【Technical Viewpoint】How to design a high-efficiency LED constant current drive circuit

This paper proposes a wide voltage input, high efficiency, high dimming ratio LED constant current drive circuit. In the hysteresis current control mode, the circuit has the advantages of simple structure, fast dynamic response, and no need for compensation circuit. Through the external pin, LED switching, analog dimming and PWM dimming can be conveniently performed. The LED constant current drive circuit is based on CSMC's 1μm40VCDMOS process and is simulated and verified using HSPICE. The results show that within the input voltage range of 8 to 30V, the circuit output current can reach a maximum of 1.2A, the output current accuracy can be controlled within 5.5%, and the power efficiency can be as high as 97%.

　　introduction

　　With the development of LED technology, high-power LEDs have been widely used in lighting decoration and lighting, and power LED driver chips are becoming more and more important. Since the brightness output of LEDs is proportional to the current passing through the LEDs, in order to ensure the consistency of brightness and color of each LED, it is necessary to design a constant current driver to make the LED current as consistent as possible.

　　Based on the characteristics of LED light emission, this paper designs a wide voltage input, high current, high dimming ratio LED constant current driver chip. The chip adopts hysteresis current control mode and can be used to drive one or more series LEDs. In the wide input voltage range of 6V~30V, the LED average current is set by sampling the high-end current. The chip output current accuracy is controlled at 5.5%. At the same time, the chip can realize analog dimming and PWM dimming through the DIM pin. The optimized chip response speed can enable the chip to achieve a very high dimming ratio.

　　This article first analyzes the overall circuit, then introduces the design of each important sub-module, and finally gives the overall simulation waveform, layout and conclusion of the chip.

　　Circuit system principle

　　Figure 1 is the overall architecture of the chip and a typical application circuit diagram.

　　The circuit includes bandgap reference, voltage regulator, high-end current sampling, hysteresis comparator, power tube M1, PWM and analog dimming modules. In addition, the chip also has built-in undervoltage and over-temperature protection circuits, which can effectively ensure the stable operation of the system under various adverse conditions.

Figure 1 Chip overall equivalent architecture diagram

　　From Figure 1, we can see that the inductor L, current sampling resistor RS, and freewheeling diode D1 form a self-oscillating continuous inductor current mode constant current LED controller. This chip adopts hysteresis current control mode, because the change of LED drive current is reflected in the change of voltage difference across RS. Therefore, when the circuit is working normally, the current in the LED is sampled through the sampling resistor RS and converted into a certain proportion of sampling voltage VCS, and then VCS enters the hysteresis comparator, and is compared with the bias voltage generated by the BIAS module to generate a PWM control signal, which is then controlled by the gate drive circuit to turn on and off the power switch tube.

　　The working principle of the circuit is analyzed in detail below. First of all, the chip is designed with two current thresholds IMAX and IMIN. When the power supply VIN is powered on, the initial current of the inductor L and the current sampling resistor RS is zero, and the LED current is also zero. At this time, the output of the CS_COMP hysteresis comparator is high, the built-in power NMOS switch tube M1 is turned on, the potential of the SW terminal is low, and the current flowing through the LED begins to rise. The current flows from VIN to ground through the inductor L, the current sampling resistor RS, the LED and the internal power switch. At this time, the current rise slope is determined by VIN, the inductor (L), and the LED voltage drop. When the LED current increases to the preset value IMAX, the output of the CS_COMP hysteresis comparator is low, and the power switch tube M1 is turned off. Due to the continuity of the inductor current, the current flows through the inductor (L), the current sampling resistor (RS), the LED and the freewheeling Schottky diode (D1) at another downward slope. When the current drops to another preset value IMIN, the power switch is turned on again, the power supply charges the inductor L, and the LED current begins to increase again. When the current increases to IMAX, the control circuit turns off the power tube and repeats the action of the previous cycle. In this way, the hysteresis control of the LED current is completed, making the average current of the LED constant. From the above analysis, it can be seen that the average drive current of the LED is determined by the internal thresholds IMAX and IMIN, so there is no feedback loop similar to the peak current control mode. Therefore, compared with the peak current control mode, the hysteresis current control mode has self-stability and does not require a compensation circuit. In addition, the peak current detection mode dynamic response adjustment generally takes several cycles, while the hysteresis current control can stabilize the dynamic response of the system in at most one cycle, so the dynamic response of the hysteresis current control is faster. Of course, the hysteresis current control mode has disadvantages such as large output ripple and variable frequency control easily produces variable frequency noise. However, in high-power LED lighting drive applications, certain ripple changes and switching frequency changes will not have a significant impact on the overall lighting performance of the LED.

