TO 6
C2 LEDS
L1
14 13 12 11 10 9 8
LIM V_IOUT RT/SYNC EN PGOOD CLKOUT SGND VLED
C10 D1
15 OVI N.C. 7

C9 R12 Q1
16 CLP DH 6
VCC
R11
17 EAOUT LX 5 RS+ VCC
C8
R7 MAX4073T
C7 18 EAN BST 4 RS- OUT
MAX16818
C1
R10
19 DIFF DL 3

20 CSN N.C. 2
R1

21 CSP PGND 1
SGND SENSE- SENSE+ SGND IN VCC VDD
22 23 24 25 26 27 28
VCC

VIN R8

C6 C5 C4

Figure 2. Typical Application Circuit for an Input-Referred Buck-Boost LED Driver (Input: 7V to 28V, Output: 1 to 6 LEDs in Series)

10 ______________________________________________________________________________________
 1.5MHz, 30A High-Efficiency, LED Driver
with Rapid LED Current Pulsing
Typical Application Circuits (continued)

MAX16818
ON/OFF

C4 R6
VIN
VCC 7V TO 28V
R3 R4 R5
VLED

C3
L1
14 13 12 11 10 9 8
LIM V_IOUT RT/SYNC EN PGOOD CLKOUT SGND VLED
C11 C1 D1
15 OVI N.C. 7

C10 R12 Q1
16 CLP DH 6

R11
17 EAOUT LX 5 LED
L2 C2 STRING
C9

C8 18 EAN BST 4
MAX16818
R10
19 DIFF DL 3 R7 R2

20 CSN N.C. 2
R1

21 CSP PGND 1
SGND SENSE- SENSE+ SGND IN VCC VDD
22 23 24 25 26 27 28
VCC

VIN R8

C7 C6 C5

Figure 3. Typical Application Circuit for a SEPIC LED Driver

______________________________________________________________________________________ 11
 1.5MHz, 30A High-Efficiency, LED Driver
with Rapid LED Current Pulsing
Typical Application Circuits (continued)
MAX16818

ON/OFF
C3 R6
VCC
R3 R4 R5 VIN
VLED 7V TO 18V

14 13 12 11 10 9 8
C2
C11 LIM V_IOUT RT/SYNC EN PGOOD CLKOUT SGND
15 OVI N.C. 7

Q1
C10 R12
16 CLP DH 6
VLED
R11 L1
17 EAOUT LX 5
C9 C4
R7 Q3
C8 18 EAN BST 4
MAX16818 LED
Q2 STRING
R10 C1
19 DIFF DL 3

20 CSN N.C. 2 D2
R2
R1
21 CSP PGND 1
SGND SENSE- SENSE+ SGND IN VCC VDD
22 23 24 25 26 27 28
VCC

VIN R8

C7 C6 C5

Figure 4. Application Circuit for a Ground-Referred Buck-Boost LED Driver

12 ______________________________________________________________________________________
 1.5MHz, 30A High-Efficiency, LED Driver
with Rapid LED Current Pulsing
Typical Application Circuits (continued)

MAX16818
VCC

R4

C3 VIN
R3 ON/OFF 7V TO 28V

14 13 12 11 10 9 8
C2
C11 LIM V_IOUT RT/SYNC EN PGOOD CLKOUT SGND
15 OVI N.C. 7

C10 R10
16 CLP DH 6

R9 L1
17 EAOUT LX 5
C9 C4
R5
C8 18 EAN BST 4 Q1
MAX16818 LED
D1 C1 STRING
R8
19 DIFF DL 3

20 CSN N.C. 2
R2
R1
21 CSP PGND 1
SGND SENSE- SENSE+ SGND IN VCC VDD
22 23 24 25 26 27 28
VCC

VIN R6

C7 C6 C5

Figure 5. Application Circuit for a Buck LED Driver

______________________________________________________________________________________ 13
 1.5MHz, 30A High-Efficiency, LED Driver
with Rapid LED Current Pulsing
Functional Diagram
MAX16818

VCC

IS

EN

0.5V x VCC

IN 5V UVLO
LDO POR
REGULATOR TEMP SENSOR
VCC
TO INTERNAL
CIRCUITS HICCUP MODE
LIM CURRENT LIMIT MAX16818
VCM
126.7kΩ

100kΩ

S Q

0.5 x VCLAMP RT
CLP R Q
Ct
AV = 34.5
CSP VCM
CA gm = 500μS VDD
CSN
PWM
AV = 4 CEA COMPARATOR BST
V_IOUT
VCLAMP VCLAMP CPWM
LOW HIGH RAMP S Q DH
SGND

2 x fS (V/s) LX
RT/SYNC CLK
OSCILLATOR R Q DL

CLKOUT
RAMP PGND
DIFF GENERATOR

SENSE- +0.6V PGOOD

N
DIFF
SENSE+ AMP 0.1 x VREF

EAOUT

ERROR AMP
EAN 0.12 x VREF
OVP LATCH
VEA

LATCH
OVP COMP
SOFT-
START VREF = 0.6V

CLEAR ON UVLO RESET OR

VCM (0.6V) ENABLE LOW

OVI

Figure 6. MAX16818 Functional Diagram

14 ______________________________________________________________________________________
 1.5MHz, 30A High-Efficiency, LED Driver
with Rapid LED Current Pulsing
Detailed Description PD = VIN x ICC

