19-0821; Rev 0; 5/07 MAX16809 Evaluation Kit The MAX16809 evaluation kit (EV kit) is a 16-channel, constant-current LED driver, capable of driving 40mA each to 16 LED strings with a total forward voltage of up to 32V. The MAX16809 EV kit is based on the MAX16809 device, which has 16 constant-current-sinking outputs with sink current settable using a single resistor and a high-performance, current-mode pulsewidth-modulator (PWM) controller, for implementing a DC-DC converter that generates the supply voltage to drive the LED strings. The MAX16809 EV kit operates at supply voltages between 9V to 16V and temperatures ranging from 0C to +70C. It features a PWM dimming control, adaptive control of the LED supply voltage, which depends upon the operating voltage of the LED strings, a built-in clock generator, and a low-current shutdown. The MAX16809 EV kit is a fully assembled and tested board. Features 9V to 16V Supply Voltage Range 40mA LED Current (Per Each LED String) Single-Resistor Current Adjust for 16 Channels Up to 32V LED String Voltage Boost Converter to Generate LED Supply Voltage Adaptive LED Supply Voltage Control Increases Efficiency PWM Dimming Control Output-Voltage-Spike Protection for InductiveOutput Lines Proven PCB Layout Ordering Information PART TEMP RANGE IC PACKAGE MAX16809EVKIT+ 0C to +70C* 38 TQFN-EP +Denotes a lead-free and RoHS-compliant EV kit. *This limited temperature range applies to the EV kit PCB only. The MAX16809 IC temperature range is -40C to +125C. EP = Exposed paddle. Component List DESIGNATION C1-C4, C6, C7, C8, C10, C11, C19-C23, C25, C26 C5, C24, C30, C31, C33 C9 C12, C13, C14 C15 QTY DESCRIPTION DESIGNATION QTY C16, C17, C18 3 22F 20%, 50V electrolytic capacitors Panasonic EEEFK1H220P 1 560pF 10%, 50V C0G capacitor (0603) KEMET C0603C561K5RACTU Murata GRM188R71H561KA01D 1 10pF 10%, 50V C0G capacitor (0603) TDK C1608C0G1H00DB Murata GRM1885C1H100JA01D 1 220pF 10%, 50V C0G capacitor (0603) KEMET C0603C221K5RACTU Murata GRM188R71H221KA01D 1 100pF 10%, 50V C0G capacitor (0603) KEMET C0603C101K5RACTU Murata GRM188R71H101KA01D 16 1nF 10%, 50V X7R capacitors (0603) Murata GRM188R71H103KA01D KEMET C0603C103K5RACTU C27 5 0.1F 10%, 50V X7R capacitors (0603) Murata GRM188R71H104KA93D TDK C1608X7R1H104K C28 1 1F 20%, 16V X7R capacitor (0805) Murata GRM21BR71C105KA01L TDK C2012X7R1C105K 3 22F 20% 35V electrolytic capacitors Panasonic EEEFK1V220R 1 1F 20%, 50V X7R capacitor (1210) KEMET C3225X7R1H105M Murata GRM32ER71H105KA01L C29 C32 DESCRIPTION ________________________________________________________________ Maxim Integrated Products For pricing, delivery, and ordering information, please contact Maxim/Dallas Direct! at 1-888-629-4642, or visit Maxim's website at www.maxim-ic.com. 1 Evaluates: MAX16809 General Description MAX16809 Evaluation Kit Evaluates: MAX16809 Component List (continued) DESIGNATION QTY C34 1 C35 D1 D2 D3-D10 D11 10nF 10%, 50V X7R capacitor (0603) KEMET C0603C103K5RACTU Murata GRM188R71H103KA01D 0 Not installed, capacitor 1 33V zener diode (SOD323) Diodes Inc. MMSZ5257BS-7 DESCRIPTION Q2 1 60V, 5.5A N-channel MOSFET Vishay Si4450DY R1 1 430 1% resistor (0603) R2 1 330k 1% resistor (0603) 8.45k 1% resistor (0603) R3 1 R4, R6, R10 3 22k 1% resistors (0603) R5 1 180k 1% resistor (0603) R7 1 10.5k 1% resistor (0603) R8 1 10 1% resistor (0603) R9 1 1.2k 1% resistor (0603) 8 15V dual zener diodes (SOT323) Diodes AZ23C15-7-F R10 1 22k 1% resistor (0603) R11 1 50k 1% resistor (0603) 20mA switching diode (SOD323) Diodes Inc. 1N4148WS-7 R12 1 75m 1% resistor (0603) R13 0 Not installed, resistor 40V small-signal Schottky diode (SOD523) Diodes Inc. SDM03U40 R14 1 2.2k 1% resistor (0603) R15 1 10k 1% resistor (0603) U1 1 MAX16809ATU+ (38-pin TQFN, 5mm x 7mm) U2 1 Digital pnp transistor ROHM DTA114WKA U3 1 Digital npn transistor ROHM DTC114WKA U4 1 Dual inverter with hysteresis Texas Instruments SN74LVC2G14DCKR -- 1 PCB: MAX16809 Evaluation Kit+ GND, GND, PWM, SHDN, VBIAS, VIN 6 J1 1 0.1in 20-pin header 1 27H, 3.2A inductor Coilcraft MSS1260-273ML 1 Switching transistor (SOT523) Diodes Inc. MMBT2222AT-7-F Q1 QTY 1A, 40V Schottky diode (SMA) Diodes Inc. CMSH1-40M Central Semiconductor CMSH1-40M 1 L1 DESIGNATION 1 1 D12 DESCRIPTION Wire loops Component Suppliers SUPPLIER PHONE FAX WEBSITE Central Semiconductor Corp. 631-435-1110 631-435-3388 www.centralsemi.com Coilcraft, Inc. 847-639-6400 847-639-1469 www.coilcraft.com Diodes Inc. 805-446-4800 805-446-4850 www.diodes.com KEMET Corp. 978-658-1663 978-658-1790 www.kemet.com Murata Mfg. Co., Ltd. 770-436-1300 770-436-3030 www.murata.com ROHM Co., Ltd. 858-625-3630 858-625-3670 www.rohm.com TDK Corp. 847-390-4373 847-390-4428 -- -- 402-563-6866 402-563-6296 Texas Instruments Inc. Vishay www.component.tdk.com www.ti.com www.vishay.com Note: Indicate that you are using the MAX16809 when contacting these component suppliers. 2 _______________________________________________________________________________________ MAX16809 Evaluation Kit Recommended Equipment * One 16V, 5A adjustable power supply * One 5V power supply * 16 LED strings with a total forward voltage 32V * One multimeter * One PWM signal generator (optional) Procedure The MAX16809 EV kit is fully assembled and tested. Follow the steps below to verify operation. Caution: Do not turn on the power supply until all connections are completed. 