LCD Display Inverter

Display Inverter / VGA Board / LCD Controller

Gallium Nitride: Complementary Efficient Solutions Covering Mid-Power Applications

[Introduction]During the 2022 American International Power Electronics Application Exhibition (APEC), Power Integrations (PI) held a new product media communication meeting. Senior technical training manager Yan Jinguang introduced the high-efficiency power supply solution for medium-power applications using combination chips. He said that the two new products adopted a two-stage architecture, the first stage is PFC (power factor correction); the second stage is DC-DC conversion. The second-stage conversion adopts LLC topology, that is, a resonant power supply.

The new product uses a HiperPFS™-5 PFC IC with an integrated 750V PowiGaN™ GaN switch as the first stage; the second stage is HiperLCS™, which is very efficient, with a total output power of 270W in resonant soft-switching operation, covering medium Power application range.

So, what are the advantages of this pair of new products, and how can they be used? Let’s look down.

Why use GaN switches?

In recent years, gallium nitride switches have become very popular. PI’s InnoSwitch3 integrates gallium nitride switches. The new products released this time also use gallium nitride switches. Yan Jinguang said that gallium nitride switches have low on-resistance and are more efficient than silicon devices, especially at low voltage input and high current. Under the same power, the lower the input voltage, the greater the current, and the higher the proportion of conduction loss. Traditional silicon devices have higher on-resistance than gallium nitride devices, so there are more losses.

At low voltages, using PI’s PowiGaN gallium nitride switches can reduce heat generation by 40 percent. Taking the 65W adapter as an example, comparing the silicon switch and the PowiGaN gallium nitride switch, the heat is significantly reduced at low voltage, and at high voltage, PowiGaN will also reduce the heat generated by the original silicon switch by 25%, and the efficiency is increased from 92% to 94%. %about. For the case that the power supply is generally small, if the heat can be greatly reduced, the temperature rise of the power supply will be improved significantly.

PowiGaN can be used to make the power supply the size of a credit card. In addition to the PowiGaN gallium nitride switch used in InnoSwitch4, ClampZero can also be used to realize the design of an active clamp flyback power supply. MinE-CAP also integrates a gallium nitride switch, which is mainly used to reduce the size of the input large electrolytic capacitor. At present, gallium nitride power devices are used in all three PI product series to realize all-gallium nitride AC-DC flyback power supply solutions.

What challenges does increasing power bring to high efficiency?

Yan Jinguang said that with the introduction of the USB PD3.1 specification, the power of the adapter power supply is increasing. When the input power exceeds 75W, the power factor requirement must be met. Generally, a PFC correction circuit is added in the front stage of the power supply, so that the current phase of the AC input terminal completely tracks the voltage.

Usually, people want the load connected across the AC grid to be purely resistive, which is characterized by the fact that the phase of the input voltage and input current is always the same, but if the load is a reactive load such as capacitive or inductive, the voltage and current will have phase difference. Voltage multiplied by current equals power. If the voltage and current phase are different, the product will become smaller, and the actual active power used by the load will be very low. Excessive power will be lost on the transmission line at the power supply end. Therefore, when the power is relatively large, a power factor correction function is required, because if the power factor is relatively low, the power seen on the grid is 100W, but only 80W can actually be used. At this time, the power factor is 0.8, so if there is a 20W line loss, you will also pay the electricity bill.

In the past, when the single-stage power supply was designed without a heat sink, the AC input was directly converted from the power supply to DC, and the actual efficiency reached 95% to realize the design without heat sink to ensure the high power density and small size of the power supply. But now that the output power has increased, so will the input power. In the case of power factor correction, the power supply is a two-stage structure, with a PFC pre-stage in the front and a DC-DC conversion stage in the back. The heatsink-free design can only be achieved if the efficiency is greater than 94%, the number of components will be more, and the two-stage power supply heat dissipation will be more difficult to handle.

The challenge is: as the power increases, power factor correction is required, and the original single-stage solution must be implemented with two stages; and the two-stage efficiency cannot be too low, otherwise the overall efficiency will drop, and the power supply will heat up seriously. Therefore, the pre-stage PFC efficiency of the two-stage power supply design must be high, and the DC-DC conversion efficiency must be as high as possible to ensure that the overall heat dissipation meets the temperature rise requirements.

How to achieve high efficiency for applications above 200W?

Yan Jinguang said that in order to meet the challenge of two-level power supply efficiency, PI has launched two products, one is HiperPFS-5, which integrates a gallium nitride power switch, and the efficiency of each level is guaranteed to be greater than 98%. Its front stage is power factor correction, which can ensure that the input current and the input voltage are in phase. The output of the front-stage power conversion is usually a high-voltage DC, which is about 400V, and is converted into the required output voltage through the secondary conversion of the latter-stage. The second stage uses PI’s second chipset product. One of the chips integrates the upper and lower tubes in the LLC topology. It uses a 600V withstand voltage FREDFET MOS tube instead of a gallium nitride switch, because LLC The application frequency is not too high, coupled with the high voltage of the busbar, the current is relatively small, and it does not reflect the advantages of low on-resistance of gallium nitride. The good reverse recovery characteristics of the body diode of the FREFET are also conducive to optimizing the performance of the LLC.

When LLC works, its power switch tube half-bridge works in resonance mode, and both MOS tubes can switch in zero voltage mode (ZVS), which can reduce the turn-on loss to zero, that is, the switch is turned on only when the voltage across the device is zero. tube, the current starts to rise. The reduction of the turn-on loss in this way of working allows the two power switches to be packaged in one IC, and only the PCB is used for heat dissipation.