MAX16818
ICC = IQ + [fSW x (QG1 + QG2)]
The MAX16818 is a high-performance average-current-
mode PWM controller for high-power, high-brightness where QG1 and QG2 are the total gate charge of the
LEDs (HBLEDs). Average current-mode control is the low-side and high-side external MOSFETs at VGATE =
ideal method for driving HBLEDs. This technique offers 5V, IQ is 3.5mA (typ), and fSW is the switching frequen-
inherently stable operation, reduces component derat- cy of the converter.
ing and size by accurately controlling the inductor cur-
Undervoltage Lockout (UVLO)
rent. The device achieves high efficiency at high
The MAX16818 includes an undervoltage lockout with
current (up to 30A) with a minimum number of external
hysteresis and a power-on-reset circuit for converter
components. The high- and low-side drivers source
turn-on. The UVLO rising threshold is internally set at
and sink up to 4A for lower switching losses while dri-
4.35V with a 200mV hysteresis. Hysteresis at UVLO
ving high-gate-charge MOSFETs. The MAX16818’s
eliminates chattering during startup.
CLKOUT output is 180° out-of-phase with respect to the
high-side driver. CLKOUT drives a second MAX16818 Most of the internal circuitry, including the oscillator,
LED driver out of phase, reducing the input-capacitor turns on when the input voltage reaches 4V. The
ripple current. MAX16818 draws up to 3.5mA of current before the
input voltage reaches the UVLO threshold.
The MAX16818 consists of an inner average current loop
representing inductor current and an outer voltage loop Soft-Start
voltage-error amplifier (VEA) that directly controls LED The MAX16818 has an internal digital soft-start for a
current. The combined action of the two loops results in monotonic, glitch-free rise of the output current. Soft-
a tightly regulated LED current. The inductor current is start is achieved by the controlled rise of the error
sensed across a current-sense resistor. The differential amplifier dominant input in steps using a 5-bit counter
amplifier senses LED current through a sense resistor in and a 5-bit DAC. The soft-start DAC generates a linear
series with the LEDs and the resulting sensed voltage is ramp from 0 to 0.7V. This voltage is applied to the error
compared against an internal 0.6V reference at the error- amplifier at a third (noninverting) input. As long as the
amplifier input. The MAX16818 will adjust the LED cur- soft-start voltage is lower than the reference voltage,
rent to within 1% accuracy to maintain emitted spectrum the system converges to that lower reference value.
of the light in HBLEDs. Once the soft-start DAC output reaches 0.6V, the refer-
ence takes over and the DAC output continues to climb
IN, VCC, and VDD
to 0.7V, assuring that it does not interfere with the refer-
The MAX16818 accepts either a 4.75V to 5.5V or 7V to
ence voltage.
28V input voltage range. All internal control circuitry
operates from an internally regulated nominal voltage of Internal Oscillator
5V (VCC). For input voltages of 7V or greater, the inter- The internal oscillator generates a clock with the fre-
nal VCC regulator steps the voltage down to 5V. The quency proportional to the inverse of RT. The oscillator
VCC output voltage is a regulated 5V output capable of frequency is adjustable from 125kHz to 1.5MHz with
sourcing up to 60mA. Bypass the VCC to SGND with better than 8% accuracy using a single resistor con-
4.7µF and 0.1µF low-ESR ceramic capacitors for high- nected from RT/SYNC to SGND. The frequency accura-
frequency noise rejection and stable operation. cy avoids the over-design, size, and cost of passive
The MAX16818 uses VDD to power the low-side and filter components like inductors and capacitors. Use
high-side drivers. Isolate VDD from VCC with a 1Ω resis- the following equation to calculate the oscillator fre-
tor and put a 1µF capacitor in parallel with a 0.1µF quency:
capacitor to ground to prevent high-current noise spikes For 120kΩ ≤ RT ≤ 500kΩ:
created by the driver from disrupting internal circuitry.
The TQFN is a thermally enhanced package and can 6.25 x 1010
RT =
dissipate up to 2.7W. The high-power packages allow fSW
the high-frequency, high-current converter to operate
from a 12V or 24V bus. Calculate power dissipation in For 40kΩ ≤ RT ≤ 120kΩ:
the MAX16818 as a product of the input voltage and the
total VCC regulator output current (ICC). ICC includes qui- 6.40 x 1010
escent current (IQ) and gate-drive current (IDD): RT =
fSW

______________________________________________________________________________________ 15
 1.5MHz, 30A High-Efficiency, LED Driver
with Rapid LED Current Pulsing
The oscillator also generates a 2VP-P voltage-ramp sig- PWM comparator (CPWM) (Figure 7). The precision CA
MAX16818