1) Connect LED strings with operating voltage of approximately 32V between VLED (pins 1-4 of J1) and OUT0-OUT15 (pins 5-20 of J1). All 16 channels should have an LED string load connected of the same type. 2) Connect the DC power supply (16V, 5A) to VIN and GND. 3) Connect a DC power supply (0 to 5V) to VBIAS and GND. 4) Turn on the power supplies and apply 10V to VIN and 3V to 5V to VBIAS. Connect SHDN and PWM to 3V to 5V. All of the LEDs should turn on. Measure the current through any LED string, which should be 40mA 7%. 5) Increase the supply voltage to 16V and the LED currents will be stable. Measure the current through any LED string, which should be 40mA 7%. 6) Apply a PWM signal with amplitude of 3V to 5V and a frequency between 100Hz and 2kHz to the PWM input. The LED brightness should increase as the PWM duty cycle increases and viceversa. 7) Connect SHDN to GND and all LEDs should turn off. Detailed Description The MAX16809 EV kit is a 16-channel, constant-current LED driver capable of driving 40mA each to 16 LED strings, with a total forward voltage of up to 32V. The MAX16809 EV kit can drive a total of 160 white LEDs in 16 strings, with operating current up to 40mA. The MAX16809 EV kit can operate at input supply voltages between 9V and 16V. The MAX16809 EV kit evaluates the MAX16809 IC, which has two major sections. The first section consists of 16 constant-current LED drivers capable of sinking up to 55mA when on and blocking up to 36V when off. The second section is a high-performance current-mode PWM controller that can control a DC-DC converter to generate the supply voltage for driving the LED strings. The MAX16809 EV kit uses the PWM controller to drive a boost converter, which takes a 9V to 16V input and generates a 33V LED supply voltage. To drive a constant current into an LED string, connect the LED string between the 33V output and any of the 16 constant-current-sink outputs. The resistor (R1) from the SET pin to ground programs the sink current of each output. The sink current of any output can be up to 55mA and the amplitude is the same value for all the outputs. The difference between the total forward voltage and the LED supply voltage drops between the constant-current-sink output and ground, and is dissipated as power in the device. The LED supply voltage generated by the boost converter in the MAX16809 EV kit is adaptive. The LED string with the highest total forward voltage dominates the control loop, and the boost-converter voltage is adjusted so that the driver associated with that string receives just enough voltage required for current drive. All the other strings, with lower total forward voltages, will have excess supply voltage, which is then dropped in the associated driver. This feedback mechanism ensures that the linear-current-control circuit dissipates the minimum possible power. An on-board inverter (U4A) is configured to generate the clock input for the MAX16809. The constant-current output-driver circuits and U4 need a 3.3V to 5V input, which should be supplied externally. If 5V is not available, it can be generated using an emitter-follower buffer from the REF output of MAX16809. Boost Converter The boost converter that generates the 33V LED supply voltage operates at a switching frequency of 350kHz in continuous-conduction mode (CCM). The current-mode PWM controller in the MAX16809 drives the external MOSFET (Q2) to control the boost converter. The MOSFET is turned on at the beginning of every switching cycle and turned off when the current through the inductor (L1) reaches the peak value set by the error-amplifieroutput voltage. Inductor current is sensed from the voltage across the ground-referenced current-sense resistor (parallel combination of R12 and R13). This currentsense information is passed on to the current-sense comparators in the MAX16809 through the CS pin. During the on period of the MOSFET, the inductor stores energy from the input supply. When the switch is turned off, the inductor generates sufficient voltage in reverse direction to discharge the stored energy to VLED. This generated voltage forms a source, in series with the input supply voltage, and drives VLED through the rectifier diode (D2). _______________________________________________________________________________________ 3 Evaluates: MAX16809 Quick Start Evaluates: MAX16809 MAX16809 Evaluation Kit As the boost converter is operated in CCM, only part of the stored energy in the inductor is discharged to VLED. The advantages of CCM include reduced input and output filtering, reduced EMI due to lower peak currents, and higher converter efficiency. However, these advantages come at the cost of a right-half-plane zero in the converter-transfer function. Compensating this zero requires reducing the system bandwidth, which affects the converter-dynamic response. As the 16-channel, constant-current-sink outputs control the current through the LEDs, slower control of VLED does not affect the LED operation. Compensation of the feedback circuit is explained in the Feedback Compensation section. An internal comparator turns off the gate