　　Circuit submodule design

　　1. Bandgap

　　Figure 2 shows a CMOS self-biased reference circuit that uses a common source and common gate current mirror to improve power supply rejection and initial accuracy. Among them, R1 and PH4 form a startup circuit. When the power is turned on, if the circuit is in a zero current state, VA is low at this time, MOS tube PH4 is turned on, and current is injected into the reference core circuit, so that the reference circuit is free from the zero degenerate bias point. When the circuit works normally, by reasonably setting the width-to-length ratio of P7 and P8, they are both in the deep linear region. Since the resistance values ​​of R2 and R3 are large, the size of VA is close to the input voltage at this time, and the MOS tube PH4 is turned off, and the startup ends. In addition, since the voltage of VA is close to the power supply voltage, after the voltage is divided by resistors R2 and R3, the voltage VB can represent the power supply voltage, so when the power supply voltage is lower than the set value, an undervoltage signal is output to turn off the power tube, which plays the role of undervoltage protection.

Figure 2 Bandgap reference voltage source circuit diagram

　　Since the input voltage of the reference circuit can reach up to 30V, and the drain-source and gate withstand voltage of ordinary MOS tubes is 5V. In order to make the current mirror more matched, P1, P2, P5, and P7 must use ordinary MOS tubes. Therefore, in order to prevent the tubes from being broken down at high voltage, thick gate oxide MOS tubes with gate-drain short-circuited between the drain and source of these tubes need to be added as protection tubes, namely PH1, PH2, and PH3.

　　2. Hysteresis Comparator (CS_COMP)

　　Figure 3 is the equivalent circuit diagram of the hysteresis comparator, where VTH_H and VTH_L are the bias reference voltages provided by the BIAS module, and CS is the sampling voltage provided by the current sampling module. The current sampling and hysteresis comparator modules are the core modules of the chip, and the hysteresis current control can be well realized through these two modules.

Figure 3 Hysteresis comparator equivalent circuit diagram

　　When the circuit is working, the high-end current sampling module samples the output current and converts it into a sampling voltage CS according to a certain ratio. When the CS voltage is greater than VTH_H, P_OFF ​​is high, P_ON is low, M1 is off and M2 is on. At this time, VTH_L is input to the negative end of COMP1_G, and since P_ON is low at this time, the power tube is turned off, the LED current begins to decrease, and the sampling voltage also begins to decrease. When the CS voltage is less than VTH_L, P_OFF ​​is low, P_ON is high, M1 is turned on, M2 is turned off, and VTH_H is input to the negative end of COMP_G. At this time, P_ON is high, the power tube is turned on, the LED current begins to increase, and the sampling voltage also begins to increase. When the CS voltage is greater than VTH_H, the hysteresis comparator module will repeat the action of the previous cycle. In this way, a square wave with a certain duty cycle can be generated through the hysteresis comparator to control the power switch tube to turn on and off, thereby effectively controlling the current size of the external LED.

　　In addition, the high-end current sampling and hysteresis comparator modules need to have a higher unit gain bandwidth GBW to increase the speed of current sampling and hysteresis comparison, which can reduce circuit delays, increase the response speed of the chip, and also improve the chip output current accuracy. 3. Analog and PWM dimming (DIM)

It is usually hoped that the brightness of LED can be adjusted at any time 　　in different applications and environments as the application and environment change, which requires the LED driver to have a dimming function. At present, the most commonly used LED dimming methods are: analog dimming, PWM dimming, digital dimming and other methods.

　　Analog dimming adjusts the brightness of LED by linearly changing the output current of LED driver. Its advantage is that it can avoid the noise and other problems caused by PWM or digital dimming. The disadvantage is that analog dimming will change the driving current of LED, thus causing color deviation of LED. PWM dimming is to repeatedly switch on and off the LED driver, output current during the period when PWM signal is enabled, and turn off the LED driver at other times. Dimming can be achieved by adjusting the duty cycle of PWM signal. The principle of PWM dimming is to use the 'visual persistence' effect of human eyes, but in order to avoid the human eye to see the flicker of LED, the frequency of PWM dimming should be above 100Hz.