nal for the PWM comparator and a 180° out-of-phase amplifies the sense voltage across RS by a factor of
clock signal for CLKOUT to drive a second LED regula- 34.5. The inverting input to the CEA senses the CA out-
tor out-of-phase. put. The CEA output is the difference between the volt-
age-error amplifier output (EAOUT) and the amplified
Synchronization voltage from the CA. The RC compensation network
The MAX16818 can be easily synchronized by con- connected to CLP provides external frequency compen-
necting an external clock to RT/SYNC. If an external sation for the CEA. The start of every clock cycle
clock is present, then the internal oscillator is disabled enables the high-side drivers and initiates a PWM on-
and the external clock is used to run the device. If the cycle. Comparator CPWM compares the output voltage
external clock is removed, the absence of clock for from the CEA with a 0V to 2V ramp from the oscillator.
32µs is detected and the circuit starts switching from The PWM on-cycle terminates when the ramp voltage
the internal oscillator. Pulling RT/SYNC to ground for at exceeds the error voltage. Compensation for the outer
least 50µs disables the converter. Use an open-collec- LED current loop varies based upon the topology.
tor transistor to synchronize the MAX16818 with the
external system clock. The MAX16818 outer LED current control loop consists
of the differential amplifier (DIFF AMP), reference volt-
Control Loop age, and VEA. The unity-gain differential amplifier pro-
The MAX16818 uses an average-current-mode control vides true differential remote sensing of the voltage
scheme to regulate the output current (Figure 7). The across the LED current set resistor, RLS. The differential
main control loop consists of an inner current loop for amplifier output connects to the inverting input (EAN) of
controlling the inductor current and an outer current the VEA. The DIFF AMP is bypassed and the inverting
loop for regulating the LED current. The inner current input is available to the pin for direct feedback. The
loop absorbs the inductor pole reducing the order of the noninverting input of the VEA is internally connected to
outer current loop to that of a single-pole system. The an internal precision reference voltage, set to 0.6V. The
current loop consists of a current-sense resistor (RS), a VEA controls the inner current loop (Figure 6). A feed-
current-sense amplifier (CA), a current-error amplifier back network compensates the outer loop using the
(CEA), an oscillator providing the carrier ramp, and a EAOUT and EAIN pins.

CCF RCF

CCFF
CSN CSP CLP VIN

CA
EAOUT MAX16818
SENSE+ 600mV IL
DIFF CEA
Z COMP

AMP
VEA CPWM DRIVE LED
SENSE- STRING
EAN COUT
VREF + VCM = 1.2V
RLS
DIFF RS

Figure 7. MAX16818 Control Loop

16 ______________________________________________________________________________________
 1.5MHz, 30A High-Efficiency, LED Driver
with Rapid LED Current Pulsing
Inductor Current-Sense Amplifier Current-Error Amplifier

MAX16818
The differential current-sense amplifier (CA) provides a (For Inductor Currents)
DC gain of 34.5. The maximum input offset voltage of The MAX16818 has a transconductance current-error
the current-sense amplifier is 1mV and the common- amplifier (CEA) with a typical gm of 550µS and 320µA
mode voltage range is 0 to 5.5V (IN = 7V to 28V). The output sink- and source-current capability. The current-
current-sense amplifier senses the voltage across a error amplifier output CLP serves as the inverting input
current-sense resistor. The maximum common-mode to the PWM comparator. CLP is externally accessible to
voltage is 3.6V when VIN = 5V. provide frequency compensation for the inner current
loops (Figure 7). Compensate (CEA) so the inductor
Inductor Peak-Current Comparator current negative slope, which becomes the positive
The peak-current comparator provides a path for fast slope to the inverting input of the PWM comparator, is
cycle-by-cycle current limit during extreme fault condi- less than the slope of the internally generated voltage
tions, such as an inductor malfunction (Figure 8). Note ramp (see the Compensation section).
the average current-limit threshold of 26.9mV still limits
the output current during short-circuit conditions. To PWM Comparator and R-S Flip-Flop
prevent inductor saturation, select an inductor with a The PWM comparator (CPWM) sets the duty cycle for
saturation current specification greater than the average each cycle by comparing the output of the current-error
current limit. Proper inductor selection ensures that only amplifier to a 2VP-P ramp. At the start of each clock
the extreme conditions trip the peak-current compara- cycle, an R-S flip-flop resets and the high-side driver
tor, such as an inductor with a shorted turn. The 60mV (DH) goes high. The comparator sets the flip-flop as
threshold for triggering the peak-current limit is twice the soon as the ramp voltage exceeds the CLP voltage,
full-scale average current-limit voltage threshold. The thus terminating the on-cycle (Figure 8).
peak-current comparator has only a 260ns delay.

VDD

PEAK-CURRENT
COMPARATOR
60mV

CLP

AV = 34.5
CSP MAX16818
CA gm = 550μS
CSN
BST
CEA
SET
VEA S Q DH
EAN CPWM
RAMP
EAOUT 2 x fS (V/s) LX
CLK
R Q DL
CLR

SHDN PGND

Figure 8. MAX16818 Phase Circuit

______________________________________________________________________________________ 17
 1.5MHz, 30A High-Efficiency, LED Driver
with Rapid LED Current Pulsing
Differential Amplifier BST
MAX16818

