How to choose the best LED driver architecture

　　Today's LED lighting has a variety of applications, from simple incandescent or cold cathode fluorescent lamp (CCFL) replacements to new architectural, industrial, medical and other applications. In order to optimize the matching of lamps and light in the application, different LED lighting applications usually have corresponding performance standard requirements.

　　To drive LEDs, engineers can choose from a wide variety of driver architectures, but each architecture has its own advantages and disadvantages, and its adaptability to specific applications varies. There are many factors to consider when choosing a driver architecture, among which cost occupies the first position, followed by isolation, dimming, flicker, color temperature, power factor, reliability, thermal management and other issues.

　　There are several basic LED driver architectures: secondary side control, primary side control, isolated/non-isolated. In addition, power factor control (PFC) is also a major performance consideration in many applications, and its solutions consist of two-stage or single-stage drivers with PFC function, or single-stage drivers without PFC function (mainly used in applications with power less than 5W). Therefore, the entire driver subsystem is the result of a series of trade-offs, with the aim of reducing the bill of materials (BOM) cost and achieving the highest efficiency while providing dimming functions and creating a temperature-controlled and fault-protected product.

　　Basic drive architecture

　　To achieve optimal isolation and control, the secondary-side control architecture monitors the output voltage/current and provides feedback signals to the primary-side driver through an optically isolated path (Figure 1). This feedback signal enables the secondary-side controller to provide better current and voltage control accuracy. Simpler primary-side control schemes eliminate the secondary-side controller and the optically isolated signal path, thereby reducing system cost and size while improving system performance. In this scheme, the primary-side driver determines the output current and voltage through primary-side waveform analysis (Figure 1). Depending on the quality of the analysis, primary-side control can match or even exceed secondary-side regulation and performance, making it a common solution for today's isolated LED drivers.

　　Figure 1: Two common LED driver schemes use secondary-side control (top) and primary-side control (bottom). Secondary-side control provides better current and voltage control accuracy, but primary-side control reduces component count and system size while improving performance.

　　The basic primary-side control circuit achieves isolation through the output stage transformer. However, to reduce component cost, non-isolated solutions use an inductor instead of a transformer and can use a buck controller instead of a primary-side driver flyback circuit (Figure 2). In non-isolated solutions, the control mechanism is simplified, but the circuit requires more complex physical isolation to prevent short circuits between the input and output. Currently, most LED driver designs use an isolated architecture. In the next year or two, advances in circuit design will provide solutions to further reduce costs.

　　Figure 2: The primary-side driver can be designed in an isolated configuration by using a transformer in the output stage, or in a nonisolated configuration by replacing the output transformer with an inductor and selecting a buck controller instead of the flyback circuit.

　　Power Factor Basics

　　When the input voltage and current are in phase and the input voltage and current waveforms are consistent, the power factor is ideally "1". When the phase difference between the input voltage and input current waveforms increases, the power factor will decrease and the system efficiency will also decrease. However, the boost converter has a built-in current waveform control function that can track the input voltage waveform to maintain a power factor close to 1.

　　To improve the power factor, a two-stage power factor correction (PFC) boost circuit can be added before the primary-side driver circuit and control circuit (Figure 3). The PFC circuit also eliminates the flicker problem caused by 2 times the line frequency. In the example, a flyback converter is used in the output stage to provide isolation for the iW3616 driver circuit. The primary-side sensing technology used in this driver chip achieves excellent line voltage and LED load current regulation without the use of secondary-side feedback circuits, while eliminating the opto-isolator feedback loop. In addition, the iW3616's real-time cycle waveform analysis technology also improves the setting response speed of the dimmer. The digital control loop can also maintain the overall operating conditions stable without the need for loop compensation devices.

　　Figure 3: By adding a two-stage power factor correction boost circuit to the iW3616 digital power controller, the driver circuit can achieve flicker-free dimming and very high power factor (>0.95).

　　PFC can also be implemented in a single-stage primary-side drive and control circuit. In such a system, the driver controls the input current waveform by modulating the input impedance, thereby regulating the power factor.

　　The two-stage PFC architecture achieves near-perfect PFC while effectively eliminating output ripple, thus significantly improving the flicker problem in LED lamps. However, the two-stage boost circuit requires more components, so the implementation cost is also higher. In contrast, although the single-stage driver with PFC function improves the power factor by modulating the input impedance, the output ripple (flicker) will also increase as the power factor increases. To compensate, the flicker must be reduced by increasing the external capacitor value. In the case where PFC regulation is not required, a simple single-stage primary-side driver can use a traditional flyback converter architecture to reduce costs.

　　Many applications also require the driver circuit to be able to interface with dimmers, but since there are already a variety of dimming technologies on the market - TRIAC-type leading and trailing edge dimmers, complex electronic dimmers, and low-voltage (0V~10V) linear control or pulse width modulation brightness control dimmers (mainly used in commercial systems) - engineers must solve many problems. New digital solutions can analyze the type of dimmer and then use optimized algorithms to control dimming. Such solutions can also eliminate flicker caused by short pulse signal interference. In contrast, traditional TRIAC-type dimmers are susceptible to false triggering and may produce unbalanced half-cycle outputs.

　　All TRIAC type dimmers have a minimum holding current requirement to keep the TRIAC on, but not all LED driver circuits have dimming capabilities. For those LED driver circuits that have dimming capabilities, the driver must load the dimmer to keep the TRIAC on continuously. Although higher loads improve dimmer compatibility, their high load current will reduce circuit efficiency. To improve efficiency again, the driver's load resistor can be replaced with a BJT or MOSFET, allowing the driver to automatically calibrate the bleeder current to ensure accurate current control in the safe operating area and reduce costs by utilizing the existing BJT or MOSFET of the boost/PFC circuit.

　　plan selection

　　Faced with so many choices, engineers usually need to carefully compare and sort out the best combination of features that the LED driver solution should have. The basic decision factors may first be dimming or non-dimming and power consumption requirements. After that, other requirements may include: PFC or no PFC (depending on the application), size requirements (whether the solution fits a specific space or printed circuit board area), reliability/operating life, and tolerable flicker (the lower the better).

　　To reduce BOM cost, consider reducing component count, using primary-side control, minimizing EMI component count, and then looking deeper into component cost - using BJTs instead of MOSFETs for drivers, reducing heat sink size and material. To some extent, these choices are also related to system reliability and operating life - the lower the component operating temperature, the longer the system operating life, especially for components such as electrolytic capacitors and LEDs themselves. Unfortunately, engineers rarely know the specific usage of LED lights, so good temperature control design is particularly important. If the heat sink is not properly sized, closed equipment with poor ventilation may lead to heat accumulation and premature failure.

　　In addition, circuit protection functions can help prevent major fusing failures such as thermal runaway and short circuits - does the driver have built-in temperature detection functions, or can a temperature sensor be added? With the help of temperature sensors, the driver circuit can implement a complex thermal closed loop - when the temperature is higher than the maximum indicator, the LED current is reduced, thereby reducing power consumption and temperature. In extreme cases, the driver can also shut down to protect itself. Many driver circuits also provide additional fault protection functions, such as LED short-circuit and open-circuit detection, overvoltage protection, soft start, and current detection resistor short-circuit protection.

　　All choices can be summarized into a few design rules of thumb related to power factor and flicker issues: If power factor and flicker are not important, use a single-stage primary-side driver without PFC function; if power factor is important, use a single-stage driver with PFC function; but if power factor and flicker are both important, the best choice is a two-stage driver with PFC function.

　　Engineers have many basic architectures to choose from, and by comparing their advantages and disadvantages, they can select the best drive solution for the target application.