pulse to the external MOSFET if the voltage at the CS pin exceeds 0.3V. The current through the inductor that produces 0.3V at the CS pin is the maximum inductor current possible (the actual current can be a little higher than this limit due to the 60ns propagation delay from the CS pin to the MOSFET drive output). This condition can happen when the feedback loop is broken, when the output capacitor charges during power-up, or when there is an overload at the output. This feature protects the MOSFET by limiting the maximum current passing through it during such conditions. The RC filter, consisting of R9 and C10, removes the voltage spike across the current-sense resistors produced by the turn-on gate current of the MOSFET and the reverse-recovery current of D2. Without filtering, these current spikes can cause sense comparators to falsely trigger and turn off the gate pulse prematurely. The filter time constant should not be higher than required (the MAX16809 EV kit uses a 120ns time constant), as a higher time constant adds additional delay to the current-sense voltage, effectively increasing the current limit. During normal operating conditions, the feedback loop controls the peak current. The error amplifier compares a scaled-down version of the LED supply voltage (VLED) with a highly accurate 2.5V reference. The error amplifier and compensation network then amplify the error signal, and the current comparator compares this signal to the sensed-current voltage to create a PWM drive output. Power-Circuit Design Initially, decide the input supply voltage range, output voltage VLED (the sum of the maximum LED total forward voltage and 1V bias voltage for the constant-current-sink output), and the output current IOUT (the sum of all the LED string currents). 4 Calculate maximum duty cycle DMAX using the following equation: DMAX = VLED + VD - VINMIN VLED + VD - VFET where VD is the forward drop of the rectifier diode D2 (~0.6V), VINMIN is the minimum input supply voltage (in this case, 9V), and VFET is the average drain-to-source voltage of the MOSFET Q2 when it is on. Select the switching frequency F SW based on the space, noise, dynamic response, and efficiency constraints. Select the maximum peak-to-peak ripple on the inductor current ILPP. For the MAX16809 EV kit, FSW is 350kHz and ILPP is 30% of the average inductor current. Use the following equations to calculate the maximum average-inductor current ILAVG and peak inductor current ILPEAK: IL AVG = IOUT 1- DMAX Since ILPP is 30% of the average-inductor current ILAVG: ILPP = IL AVG x 0.3 x 2 ILPEAK = IL AVG + ILPP 2 Calculate the minimum inductance value LMIN with the inductor current ripple set to the maximum value: LMIN = (VINMIN - VFET ) x DMAX FSW x ILPP Choose an inductor that has a minimum inductance greater than this calculated value. Calculate the current-sense resistor (R12 in parallel with R13) using the equation below: RCS = 0.3 x 0.75 ILPEAK where 0.3V is the maximum current-sense signal voltage. The factor 0.75 is for compensating the reduction of maximum current-sense voltage due to the addition of slope compensation. Check this factor and adjust after the slope compensation is calculated. See the Slope Compensation section for more information. _______________________________________________________________________________________ MAX16809 Evaluation Kit The continuous drain-current rating of the selected MOSFET when the case temperature is at +70C should be greater than that calculated by the following equation. The MOSFET must be mounted on a board, as per manufacturer specifications, to dissipate the heat: IL SAT = 1.1 x ILPEAK Calculate the output capacitor COUT (parallel combination of C16, C17, C18, and C24) using the following equation: COUT = DMAX x IOUT VLEDPP x FSW where VLEDPP is the peak-to-peak ripple in the LED supply voltage. The value of the calculated output capacitance will be much lower than what is actually necessary for feedback loop compensation. See the Feedback Compensation section to calculate the output capacitance based on the compensation requirements. Calculate the input capacitor CIN (parallel combination of C12, C13, C14, and C5) using the following equation: ILPP CIN = 8 x FSW x VINPP where VINPP is the peak-to-peak input ripple voltage. This equation assumes that input capacitors supply most of the input ripple current. Selection of Power Semiconductors The switching MOSFET (Q2) should have a voltage rating sufficient to withstand the maximum output voltage, together with the diode drop of D2, and any possible overshoot due to ringing caused by parasitic inductances and capacitances. Use a MOSFET with voltage rating higher than that calculated by the following equation: VDS = (VLED + VD ) x1.3 The factor of 1.3 provides a 30% safety margin. IL AVG2 x 1.3 IDRMS = DMAX The MOSFET dissipates power due to both switching losses, as well as conduction losses. Use the following equation to calculate the conduction losses in the MOSFET: PCOND = IL AVG2 x RDSON DMAX where RDSON is the on-state drain-source resistance of the MOSFET with an assumed junction temperature of 100C. Use the following equation to calculate the switching losses in the MOSFET: PSW = 1 IL AVG x VLED2 