The DIFF AMP facilitates remote sensing at the load The MAX16818 uses VDD to power the low- and high-
(Figure 7). It provides true differential LED current side MOSFET drivers. The high-side driver derives its
(through the RLS sense resistor) sensing while rejecting power through a bootstrap capacitor and VDD supplies
the common-mode voltage errors due to high-current power internally to the low-side driver. Connect a
ground paths. The VEA provides the difference 0.47µF low-ESR ceramic capacitor between BST and
between the differential amplifier output (DIFF) and the LX. Connect a Schottky rectifier from BST to VDD. Keep
desired LED current-sense voltage. The differential the loop formed by the boost capacitor, rectifier, and IC
amplifier has a bandwidth of 3MHz. The difference small on the PCB.
between SENSE+ and SENSE- is regulated to 0.6V.
Connect SENSE+ to the positive side of the LED current- Protection
sense resistor and SENSE- to the negative side of the The MAX16818 includes output overvoltage protection
LED current-sense resistor (which is often PGND). (OVP). During fault conditions when the load goes to
high impedance (opens), the controller attempts to
MOSFET Gate Drivers (DH, DL) maintain LED current. The OVP protection disables the
The high-side (DH) and low-side (DL) drivers drive the MAX16818 whenever the voltage exceeds the thresh-
gates of external n-channel MOSFETs (Figures 1–5). old, protecting the external circuits from undesirable
The drivers’ 4A peak sink- and source-current capabili- voltages.
ty provides ample drive for the fast rise and fall times of
the switching MOSFETs. Faster rise and fall times result Current Limit
in reduced cross-conduction losses. Due to physical The VEA output is clamped to 930mV with respect to
realities, extremely low gate charges and R DS(ON) the common-mode voltage (V CM). Average-current-
resistance of MOSFETs are typically exclusive of each mode control has the ability to limit the average current
other. MOSFETs with very low RDS(ON) will have a high- sourced by the converter during a fault condition. When
er gate charge and vice versa. Choosing the high-side a fault condition occurs, the VEA output clamps to
MOSFET (Q1) becomes a trade-off between these two 930mV with respect to the common-mode voltage
attributes. Applications where the input voltage is much (0.6V) to limit the maximum current sourced by the con-
higher than the output voltage result in a low duty cycle verter to ILIMIT = 26.9mV / RS. The hiccup current limit
where conduction losses are less important than overrides the average current limit. The MAX16818
switching losses. In this case, choose a MOSFET with includes hiccup current-limit protection to reduce the
very low gate charge and a moderate R DS(ON). power dissipation during a fault condition. The hiccup
Conversely, for applications where the output voltage is current-limit circuit derives inductor current information
near the input voltage resulting in duty cycles much from the output of the current amplifier. This signal is
greater than 50%, the RDS(ON) losses become at least compared against one half of V CLAMP(EA) . With no
equal, or even more important than the switching losses. resistor connected from the LIM pin to ground, the hic-
In this case, choose a MOSFET with very low RDS(ON) cup current limit is set at 90% of the full-load average
and moderate gate charge. Finally, for the applications current limit. Use REXT to increase the hiccup current
where the duty cycle is near 50%, the two loss compo- limit from 90% to 100% of the full load average limit.
nents are nearly equal, and a balanced MOSFET with The hiccup current limit can be disabled by connecting
moderate gate charge and RDS(ON) work best. LIM to SGND. In this case, the circuit follows the aver-
age current-limit action during overload conditions.
In a buck topology, the low-side MOSFET (Q2) typically
operates in a zero voltage switching mode, thus it does Overvoltage Protection
not have switching losses. Choose a MOSFET with very The OVP comparator compares the OVI input to the
low RDS(ON) and moderate gate charge. overvoltage threshold. A detected overvoltage event
Size both the high-side and low-side MOSFETs to han- latches the comparator output forcing the power stage
dle the peak and RMS currents during overload condi- into the OVP state. In the OVP state, the high-side
tions. The driver block also includes a logic circuit that MOSFET turns off and the low-side MOSFET latches on.
provides an adaptive nonoverlap time to prevent shoot- Connect OVI to the center tap of a resistor-divider from
through currents during transition. The typical nonover- VLED to SGND. In this case, the center tap is compared
lap time between the high-side and low-side MOSFETs against 1.276V. Add an RC delay to reduce the sensitivity
is 35ns. of the overvoltage circuit and avoid nuisance tripping of
the converter. Disable the overvoltage function by con-
necting OVI to SGND.

18 ______________________________________________________________________________________
 1.5MHz, 30A High-Efficiency, LED Driver
with Rapid LED Current Pulsing
Applications Information span from the output to the input. This effectively

