LED lighting driver design based on buck converter

Although the peak current mode controlled discontinuous buck-boost converter is a good choice for LED drivers when the output voltage may be higher or lower than the input voltage. However, when using this buck-boost converter to design a driver, changes in the LED voltage will change the LED current, and an open LED will cause excessive voltage at the output, thus damaging the converter. This article will discuss this converter design for LEDs in detail and provide a variety of methods to overcome its inherent shortcomings.

Light-emitting diodes (LEDs) have been used for many years, and with the latest technological advances, they are becoming a strong competitor in the lighting market. New high-brightness LEDs have a long life (about 100,000 hours) and high efficiency (about 30 lumens/watt). Over the past thirty years, the light output brightness of LEDs has doubled every 18 to 24 months, and this growth momentum will continue. This trend is called Haitz's Law, which is equivalent to Moore's Law for LEDs.

Electrically speaking, LEDs are similar to diodes in that they also conduct electricity in one direction (although their reverse blocking capability is not very good and high reverse voltages can easily damage LEDs), and have low dynamic impedance VI characteristics similar to conventional diodes. In addition, LEDs generally have a rated current for safe conduction (high-brightness LEDs are generally rated for 350mA or 700mA). When passing the rated current, the difference in the forward voltage drop of the LED may be relatively large, and the voltage drop of a 350mA white light LED is usually between 3 and 4V.

Driving LEDs requires controlled DC current. To extend the life of LEDs, the ripple in the LED current must be low, because high ripple current will cause the LED to generate large resistive power consumption, reducing the LED life. LED drive circuits need to be more efficient, because the overall efficiency depends not only on the LED itself, but also on the drive circuit. Switching converters operating in current control mode are ideal drive solutions that meet the high power and high efficiency requirements of LED applications.

Driving multiple LEDs also requires careful consideration. Figure 1 shows a series-parallel connection circuit for LEDs. Figure 1(a) shows a parallel connection circuit for LEDs. Figure 1(h) shows a series connection circuit for LEDs. Since the dynamic impedance and forward voltage drop of each LED are different, it is impossible to ensure that the current flowing through the LEDs is the same without an external current-sharing circuit (such as a current mirror). In addition, since a failure of one LED will disconnect the LED string, all LED currents will be distributed among the remaining LED strings, which will increase the current in the LED string and may damage the LED. Therefore, for the above two reasons, parallel LED circuits such as Figure 1(a) are generally not used in designs.

Therefore, it is better to connect the LEDs in series. However, the disadvantage of this method is that if one LED fails, the entire LED string will stop working. A simple way to keep the remaining LED strings working is to connect a Zener diode (whose rated voltage is greater than the maximum voltage of the LED) in parallel with each (or each group of) LEDs, as shown in Figure 1(b). In this way, if any LED fails, its current will flow to the corresponding Zener diode, and the rest of the LED string can still work normally.

Basic single-stage switching converters can be divided into three categories: buck converters, boost converters, and buck-boost converters. Buck converters (Figure 2(a)) are ideal when the voltage across the LED string is lower than the input voltage; boost converters (Figure 2(b)) are more appropriate when the input voltage is always lower than the string output voltage; and buck-boost converters (Figure 2(c)) are more appropriate when the output voltage may be higher or lower than the input voltage (caused by output or input variations). The disadvantage of a boost converter is that any transients in the input voltage (which can increase the input voltage and exceed the output voltage) will cause a large current to flow through the LED (due to the low dynamic impedance of the load), damaging the LED. A buck-boost converter can also be used instead of a boost converter, because transients in the input voltage will not affect the LED current.

Working Principle of Buck-Boost Converter

Buck-boost converters are a good choice for LED drivers in low-voltage applications. The reasons include that they can drive LED strings with voltages above and below the input voltage (boost and buck), high efficiency (easily reach over 85%), discontinuous operation mode can suppress input voltage changes (provide excellent line voltage regulation), peak current control mode allows the converter to regulate LED current without complex compensation (simplifies design), it is easy to achieve linear and PWM LED brightness adjustment, switching transistor failure will not damage the LED, etc. Figure 2 shows the connection circuit of buck, boost and buck-boost converters with LED strings.

However, this method still has disadvantages: one is the problem of peak current control, because the buck-boost converter using discontinuous current mode is a constant power converter. Therefore, any change in the LED string voltage will cause a corresponding change in the LED current; another problem is that the LED open circuit state will generate a high voltage in the circuit that will damage the converter; in addition, additional circuits are required to convert the constant power converter into a constant current converter, and the converter needs to be protected under no-load conditions.

Figure 3 shows a specific buck-boost converter application circuit. The controller has a built-in oscillator for setting the switching frequency. At the beginning of the switching cycle, Q1 is turned on. As the input voltage VIN is applied to the inductor, the inductor current (iL (t)) starts to rise from zero (initial steady state). When the inductor current rises to the preset current value (ipk), Q1 is turned off. The switch on time (ton) is determined by the following formula:

ton=ipkL/VIN

At this point, the total energy (J) stored in the inductor is:

J=Li2pk/2

Thus, although the switch is turned off at this time, the current flowing through the inductor is not interrupted. This causes the diode D1 to conduct and generate an output voltage (-Vo) across the inductor. This negative voltage causes the inductor current to drop rapidly. After a certain time tOFF, the inductor current tends to zero. This time can be calculated by the following formula:

tOFF=ipkL/VO

To make the converter operate in discontinuous conduction mode, the sum of the switch on-time and the inductor current fall time must be less than or equal to the switching period TS to ensure that the inductor current can start from zero in the next switching cycle.