x CGD x FSW 1 x + 2 IGON IGOFF where IGON and IGOFF are the gate currents of the MOSFET (with V GS equal to the threshold voltage) when it is turned on and turned off, respectively, and CGD is the gate-to-drain MOSFET capacitance. Choose a MOSFET that has a higher power rating than that calculated by the following equation when the MOSFET case temperature is at +70C: PTOT = PCOND + PSW The MAX16809 EV kit uses a Schottky diode as the boost-converter rectifier (D2). A Schottky rectifier diode produces less forward drop and puts the least burden on the MOSFET during reverse recovery. If a diode with considerable reverse-recovery time is used, it should be considered in the MOSFET switching-loss calculation. The Schottky diode selected should have a voltage rating 20% above the maximum boost-converter output voltage. The current rating of the diode should be greater than ID in the following equation: IL AVG2 ID = 1- DMAX x 1.2 _______________________________________________________________________________________ 5 Evaluates: MAX16809 The saturation current limit of the selected inductor (ILSAT) should be greater than the value given by the equation below. Selecting an inductor with 10% higher ILSAT rating is a good choice: Evaluates: MAX16809 MAX16809 Evaluation Kit Slope Compensation When the boost converter operates in CCM with more than 50% duty cycle, subharmonic oscillations occur if slope compensation is not implemented. Subharmonic oscillations do not allow the PWM duty cycle to settle to a peak current value set by the voltage-feedback loop. The duty cycle oscillates back and forth about the required value, usually at half the switching frequency. Subharmonic oscillations die out if a sufficient negative slope is added to the inductor peak current. This means that for any peak current set by the feedback loop, the output pulse terminates sooner than normally expected. The minimum slope compensation that should be added to stabilize the current loop is half of the worst-case (max) falling slope of inductor current. Adding a ramp to the current-sense signal, with positive slope in sync with the switching frequency, can produce the desired function. The greater the duty cycle, the greater the added voltage, and the greater the difference between the set current and the actual inductor current. In the MAX16809 EV kit, the oscillator ramp signal is buffered using Q1 and added to the current-sense signal with proper scaling to implement the slope compensation. Follow the steps below to calculate the component values for slope compensation. Calculate the worst-case falling slope of the inductor current using the following equation: IL SLOPE = (VLEDMAX + VD - VINMIN ) LMIN From the inductor current falling slope, find its equivalent voltage slope across the current-sense resistor RCS (R12 parallel with R13) using the following equation: VSLOPE = IL SLOPE x RCS The minimum voltage slope that should be added to the current-sense waveform is half of VSLOPE for ensuring stability up to 100% duty cycle. As the maximum continuous duty cycle used is less than 100%, the minimum required compensation slope becomes: V x (2DMAX - 1) x 1.1 VCSLOPE = SLOPE DMAX The factor 1.1 provides a 10% margin. Resistors R9 and R10 determine the attenuation of the buffered voltage slope from the emitter of Q1. The forward drop of 6 signal diode D11, together with the VBE of Q1, almost cancel the 1.1V offset of the ramp waveform. Calculate the approximate slope of the oscillator ramp using the following equation: VRSLOPE = 1.7 x FSW where 1.7V is the ramp amplitude and F SW is the switching frequency. Select the value of R9 such that the input bias current of the current-sense comparators does not add considerable error to the current-sense signal. The value of R10 for the slope compensation is given by the equation: VRSLOPE R10 = - 1 x R9 VCSLOPE LED Driver The MAX16809 features a 16-channel, constant-current LED driver, with each channel capable of sinking up to 55mA of LED current. The LED strings are connected between VLED and the constant-current-sink outputs to drive regulated current through LED strings. The current through all 16 channels is controlled through resistor (R1) from the SET pin to ground. The MAX16809 EV kit sets the current through each string at 40mA and the maximum LED supply voltage to 33V. The MAX16809 EV kit drives LED strings with a total forward voltage of up to 32V. A 4-wire serial interface with four inputs (DIN, CLK, LE, and OE) individually control the constant-current outputs. In the MAX16809 EV kit, a 50kHz clock signal, generated by U4A, clocks 16 1s into the internal shift register by tying DIN and LE to 5V. The clock-generation circuit can be avoided if a microcontroller provides the function. The output enable (OE) can provide PWM dimming. An inverted PWM signal, generated by the inverter U4B, is necessary to drive the OE pin. When the PWM signal is low (LED drivers off), it also influences the feedback with the network formed by R6 and D12. See the Adaptive LED Supply Voltage Control section for more details. If an inverted PWM signal is available, use the circuit shown in Figure1 to drive the OE input and feedback