MAX16818
removes the boost-only restriction of the regulator in
Application Circuit Descriptions Figure 1, allowing the voltage across the LEDs to be
This section provides some detail regarding the appli- greater than or less than the input voltage. LED current
cation circuits in the Simplified Diagram and Figures sensing is not ground-referenced, so a high-side cur-
1–5. The discussion includes some description of the rent-sense amplifier is used to measure current.
topology as well as basic attributes.
SEPIC LED Driver
High-Frequency LED Current Pulser Figure 3 shows the MAX16818 configured as a SEPIC
The Simplified Diagram shows the MAX16818 providing LED driver. While buck topologies require the output to
high-frequency, high-current pulses to the LEDs. The be lesser than the input, and boost topologies require
basic topology must be a buck, since the inductor the output to be greater than the input, a SEPIC topolo-
always connects to the load in that configuration (in all gy allows the output voltage to be greater than, equal
other topologies, the inductor disconnects from the to, or less than the input. In a SEPIC topology, the volt-
load at one time or another). The design minimizes the age across C1 is the same as the input voltage, and L1
current ripple by oversizing the inductor, which allows and L2 are the same inductance. Therefore, when Q1
for a very small (0.01µF) output capacitor. When MOS- conducts (on-time), both inductors ramp up current at
FET Q3 turns on, it diverts the current around the LEDs the same rate. The output capacitor supports the out-
at a very fast rate. Q3 also discharges the output put voltage during this time. During the off-time, L1 cur-
capacitor, but since the capacitor is so small, it does rent recharges C1 and combines with L2 to provide
not stress the MOSFET. Resistor R1 senses the LED/Q3 current to recharge C2 and supply the load current.
current and there is no reaction to the short that Q3 Since the voltage waveform across L1 and L2 are
places across the LEDs. This design is superior in that exactly the same, it is possible to wind both inductors
it does not attempt to actually change the inductor cur- on the same core (a coupled inductor). Although volt-
rent at high frequencies and yet the current in the LEDs ages on L1 and L2 are the same, RMS currents can be
varies from zero to full in very small periods of time. The quite different so the windings may have a different
efficiency of this technique is very high. Q3 must be gauge wire. Because of the dual inductors and seg-
able to dissipate the LED current applied to its RDS(ON) mented energy transfer, the efficiency of a SEPIC con-
at some maximum duty cycle. If the circuit needs to verter is somewhat lower than standard bucks or
control extremely high currents, use paralleled boosts. As in the boost driver, the current-sense resis-
MOSFETs. PGOOD is low during LED pulsed-current tor connects to ground, allowing the output voltage of
operation. the LED driver to exceed the rated maximum voltage of
the MAX16818.
Boost LED Driver
In Figure 1, the external components configure the Ground-Referenced Buck/Boost LED Driver
MAX16818 as a boost converter. The circuit applies the Figure 4 depicts a buck/boost topology. During the on-
input voltage to the inductor during the on-time, and time with this circuit, the current flows from the input
then during the off-time the inductor, which is in series capacitor, through Q1, L1, and Q3 and back to the
with the input capacitor, charges the output capacitor. input capacitor. During the off-time, current flows up
Because of the series connection between the input through Q2, L1, D1, and to the output capacitor C1.
voltage and the inductor, the output voltage can never This topology resembles a boost in that the inductor
go lower than the input voltage. The design is nonsyn- sits between the input and ground during the on-time.
chronous, and since the current-sense resistor con- However, during the off-time the inductor resides
nects to ground, the power supply can go to any output between ground and the output capacitor (instead of
voltage (above the input) as long as the components are between the input and output capacitors in boost
rated appropriately. R2 again provides the sense voltage topologies), so the output voltage can be any voltage
the MAX16818 uses to regulate the LED current. less than, equal to, or greater than the input voltage. As
compared to the SEPIC topology, the buck/boost does
Input-Referenced LED Driver not require two inductors or a series capacitor, but it
The circuit in Figure 2 shows a step-up/step-down reg- does require two additional MOSFETs.
ulator. It is similar to the boost converter in Figure 1 in
that the inductor is connected to the input and the Buck Driver with Synchronous Rectification
MOSFET is essentially connected to ground. However, In Figure 5, the input voltage can go from 7V to 28V and,
rather than going from the output to ground, the LEDs because of the ground-based current-sense resistor, the

______________________________________________________________________________________ 19
 1.5MHz, 30A High-Efficiency, LED Driver
with Rapid LED Current Pulsing
output voltage can be as high as the input. The synchro- surface-mount inductor series available from various
MAX16818

nous MOSFET keeps the power dissipation to a minimum, manufacturers.

especially when the input voltage is large when com- For example, for a buck regulator and 2 LEDs in series,
pared to the voltage on the LED string. It is important to calculate the minimum inductance at VIN(MAX) = 13.2V,
keep the current-sense resistor, R1, inside the LC loop, VLED = 7.8V, ΔIL = 400mA, and fSW = 330kHz:
so that ripple current is available. To regulate the LED
current, R2 creates a voltage that the differential amplifier Buck regulators:
compares to 0.6V. If power dissipation is a problem in R2,
add a noninverting amplifier and reduce the value of the (13.2 − 7.8) x 7.8
LMIN = = 24.2μH
sense resistor accordingly. 13.2 x 330k x 0.4
Inductor Selection For a boost regulator with four LEDs in series, calculate
The switching frequencies, peak inductor current, and the minimum inductance at VIN(MAX) = 13.2V, VLED =
allowable ripple at the output determine the value and 15.6V, ΔIL =400mA, and fSW = 330kHz:
size of the inductor. Selecting higher switching frequen-
cies reduces the inductance requirement, but at the Boost regulators:
cost of lower efficiency. The charge/discharge cycle of
(15.6 − 13.2) x 13.2
the gate and drain capacitances in the switching LMIN = = 15.3μH
MOSFETs create switching losses. The situation wors- 15.6 x 330k x 0.4
ens at higher input voltages, since switching losses are
proportional to the square of the input voltage. The The average-current-mode control feature of the
MAX16818 can operate up to 1.5MHz, however for MAX16818 limits the maximum peak inductor current
VIN > +12V, use lower switching frequencies to limit the and prevents the inductor from saturating. Choose an
switching losses. inductor with a saturating current greater than the
The following discussion is for buck or continuous worst-case peak inductor current. Use the following
boost-mode topologies. Discontinuous boost, buck- equation to determine the worst-case inductor current:
boost, and SEPIC topologies are quite different in
VCL ΔICL
regards to component selection. ILPEAK = +
RS 2
Use the following equations to determine the minimum
inductance value:
where R S is the inductor sense resistor and V CL =
Buck regulators: 0.0282V.
(VINMAX − VLED) x VLED Switching MOSFETs
LMIN =
VINMAX x fSW x ΔIL When choosing a MOSFET for voltage regulators, con-
sider the total gate charge, RDS(ON), power dissipation,
Boost regulators: and package thermal impedance. The product of the
MOSFET gate charge and on-resistance is a figure of
(VLED − VINMAX) x VINMAX merit, with a lower number signifying better perfor-
LMIN = mance. Choose MOSFETs optimized for high-frequen-
VLED x fSW x ΔIL
cy switching applications.
where VLED is the total voltage across the LED string. The average current from the MAX16818 gate-drive
As a first approximation choose the ripple current, ΔIL, output is proportional to the total capacitance it drives
equal to approximately 40% of the output current. at DH and DL. The power dissipated in the MAX16818
Higher ripple current allows for smaller inductors, but it is proportional to the input voltage and the average
also increases the output capacitance for a given volt- drive current. See the IN, V CC, and V DD section to
age ripple requirement. Conversely, lower ripple cur- determine the maximum total gate charge allowed from
rent increases the inductance value, but allows the the combined driver outputs. The gate-charge and
output capacitor to reduce in size. This trade-off can be drain-capacitance (CV 2) loss, the cross-conduction
altered once standard inductance and capacitance val- loss in the upper MOSFET due to finite rise/fall times,
ues are chosen. Choose inductors from the standard and the I2R loss due to RMS current in the MOSFET
RDS(ON) account for the total losses in the MOSFET.