　　Since the average LED current will not be changed, PWM dimming will not change the LED color.

Figure 4 Analog dimming equivalent circuit diagram

　　Figure 4 shows the analog dimming equivalent circuit diagram. Figure 4 is a differential input structure. The input V1 is a fixed level of 2.5V, and V2 is the level of the DIM pin input after resistor voltage division. Since this circuit only works under large signal conditions, its large signal is analyzed first. The current mirror composed of N1 and N2 tubes forces the currents of the two paths to be equal, then:

　　When the voltage is greater than V1, since the voltage at point L2 is low, N3 and N4 are cut off. The output Io is zero and there is no dimming effect. When V2 decreases to 2.5V, the currents on both sides are equal and the output is also zero. At this time, if V2 decreases from 2.5V by ΔV, it can be seen from formula (3) that the difference between the voltages L1 and L3 increases by ΔV. The current generated by the voltage difference on the resistor is mirrored by N3 and N4 to obtain the output current Io. This current will enter the current sampling module and affect the size of the current sampling voltage CS, thereby changing the output current.

　　Figure 5 shows the simulation diagram of the chip analog dimming process. It can be seen from the figure that when the DIM pin voltage gradually decreases, the LED average current IL also begins to decrease at a certain ratio. When the DIM pin voltage is lower than 0.3V, the power tube is turned off and the LED current drops to zero. This shows that the analog dimming module can well control the LED drive current.

Figure 5 Simulation diagram of dimming process

　　Figure 6 shows the equivalent circuit diagram of PWM dimming. By adding a PWM signal with a variable duty cycle to the DIM pin, the output current can be changed to achieve PWM dimming.

Figure 6 PWM dimming equivalent circuit diagram　　In Figure 6, when DIM changes from high to low and is less than VT_L, the enable changes EN to high. At this time, VT is selected as VT_H. When DIM changes from low to high and is higher than VT_H, the conversion is enabled and a certain voltage hysteresis is achieved. If the input signal is a PWM signal, the above working process is also passed, so that the EN output is also a PWM signal, which controls the switch of the internal power tube, thereby achieving the purpose of controlling the output current.

　　Figure 7 shows the output current waveform when the DIM pin inputs a PWM square wave with a typical value of 20kHz and a duty cycle of 50%. From the figure, it can be seen that when a square wave with a certain duty cycle is input to the DIM pin, the average current of the LED is proportional to the duty cycle of the PWM square wave. Therefore, by setting the duty cycle of the PWM square wave, the size of the average current of the LED can be changed.

　Figure 7 PWM dimming waveform

　　It can also be seen from the above figure that when outputting an inductor current cycle, the PWM square wave has the minimum duty cycle, which is about 4%. At this time, the maximum dimming ratio is 25:1. Obviously, the longer the cycle and the lower the frequency of the PWM square wave for digital dimming, the higher the dimming ratio obtained. However, considering the visual persistence effect of the human eye, in order to prevent the output LED current frequency from being too low and causing flickering, the minimum fDIM=100Hz is generally set in application. At this time, the maximum dimming ratio can be as high as 5000:1.

　　Simulation Results

　　Based on the 1μm40VCSMC process model, this paper uses HSPICE software to simulate and verify the overall chip.

　　Table 1 shows the LED output current accuracy under typical conditions, when the sampling resistor RS = 0.33ohm, the inductor L = 100μH, different power supply voltages, and different numbers of LED connections. The chip will cause errors in the output current due to factors such as sampling delay, sampling accuracy, and driver level delay. Under different power supply voltages and load conditions, it can be seen from Table 1 that the output current accuracy can be well controlled within 5.5%. At the same time, it can also be seen that to achieve better current accuracy, the corresponding power supply voltage needs to match it under a fixed load.

Table 1 Output current accuracy

　　Conclusion

　　Based on the 1μm40VCSMC high-voltage process, this paper designs a wide voltage input, high current, high dimming ratio LED constant current buck driver chip. In the hysteresis current control mode, the chip has the advantages of simple structure, fast dynamic response, and no need for compensation circuit. Through the DIM pin, the chip can conveniently perform LED switching, analog dimming, and wide range PWM dimming. The simulation results show that when the input voltage changes from 8V to 30V, the maximum deviation of the chip output current does not exceed 5.5%. In addition, when the chip drives 7 LEDs, the efficiency can be as high as 97%.