network. _______________________________________________________________________________________ MAX16809 Evaluation Kit Evaluates: MAX16809 COMP C28 10pF R5 180k C29 220pF VLED R2 330k R11 50k D1 33V R7 10.5k FB 3 D10 AZ23C15-7 3 D3 AZ23C15-7 R6 22k VBIAS C30 0.1F 5 4 3 U4B R15 10k SN74LVC2G14 D12 SDM03U40 1 2 1 2 2 OUT0 OUT1 PWM OUT14 OUT15 OEB Figure 1. Inverting PWM Drive Circuit Output Current Setting The amplitude of the output sink currents for all 16 channels is set to the same value by the resistor (R1) from the SET pin to ground. The minimum allowed value of RSET is 311, which sets the output currents to 55mA. The maximum allowed value of RSET is 5k. The MAX16809 EV kit uses 430 for RSET, which sets the output current to 40mA. To set a different output current, use the following equation: RSET = 17100 IOUT where RSET is the current-setting resistor (R1) value in ohms and IOUT is the desired output current in milliamps. Adaptive LED Supply Voltage Control To reduce power dissipation in the IC, the MAX16809 EV kit features adaptive control of VLED based on the operating voltage of the LED strings. The constant-current-sink outputs can sink stable currents with output voltages as low as 0.8V. The voltage at each of the 16 outputs will be the difference between VLED and the total forward voltage of the LED string connected to that output. The MAX16809 EV kit implements a feedback mechanism to sense the voltage at each of the 16 constant-current-sink outputs. Using dual zener diodes (D3-D10), the MAX16809 EV kit selects the lowest driver voltage (with the greatest LED string voltage) to regulate. The boost-converter PWM then adjusts so that VLED is high enough for this sink output to settle to approximately 0.8V. All the other strings have sufficient voltage, as their total forward voltages are lower. The feedback mechanism ensures that the IC dissipates the minimum possible power. For adaptive control to function efficiently connect LED strings to all 16 channels and use an equal number of LEDs from the same bin in each string. If some of the 16 channels are not used, then the zener diodes (D3-D10) should be removed from the unused channels. Use the equation below to calculate the value of R2 to get the required minimum voltage at the sink outputs: R2 = (VFLED + VS - 2.5) x R7 2.5 - VDZ - VS where 2.5V is the feedback reference, VDZ is the forward drop of the ORing diode (D3-D10), VS = 0.5V is the required sink-output voltage, and VFLED is the nominal total forward voltage of the LED strings. Select the value of R2 such that R7 is approximately 10k. The zener diodes (D3-D10) also provide output overvoltage protection. If an LED string gets partially or fully shorted, making the sink-output voltage go high, the 15V zener diode connected to that output conducts in reverse direction, and limits the VLED voltage. Under this condition, the other LED strings might not turn on. When the outputs are off, the LED drivers are at high impedance and the feedback network now combines R6 and D12 to provide a path for the feedback current and to control VLED. Use the following equation to _______________________________________________________________________________________ 7 Evaluates: MAX16809 MAX16809 Evaluation Kit calculate the value of R6 to get the required LED supply voltage during PWM off time: R6 = R2 x (2.5 - 0.4) VLEDOFF - 2.5 where 2.5V is the feedback-reference voltage, 0.4V is the total voltage dropped by D4 and PWM input, and VLED OFF is the desired LED supply voltage during PWM off time. VLEDOFF should be set to the worst-case LED string voltage plus some additional headroom for the LED drivers (0.8V), as well as a reserve voltage (approximately 1V). The reserve voltage allows the MAX16809 to provide current for very short PWM dimming on-pulses. With pulses as low as 2s, the VLED control loop is not able to react, and the output capacitors provide all the current. For longer PWM dimming pulses, the control loop reacts and the supply operates at the adaptive voltage level. During an open LED condition, the 33V zener diode (D1) limits the maximum LED supply voltage to 35.5V. If VLED attempts to increase beyond this level, D1 conducts in reverse direction and pulls the FB pin high, which causes the boost regulator to cut back on the PWM signal and reduce the output voltage. PWM Dimming The PWM dimming controls the LED brightness by adjusting the duty cycle of the PWM input signal. A high voltage at the PWM input enables the output current; a low voltage turns off the output current. Connect a signal with peak amplitude of 3V to 5V and with frequency from 100Hz to 2kHz to the PWM input and vary the duty cycle to adjust the LED brightness. The LED brightness increases when the duty cycle increases and vice versa. If an inverted PWM signal is available, use that signal to implement PWM dimming, as shown in Figure 1. Feedback Compensation Like any other circuit with feedback, the boost converter that generates the supply voltage for the LED strings needs to be compensated for stable control of its output voltage. As the boost converter is operated in continuous-conduction mode, there exists a right-halfplane (RHP) zero in the power-circuit transfer function. This zero adds a 20dB/decade gain together