20 ______________________________________________________________________________________
 1.5MHz, 30A High-Efficiency, LED Driver
with Rapid LED Current Pulsing
Buck Regulator Boost Regulator

MAX16818
Estimate the power loss (PDMOS_) caused by the high-side Estimate the power loss (PDMOS_) caused by the MOS-
and low-side MOSFETs using the following equations: FET using the following equations:

PDMOS − HI = (QG x VDD x fSW ) + PDFET = (QG x VDD x fSW ) +

⎛ VIN x IOUT x (tR + tF) x fSW ⎞
⎛ VIN x IOUT x (tR + tF) x fSW ⎞ ⎜ ⎟ + (RDS(ON) x IRMS − HI2)
⎝ 2 ⎠
⎜ ⎟
⎝ 2 ⎠ D
IRMS − HI = (IVALLEY2 + IPK 2 + IVALLEY x IPK ) x
+ (RDS(ON) x IRMS − HI )
2 3

where QG, RDS(ON), tR, and tF are the upper-switching For a boost regulator in continuous mode, D = VLEDs /
MOSFET’s total gate charge, on-resistance at maximum (VIN + VLEDs), IVALLEY = (IOUT - ΔIL / 2) and IPK =
operating temperature, rise time, and fall time, respectively. (IOUT + ΔIL / 2).
The voltage across the MOSFET:
D
IRMS − HI = (IVALLEY2 + IPK 2 + IVALLEY x IPK ) x VMOSFET = VLED + VF
3
where VF is the maximum forward voltage of the diode.
For the buck regulator, D = V LEDs / V IN, I VALLEY = The output diode on a boost regulator must be rated to
(IOUT - ΔIL / 2) and IPK = (IOUT + ΔIL / 2). handle the LED series voltage, VLED. It should also
have fast reverse-recovery characteristics and should
PDMOS − LO = (QG x VDD x fSW ) + handle the average forward current that is equal to the
LED current.
(RDS(ON) x IRMS − LO2)
(1− D)
Input Capacitors
IRMS − LO = (IVALLEY2 + IPK2 + IVALLEY x IPK) x For buck regulator designs, the discontinuous input
3
current waveform of the buck converter causes large
For example, from the typical specifications in the ripple currents in the input capacitor. The switching fre-
Applications Information section with VOUT = 7.8V, the quency, peak inductor current, and the allowable peak-
high-side and low-side MOSFET RMS currents are to-peak voltage ripple reflected back to the source
0.77A and 0.63A, respectively, for a 1A buck regulator. dictate the capacitance requirement. Increasing
Ensure that the thermal impedance of the MOSFET switching frequency or paralleling out-of-phase con-
package keeps the junction temperature at least +25°C verters lowers the peak-to-average current ratio, yield-
below the absolute maximum rating. Use the following ing a lower input capacitance requirement for the same
equation to calculate the maximum junction tempera- LED current. The input ripple is comprised of ΔV Q
ture: TJ = (PDMOS x θJA) + TA, where θJA and TA are (caused by the capacitor discharge) and ΔV ESR
the junction-to-ambient thermal impedance and ambi- (caused by the ESR of the capacitor). Use low-ESR
ent temperature, respectively. ceramic capacitors with high-ripple-current capability at
To guarantee that there is no shoot-through from VIN to the input. Assume the contributions from the ESR and
PGND, the MAX16818 produces a nonoverlap time of capacitor discharge are equal to 30% and 70%, respec-
35ns. During this time, neither high- nor low-side MOS- tively. Calculate the input capacitance and ESR required
FET is conducting, and since the output inductor must for a specified ripple using the following equation:
maintain current flow, the intrinsic body diode of the
ΔVESR
low-side MOSFET becomes the conduction path. Since ESRIN =
this diode has a fairly large forward voltage, a Schottky ⎛ ΔIL ⎞
⎜IOUT + ⎟
diode (in parallel to the low-side MOSFET) diverts current ⎝ 2 ⎠
flow from the MOSFET body diode because of its lower
forward voltage, which, in turn, increases efficiency.