In fact, (tON+tOFF) is at its maximum value at minimum input voltage and maximum output voltage. Therefore, ensuring that the converter operates in discontinuous conduction mode at these voltages ensures that the following condition is met in all cases: tON+tOFF≤Ts

The power (Pin) obtained by the converter from the input is the product of the energy in the inductor and the switching frequency f:

Pin=fsLi2pk/2

Assuming the LED string voltage (VO) is constant and the efficiency is 100%, the LED current (iLED) is:

iLED=PIN/VLED=Li2pkfs/2V

In peak current control mode, ipk is usually a fixed value. Therefore, the LED current is completely independent (theoretically) of the input voltage. With a fixed ipk, a rise (fall) in the input voltage causes an inversely proportional decrease (increase) in the transistor's on-time, which provides very good line voltage regulation. In practice, the delay between the control IC detecting the current peak and the GATE pin actually turning off causes input power variations. Shorter on-times will have more errors due to the delay time, as the delay time will account for a significant portion of the on-time.

In reality, the LED current is inversely proportional to the voltage across the LED string. A circuit with a nominal output of 20 V and 350 mA will produce 700 mA at a 10 V output voltage, which is obviously not the desired result. However, by making the switching frequency proportional to the output voltage, the above formula provides a way to convert a constant power converter into a constant voltage converter.

Assume fs=KVO, where K is a constant, then:

iLED=kLi2pk／2

This way, iLED will be independent of input and output voltage.

Another disadvantage of the flyback converter is that it is susceptible to the output open circuit condition. When the LED is open circuit, the energy stored in the inductor will be transferred to the output capacitor at the end of each switch on time. In this way, the load without capacitor discharge will cause the voltage across the capacitor to gradually rise, eventually exceeding the nominal value of the device and damaging the power stage. Therefore, additional circuitry can be added to provide output voltage feedback and overvoltage protection.

Output voltage feedback

Figure 4 shows an additional circuit that can be used to implement overvoltage protection and LED open circuit protection. In fact, many peak current mode controller ICs have a dedicated RT pin. The resistor connected to this pin can be used to set the internal current, which is used to charge the oscillator capacitor (which can be internal or external). The ramp voltage on the oscillator capacitor controls the switching frequency, so that the switching frequency is proportional to the output current of the RT pin. The smaller (larger) the resistance, the larger (smaller) the current, and the higher (lower) the switching frequency. Based on this principle, the output voltage feedback can be used to adjust the switching frequency.

In the circuit shown in Figure 4, resistors R3 and R4 form a voltage divider. The voltage across R5 is the voltage across R4 minus the voltage drop (Vbe) between the base and emitter of transistor Q2. Therefore, the current flowing through R5 (IR5) is:

This current is obtained from pin RT of the control IC using a matched transistor pair.

Resistor R2 in Figure 4 is used to start the converter. In the start-up state, the output voltage is zero, so IR5 is also zero. Since there is no current from the controller RT pin, the converter cannot start. Increasing resistor R2 can obtain a small amount of current in the start-up state and make R2 sized to meet:

IR5>>V(RT)／R2

Where V(RT) is the voltage on the controller RT pin. Meeting this condition ensures the start-up of the converter and minimizes the error caused by R2. If R3=R4 is selected, then:

IR5>>VO/2R5

It is assumed here that the output voltage is much larger than the base-emitter voltage drop of Q2.

Thus, according to the above formulas, the output LED current can be obtained as:

iLED=KICLi2pk／(2×2R5)

In this way, the LED current is no longer determined by the input or output voltage. Overvoltage protection is added by using resistor R6, transistor Q3 and Zener diode D2. In the LED open circuit state, when the switch is on, the inductor stores energy, and when the switch is off, the energy is transferred to the output capacitor. Because there is not enough load for the capacitor to discharge, the output voltage will gradually increase in each cycle. When the voltage rises to exceed the Zener diode's forward voltage, the Zener diode branch circuit formed by D2 and R6 begins to conduct. This also provides a path for current to flow through the base of Q3, turning Q3 on. At this time, resistor R4 is actually shorted. Therefore, the PN junction of the base emitter of Q2 will turn off, resulting in zero current in R5. This will stop the controller's internal oscillation until the output voltage drops below the Zener diode voltage, and the above process continues. This burst mode can minimize the average power in the LED open circuit state. This overvoltage protection method will force the control IC into a low-frequency, low-power operation mode.

The current in the Zener diode resistor branch circuit must be able to produce a voltage across R6 large enough to bias the PN junction between the transistor base and emitter.

Conclusion

In switching LED drivers with output current feedback, feedback compensation is generally required to stabilize the converter and regulate the current to the desired current value. The transient response performance of these feedback schemes is limited and cannot meet the fast on/off transient response required for PWM brightness adjustment of LEDs. However, the converter described in this article does not require any feedback compensation. The only feedback information used by this control scheme is the peak current flowing through the MOSFET through the sense resistor. Because the converter stores the required energy in each cycle, it can respond instantly to transients. Therefore, it can easily work with PWM brightness adjustment schemes.

The buck-boost converter is an effective solution for low DC voltage input LED drivers. It can drive LED strings regardless of whether the output voltage is higher or lower than the input voltage. In addition, small and inexpensive additional circuitry can be added to the converter to overcome load regulation and no-load conditions. The converter is easy to implement and has no