with a 90degree phase lag, which is difficult to compensate. The easiest way to avoid this zero is to roll off the loop gain 8 to 0dB at a frequency less than half of the RHP zero frequency with a -20dB/decade slope. For a boost converter, the worst-case RHP zero frequency (FZRHP) is given by the following equation: FZRHP = VLED(1- DMAX )2 2 x L x IO where DMAX is the maximum duty cycle, L is the inductance of the inductor, and I O is the output current, which is the sum of all the LED string currents. The boost converter used in the MAX16809 EV kit is operated with current-mode control. There are two feedback loops within a current-mode-controlled converter: an inner loop that controls the inductor current and an outer loop that controls the output voltage. The amplified voltage error produced by the outer voltage loop is the input to the inner current loop that controls the peak inductor current. The internal current loop converts the double-pole 2ndorder system, formed by the inductor and the output capacitor COUT, to a 1st-order system having a single pole consisting of the output filter capacitor and the output load. As the output load is a constant current (i.e., very high Thevenin impedance), this pole is located near the origin (0Hz). The phase lag created by the output pole for any frequency will be 90 degrees. Since the power-circuit DC gain is limited by other factors, the gain starts falling at -20dB/decade from a non-zero frequency before which the power-circuit gain stabilizes. Total gain of the feedback loop at DC is given by the following equation: GTOT = GP x GEA x GFB where GP is the power-circuit DC gain, and GEA is the error-amplifier open-loop DC gain, typically 100dB. GFB is the gain of the feedback network for adaptive control of the VLED, which is seen from VLED to the erroramplifier input (FB pin). The adaptive control senses the voltages at the 16 constant-current-sink outputs and adjusts the feedback to control these voltages to a minimum value (Figure 2). As the LEDs carry constant current, the voltage across the LEDs does not change with variations in VLED. Any change in VLED directly reflects to the constant-current-sink outputs and to the error-amplifier input, making GFB equal to unity. _______________________________________________________________________________________ MAX16809 Evaluation Kit GP = VLED EA OUT Calculate the power-circuit DC gain using the following equation: 1 GP = 2 VIN I + O x RCS x 3 2 V IN 2 x L x FSW x VLED where RCS is the current-sense resistor, F SW is the switching frequency, and the factor 3 is to account for the attenuation of error-amp output before it is fed to the current-sense comparator. The power-circuit gain is lowest at the minimum input supply voltage and highest at the maximum input supply voltage. Any input supply voltage between 9V and 16V can be used for power-circuit gain calculation, as the final compensation values obtained are the same. Calculate the frequency FP2, at which the power-circuit gain starts falling, at -20dB/decade using the following equation: FP2 = (1- DMAX ) 2 x COUT x 3 x RCS x GP where COUT is the output filter capacitor, which is the parallel combination of C16, C17, C18, and C24. Adjust the output capacitance so that the product of FP2 and GP is below FZRHP / 6. The value of output capacitance obtained this way will be much greater than the value obtained using the maximum output voltage ripple specification. The compensation strategy is as follows. The gain-frequency response of the feedback loop should cross 0dB at or below half of the RHP zero frequency, with a slope of -20dB/decade for the feedback to be stable and have sufficient phase margin. The compensation network from COMP pin to FB pin of the MAX16809 (formed by R5, C28, C29, and R11) offers one dominant pole (P1), a zero (Z1), and a high-frequency pole (P3). There are two very low frequency poles and a zero in the loop before the crossover frequency. The function of the zero (Z1) is to compensate for the output pole and to reduce the slope of the loop gain from -40dB/decade to -20dB/decade, and also to reduce the phase lag by 90 degrees. Choose the crossover frequency to be half of the worstcase RHP zero frequency: FC = FZRHP 2 Place the zero (Z1) at one-third of the crossover frequency, so that the phase margin starts improving from a sufficiently lower frequency: FZ1 = FC 3 Use the following equation to calculate the dominant pole location, so that the loop gain crosses 0dB at FC: FP1 = FZRHP x FZ1 2 x GTOT x FP2 Since the open-loop gain of the error amplifier can have variations, the dominant pole location can also vary from device to device. In the MAX16809 EV kit, the dominant pole location is decided by the error-amplifier gain, so the combined effect is a constant-gain-bandwidth product. Select the value of R11 such that the input bias current of the error amplifier does not cause considerable drop across it. The effective AC impedance seen from the FB pin is the sum of R11 and R7. It is preferable to keep R7 much lower, compared to R11, to have better control on the AC impedance. Find C29 using the following equation: C29 = 1 2 x GEA x (R11+ R7) x FP1 The location of the zero (Z1) decided by R5 and