______________________________________________________________________________________ 21
 1.5MHz, 30A High-Efficiency, LED Driver
with Rapid LED Current Pulsing
Buck: Current Limit
MAX16818

I x D(1− D) In addition to the average current limit, the MAX16818

CIN = OUT also has hiccup current limit. The hiccup current limit is
ΔVQ x fSW
set to 10% below the average current limit to ensure that
where IOUT is the output current of the converter. For the circuit goes in hiccup mode during continuous out-
example, at VIN = 13.2V, VLED = 7.8V, IOUT = 1A, ΔIL = put short circuit. Connecting a resistor from LIM to
0.4A, and fSW = 330kHz, the ESR and input capaci- ground increases the hiccup current limit, while shorting
tance are calculated for the input peak-to-peak ripple LIM to ground disables the hiccup current-limit circuit.
of 100mV or less yielding an ESR and capacitance
value of 25mΩ and 10µF. Average Current Limit
The average-current-mode control technique of the
For boost regulator designs, the input-capacitor current MAX16818 accurately limits the maximum output current.
waveform is dominated by the inductor, a triangle wave The MAX16818 senses the voltage across the sense
a magnitude of ΔIL. For simplicity’s sake, the current resistor and limit the peak inductor current (I L-PK )
waveform can be approximated by a square wave with accordingly. The on-cycle terminates when the current-
a magnitude that is half that of the triangle wave. sense voltage reaches 25.5mV (min). Use the following
Calculate the input capacitance and ESR required for a equation to calculate the maximum current-sense resis-
specified ripple using the following equation: tor value:
ΔVESR
ESRIN = 0.0255
ΔIL RS =
IOUT
Boost: 0.75 x 10 − 3
PDR =
ΔIL RS
x D
CIN = 2 where PDR is the dissipation in the series resistors.
ΔVQ x fSW Select a 5% lower value of RS to compensate for any
Duty cycle, D, for a boost regulator is equal to (VOUT - parasitics associated with the PCB. Also, select a non-
VIN) / VOUT. As an example, at VIN = 13.2V, VLED = inductive resistor with the appropriate power rating.
15.6V, IOUT = 1A, ΔIL = 0.4A, and fSW = 330kHz, the
Hiccup Current Limit
ESR and input capacitance are calculated for the input
The hiccup current-limit value is always 10% lower than
peak-to-peak ripple of 100mV or less yielding an ESR
the average current-limit threshold, when LIM is left
and capacitance value of 250mΩ and 1µF, respectively.
unconnected. Connect a resistor from LIM to SGND to
Output Capacitor increase the hiccup current-limit value from 90% to
For buck converters, the inductor always connects to 100% of the average current-limit value. The average
the load, so the inductance controls the ripple current. current-limit architecture accurately limits the average
The output capacitance shunts a fraction of this ripple output current to its current-limit threshold. If the hiccup
current and the LED string absorbs the rest. The current limit is programmed to be equal or above the
capacitor reactance (which includes the capacitance average current-limit value, the output current does not
and ESR) and the dynamic impedance of the LED reach the point where the hiccup current limit can trig-
diode string form a conductance divider that splits the ger. Program the hiccup current limit at least 5% below
ripple current between the LEDs and the capacitor. In the average current limit to ensure that the hiccup cur-
many cases, the capacitor is very large as compared to rent-limit circuit triggers during overload. See the
the ESR, and this divider reduces to the ESR and the Hiccup Current Limit vs. R EXT graph in the Typical
LED resistance. Operating Characteristics.
Boost converters place a harsher requirement on the
output capacitors as they must sustain the full load dur-
ing the on-time of the MOSFET and are replenished
during the off-time. The ripple current in this case is the
full load current, and the holdup time is equal to the
duty cycle times the switching period.

22 ______________________________________________________________________________________
 1.5MHz, 30A High-Efficiency, LED Driver
with Rapid LED Current Pulsing
Compensation In order to choose CCF, the external loop gain must be

MAX16818
The main control loop consists of an inner current loop considered. The following equation describes the over-
(inductor current) and an outer LED current loop. The all loop gain for a buck regulator, which is the ratio of a
MAX16818 uses an average current-mode control small-signal change in the output of amplifier CA to the
scheme to regulate the LED current (Figure 7). The VEA output of amplifier CEA:
output provides the controlling voltage for the current External Loop Buck:
source. The inner current loop absorbs the inductor
pole reducing the order of the LED current loop to that ΔVCA R x VIN x AV
of a single-pole system. = S
ΔVCEA VRAMP x sL
The major consideration when designing the current
control loop is making certain that the inductor down- where AV is the gain of the current amplifier (34.5) and
slope (which becomes an upslope at the output of the V RAMP is voltage peak (2V) of the internal ramp.
CEA) does not exceed the internal ramp slope. This is a Multiplying the external loop gain with the CEA amplifier
necessary condition to avoid subharmonic oscillations gain gives the total loop equation and solves for the fre-
similar to those in peak current mode with insufficient quency that yields a gain of 1 results in:
slope compensation. This requires that the gain at the
output of the CEA be limited based on the following Total Loop Buck:
equation (Figure 6):
VIN x fSW
Buck: fCMAX =
2πVOUT
VRAMP × fSW × L
RCF ≤ To be stable, the gain of the CEA amplifier must have a
A V × RS × VOUT × gm zero placed before fCMAX. CCF creates a pole at the
fSW x L origin and the combination of RCF and CCF creates the
RCF ≤ 105 zero. Lower frequency zeros result in less bandwidth,
RS x VOUT
but greater phase margin. The pole created by CCFF
where VRAMP = 2V, gm = 550µs, AV = 34.5. (in conjunction with RCF) is for noise reduction and can
be placed well past the crossover frequency.
Boost:
The following equation describes the external loop gain
for a boost regulator:
VRAMP × fSW × L
RCF ≤ External Loop Boost:
A V × RS × (VOUT − VIN ) × gm
fSW x L ΔVCA R x VOUT x AV
RCF ≤ 105 = S
RS x (VOUT − VIN) ΔVCEA VRAMP x sL
Solving for the gain of the CEA amplifier, To get the total loop gain for a boost regulator, multiply
Buck: the external loop gain with the gain of the CEA amplifier
to arrive at the following:
ΔVCEA V x fSW x L
gm × RCF = = RAMP Total Loop Boost:
ΔVCA VOUT x RS x AV
fSW x VOUT
fCMAX =
Boost: 2π (VOUT − VIN)