C29 is given by the following equation: FZ1 = 1 2 x R5 x C29 Place the high-frequency pole (P3), formed by C28, C29, and R5, at half the switching frequency to provide further attenuation to any high-frequency signal propagating through the system. The location of the high-frequency pole (FP3) is given by the following equation, and should be used to calculate the value of C28: FP3 = 1 1 1 + 2 x R5 x C28 C29 -1 _______________________________________________________________________________________ 9 Evaluates: MAX16809 The DC gain of the power circuit is expressed as the change in the output voltage, with respect to the change in error-amplifier output voltage. As the boost converter in the MAX16809 EV kit drives a constantcurrent load, the power-circuit DC gain is calculated based on a constant-current load: Evaluates: MAX16809 MAX16809 Evaluation Kit The MAX16809 EV kit uses electrolytic capacitors at the output for filtering, so the zero produced by the ESR of the capacitors can be low enough to be within or near the crossover frequency. This zero should be compensated using an additional pole (P4) placed at the ESR zero location. The ESR zero frequency is calculated using the following equation: 1 FZESR = 2 x ESR x COUT Use the following equation to calculate the value of C35 to place the pole (P4) at the ESR zero frequency: 1 C35 = 2 x FZESR x R7 If ceramic capacitors are used at the output for filtering, the frequency of zero produced by the ESR and the capacitance will be above the crossover frequency (0dB gain frequency) of the feedback loop and need not be considered in the compensation design. Layout Considerations LED driver circuits based on the MAX16809 device use a high-frequency switching converter to generate the supply voltage for LED strings. Proper care must be taken while laying out the circuit to ensure proper operation. The switching-converter part of the circuit has nodes with very fast voltage changes, producing highfrequency electric fields, and branches with fast current changes, producing high-frequency magnetic fields. As the circuit converts power, the amplitude of these fields will be high and can easily couple to sensitive parts of the circuit, creating undesirable effects. Follow the guidelines below to reduce noise as much as possible: 1) Connect the bypass capacitors from REF and VCC as close as possible to the device and connect the capacitor grounds to the analog ground plane using vias close to the capacitor terminals. Connect the AGND pin of the device to the analog ground plane using a via close to the pin. Lay the analog ground plane on the inner layer, preferably next to the top layer. Use the analog ground plane to cover the entire area under critical-signal components for the power converter. 2) Keep the oscillator timing capacitor and resistor very close to the RTCT pin and make the connection as short as possible. Connect the ground of the timing capacitor to the analog ground plane using a via close to the capacitor terminal. Make sure that no switching node is present near the RTCT node and keep the area of the copper connected to the pin small. Keep the REF connection to the timing resistor short and away from any switching node. 10 3) Have a power ground plane for the switching-converter power circuit under the power components (input filter capacitor, output filter capacitor, inductor, MOSFET, rectifier diode, and current-sense resistor). Connect all the ground connections to the power ground plane using vias close to the terminals. 4) There are two loops in the power circuit that carry high-frequency switching currents. One loop is when the MOSFET is on (from the input filter capacitor positive terminal, through the inductor, the MOSFET, and the current-sense resistor, to the input capacitor negative terminal). The other loop is when the MOSFET is off (from the input capacitor positive terminal, through the inductor, the rectifier diode, and the output filter capacitor, to the input capacitor negative terminal). Analyze these two loops and make the loop areas as small as possible. Wherever possible, have a return path on the power ground plane for the switching currents on the top-layer copper tracks, or through power components. This reduces the loop area considerably and provide a low inductance path for the switching currents. Reducing the loop area also reduces radiation during switching. 5) The gate-drive current of the MOSFET is another high-frequency switching current to consider. There are two major loops: one during the MOSFET turn-on edge and the second during the turn-off edge. The MOSFET turn-on loop is from the VCC bypass capacitor positive terminal, through the MOSFET driver in the device, the gate-drive resistor, the MOSFET gate to source (CGS and CGD), and the currentsense resistor to the VCC bypass capacitor negative terminal. There is no direct path for the current from the current-sense resistor to return to the VCC bypass capacitor through the ground plane, as the VCC bypass capacitor is connected to the analog ground plane and the current-sense resistor is connected to the power ground plane. The best solution is to connect the analog ground plane to the power ground plane directly under the MOSFET gate-drive track. This ensures that the turn-off current also has a return path on the ground plane. 