ΔVCEA VRAMP x fSW x L

gm × RCF = = As in the buck regulator, the zero created by RCF and
ΔVCA (VOUT − VIN) x RS x AV CCF sits at a frequency lower than fCMAX to maintain
stable operation.

______________________________________________________________________________________ 23
 1.5MHz, 30A High-Efficiency, LED Driver
with Rapid LED Current Pulsing
Power Dissipation 7) Avoid long traces between the VDD bypass capaci-
MAX16818

The TQFN is a thermally enhanced package and can dis- tors, the driver output of the MAX16818, the MOS-
sipate about 2.7W. The high-power package makes the FET gates, and PGND. Minimize the loop formed by
high-frequency, high-current LED driver possible to oper- the VCC bypass capacitors, bootstrap diode, boot-
ate from a 12V or 24V bus. Calculate power dissipation in strap capacitor, the MAX16818, and the upper
the MAX16818 as a product of the input voltage and the MOSFET gate.
total VCC regulator output current (ICC). ICC includes qui- 8) Distribute the power components evenly across the
escent current (IQ) and gate drive current (IDD): board for proper heat dissipation.
PD = VIN x ICC 9) Provide enough copper area at and around the
ICC = IQ + [fSW x (QG1 + QG2)]
switching MOSFETs, inductor, and sense resistors
to aid in thermal dissipation.
where QG1 and QG2 are the total gate charge of the low- 10) Use wide copper traces (2oz) to keep trace induc-
side and high-side external MOSFETs at VGATE = 5V, IQ tance and resistance low to maximize efficiency.
is estimated from the Supply Current (IQ) vs. Frequency Wide traces also cool heat-generating components.
graph in the Typical Operating Characteristics, and fSW
is the switching frequency of the LED driver. For boost
drivers, only consider one gate charge, QG1.
Use the following equation to calculate the maximum Pin Configuration
power dissipation (PDMAX) in the chip at a given ambi-
ent temperature (TA):

EAOUT
TOP VIEW

DIFF
PDMAX = 34.5 x (150 - TA) mW.

CSN
CSP

EAN

CLP

OVI
PCB Layout Guidelines 21 20 19 18 17 16 15
Use the following guidelines to layout the switching SGND 22 14 LIM
voltage regulator:
SENSE- 23 13 V_IOUT
1) Place the IN, V CC , and V DD bypass capacitors
SENSE+ 24 12 RT/SYNC
close to the MAX16818.
SGND 25 11 EN
2) Minimize the area and length of the high current MAX16818
loops from the input capacitor, upper switching IN 26 10 PGOOD
MOSFET, inductor, and output capacitor back to VCC 27 9 CLKOUT
the input capacitor negative terminal. * EXPOSED PAD
VDD 28 8 SGND
3) Keep short the current loop formed by the lower +
switching MOSFET, inductor, and output capacitor. 1 2 3 4 5 6 7

4) Place the Schottky diodes close to the lower

PGND

N.C.

DL

BST

LX

DH

N.C.

MOSFETs and on the same side of the PCB. TQFN

5) Keep the SGND and PGND isolated and connect
them at one single point.
6) Run the current-sense lines CSP and CSN 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 Chip Information
cross these critical signal lines through power cir- TRANSISTOR COUNT: 5654
cuitry. Sense the current right at the pads of the
current-sense resistors. PROCESS: BiCMOS

24 ______________________________________________________________________________________
 1.5MHz, 30A High-Efficiency, LED Driver
with Rapid LED Current Pulsing
Package Information

MAX16818
(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.)

QFN THIN.EPS
PACKAGE OUTLINE,
16, 20, 28, 32, 40L THIN QFN, 5x5x0.8mm
1
21-0140 K 2

______________________________________________________________________________________ 25
 1.5MHz, 30A High-Efficiency, LED Driver
with Rapid LED Current Pulsing
Package Information (continued)
MAX16818

(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.)

PACKAGE OUTLINE,
16, 20, 28, 32, 40L THIN QFN, 5x5x0.8mm
2
21-0140 K 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.

26 ____________________Maxim Integrated Products, 120 San Gabriel Drive, Sunnyvale, CA 94086 408-737-7600

© 2006 Maxim Integrated Products is a registered trademark of Maxim Integrated Products, Inc.

Heaney