6) The drain node of the MOSFET is a switching node. Keep this node area small to reduce radiation and capacitive coupling to other sensitive parts of the circuit. However, the track should be wide enough to carry the large switching currents. 7) Keep the node area and track length on the FB pin small to reduce any noise pickup. ______________________________________________________________________________________ MAX16809 Evaluation Kit Power Dissipation The MAX16809 dissipates power during normal operating conditions. The heat transferred to the exposed pad from the die should be properly dissipated to the board to prevent the device from entering into thermal shutdown. The exposed pad land area on the top layer should be of the same size as that of the exposed pad. Thermal vias are used to carry the heat from the exposed pad to other layers of the board and spread it across the board area through copper planes. Thermal vias should have a 0.4mm hole size and should be placed at a distance of 1mm from center to center. For a four-layer board, these vias should be connected to the bottom ground plane and to one internal ground plane. Do not use thermal relief for the thermal vias; instead, use solid copper to get the minimum thermal resistance. Use the following equation to calculate the total power dissipated in the MAX16809 device during normal operation: PD = VSN x IN + IB x VIN N = 0 - 15 where IN is the LED current in channel N, VS is the operating voltage of each of the LED driver outputs with respect to GND pins, IB is the input bias current of MAX16809 including the average of MOSFET drive current, and VIN is the input supply voltage. To dissipate 1W of power, the exposed pad of the device should be connected to a minimum of two square inches of copper ground plane with 70 copper thickness. ______________________________________________________________________________________ 11 Evaluates: MAX16809 8) Connect the power ground plane for the constantcurrent LED driver part of the circuit to the boostconverter output filter capacitor negative terminal. 12 G2 SHDN PWM GND VBIAS GND VIN VBIAS C12 22F 35V C13 22F 35V VIN C14 22F 35V EEEFK1V220R R12 75m 1% R13 OPEN 3 2 1 Q2 Si4450DY 5 6 7 8 C15 1F 50V 4 C32 100pF R9 1.2k 1% R4 22k R8 1% 10 1% D2 CMSH1-40M C17 22F 50V C34 10nF R10 22k 1% D11 1N4148 2 1 C18 22F 50V 2 - 5 + C27 560pF R3 8.45k 1% 1 U4A R14 2.2k 1% Q1 MMBT2222 C16 22F 50V EEEFK1H220P C24 L1 0.1F 27H 50V MSS1278 G1 G1 SN74LVC2G14 6 23 OUT2 26 27 28 OUT5 OUT6 OUT7 R15 10k 1% 31 30 29 25 OUT4 24 22 OUT3 21 OUT1 20 OUT0 CLK C33 0.1F VLED 18 17 16 15 14 G1 13 U1 MAX16809 2 5 32 - + 33 36 C30 0.1F 35 SN74LVC2G14 4 34 OEB 37 OE G2 G1 1 2 3 4 5 6 7 8 9 VIN OUT8 OUT9 OUT10 OUT11 OUT12 OUT13 10 OUT14 11 OUT15 12 EP C31 0.1F 38 N.C. FB COMP OUT8 OUT9 OUT10 OUT11 OUT12 OUT13 OUT14 OUT15 N.C. AGND OUT VCC N.C. REF N.C. 3 U4B N.C. CS RTCT OUT7 OUT6 OUT5 OUT4 OUT3 OUT2 OUT1 OUT0 LE EP R1 430 1% CLK DIN PGND PGND V+ SET DOUT 19 CLK OUT C9 1F GND 1 OUT 3 IN 2 U2 C35 OPEN VLED DTA114WKA G1 VBIAS R5 180k 1% R11 C29 50k 220pF 1% OE DTC114WKA 2 IN U3 1 GND 3 C28 10pF C5 0.1F D12 SDM03U40 R6 22k 1% R2 330k 1% R7 10.5k 1% D1 33V 1 2 2 1 1 2 2 1 1 2 3 D6 AZ23C15-7 6 1 2 3 4 5 9 AZ23C15-7 D8 AZ23C15-7 D9 AZ23C15-7 D10 20 AZ23C15-7 D7 19 OUT8 18 17 16 15 14 13 12 OUT9 OUT10 OUT11 OUT12 OUT13 OUT14 OUT15 OUT0 11 OUT2 OUT1 10 OUT3 8 2 1 3 D5 AZ23C15-7 C23 1nF C6 1nF C1 1nF 7 1 2 3 D4 AZ23C15-7 C26 1nF C22 1nF C11 1nF C8 1nF C3 1nF OUT4 2 1 C25 1nF 3 D3 AZ23C15-7 C20 1nF C21 1nF C19 1nF C10 1nF C7 1nF C2 1nF OUT5 OUT6 OUT7 OUT0 OUT1 OUT2 OUT3 OUT4 OUT5 OUT6 OUT7 OUT8 OUT9 OUT10 OUT11 OUT12 OUT13 OUT14 OUT15 VLED VLED J1A C4 1nF Evaluates: MAX16809 MAX16809 Evaluation Kit Figure 2. MAX16809 EV Kit Schematic ______________________________________________________________________________________ MAX16809 Evaluation Kit Evaluates: MAX16809 Figure 3. MAX16809 EV Kit Component Placement Guide--Component Side ______________________________________________________________________________________ 13 Evaluates: MAX16809 MAX16809 Evaluation Kit Figure 4. MAX16809 EV Kit PCB Layout--Component Side 14 ______________________________________________________________________________________ MAX16809 Evaluation Kit Evaluates: MAX16809 Figure 5. MAX16809 EV Kit PCB Layout--Inner Layer 1 ______________________________________________________________________________________ 15 Evaluates: MAX16809 MAX16809 Evaluation Kit Figure 6. MAX16809 EV Kit PCB Layout--Inner Layer 2 16 ______________________________________________________________________________________ MAX16809 Evaluation Kit Evaluates: MAX16809 Figure 7. MAX16809 EV Kit PCB Layout--Solder Side Maxim cannot assume responsibility for use of any circuitry other than circuitry entirely embodied in a Maxim product. No circuit patent licenses are implied. Maxim reserves the right to change the circuitry and specifications without notice at any time. Maxim Integrated Products, 120 San Gabriel Drive, Sunnyvale, CA 94086 408-737-7600 ____________________ 17 (c) 2007 Maxim Integrated Products is a registered trademark of Maxim Integrated Products, Inc.