The second IC is the secondary controller, which is connected between the primary and secondary of the secondary power supply, and transmits the feedback signal of the secondary to the power conversion device through FluxLink™. The two chips must work together. Only in this way can the switching sequence of the primary half-bridge switch tube and the secondary output synchronous rectifier tube be optimized.

He explained that the overall power architecture has three chips, which are actually two sets of ICs, one for functional factor correction and the other for the LLC chipset. Since many power-consuming functions have been integrated into the IC, the total no-load power consumption can be less than 40mW, which is very helpful for some adapter applications, which can improve power density with a small number of components and reduce unnecessary power consumption. power consumption to minimize standby power consumption.

Active PFC with very few components

The front-stage active PFC in the two-stage architecture works in a variable frequency DCM (discontinuous conduction mode) mode; the power factor correction conversion can be designed to work in different working modes, which can be continuous mode, critical mode or discontinuous mode model. For smaller power applications, such as within 250W, CRM (critical mode) or DCM is usually used, because the inductance is relatively small and the output diode has no reverse recovery problem. Although CCM (Continuous Conduction Mode) can achieve higher efficiency, the reverse recovery problem of the boost diode must be considered, not only the cost of the diode will increase, but the high frequency oscillation that occurs during reverse recovery will also have an impact on EMI , At the same time, the cost and volume of the boost Inductor in the continuous mode will increase due to the relatively large inductance.

The HiperPFS-5 released this time works in discontinuous mode, and works in critical mode when the input voltage is the lowest. This design not only considers the cost of the overall solution, but also considers the space occupied by the power supply itself, which is conducive to realizing a design with high power density and small volume. The chip’s unique way of controlling the on-time and off-time of the power tube separately can ensure that the operating frequency at the voltage peak is the highest when the 90VAC low-voltage input is input, which can greatly reduce the inductance of the required boost inductor, thereby reducing the inductance volume. The traditional critical work control method often has the lowest switching frequency at the peak of the AC input voltage, so a boost inductor with a higher inductance has to be used.

If you want to track the input voltage with the input current, you must know the input voltage waveform. HiperPFS-5 adopts a digital voltage sampling method, that is, there is an ADC (analog-to-digital conversion) inside the IC, which can filter out the slight distortion in the waveform. For example, if the front-end input voltage comes from a generator or UPS power supply, its input is often not an ideal sinusoidal waveform.

X-capacitor discharge is a safety requirement for preventing electric shock. In DCM mode, due to the large peak current, there may be a problem of relatively high differential mode EMI components. The effective way to suppress differential mode EMI is to use X capacitors, and excessively large capacitors A corresponding discharge resistor is required to ensure that the X capacitor can be discharged for a short time after the AC power failure. If a discharge resistor with a smaller resistance value is always connected across the AC input terminals, the standby or no-load power consumption performance of the power supply will be deteriorated when it is working. PI has also integrated the X-capacitor discharge control function of the previous CapZero IC into the HiperPFS-5. The internal switch will be turned on after the AC is powered off, and the discharge resistor is connected to both ends of the X capacitor to discharge. During the normal operation of the power supply, the discharge resistor itself has no power consumption.

Another feature of the new product is the use of self-power supply, which can save external power supply circuits. The IC can be powered using the following DC-DC or another winding. In PFC circuits, most solutions on the market use charge pumps with three or four components. PI’s solution can realize self-powering of the drain, and the controller can be powered internally, which saves some components and makes the solution more concise.

HiperPFS-5 integrates a 750V withstand voltage PowiGaN gallium nitride switch, which can withstand lightning strikes and prevent the DC-DC conversion behind the direct impact of the PFC. At 230V full load, the efficiency can reach 98.3%, which helps to reduce temperature rise and achieve overall system efficiency.

Another technology is Power Factor Enhancement (PFE), which guarantees a power factor greater than 0.96 at high voltage input and 20% load. This function is that in the case of high voltage input and light load, the chip will adjust the waveform of the input current by itself to compensate for the waveform distortion and power factor deterioration caused by excessive input X capacitance. The optimized quasi-resonant (QR) mode can improve turn-on losses, especially at high input voltages.

In a word, the unique control engine of HiperPFS-5 can achieve more than 98% efficiency, variable frequency can ensure light load efficiency, quasi-resonant mode can reduce the turn-on loss when high voltage input; discontinuous mode can use smaller inductance boost Inductance, the volume will be reduced accordingly; the frequency sliding technology can ensure high efficiency at heavy and light loads, and the frequency change can also reduce EMI; the output power of the GaN switch can reach 240W, covering most medium power applications; ultra-thin InSOP™ -28F package with bare horizontal plate exposed below, which helps to dissipate heat, and the chip height is only 1.9mm.

How to ensure the best conversion efficiency?

The newly introduced HiperLCS-2 chipset adopts LLC resonance control and synchronous rectification scheme. Why does such a combination ensure optimal conversion efficiency?

Yan Jinguang said that the first generation of HiperLCS products did not have synchronous rectification, and now the control chip is bridged between the primary and secondary of the LLC conversion. Utilize the high-bandwidth transmission characteristics of FluxLink to achieve fast and accurate switching timing control. Maximize the advantages of synchronous rectification. Synchronous rectification is widely used in flyback power supplies, mainly to improve rectification efficiency. It has lower conduction losses than diodes and has higher overall power supply efficiency. Since the power range of LLC topology applications is larger than that of flyback, the use of synchronous rectification is also the key to further improving efficiency when the output current is high.

The two 600V withstand voltage FREDFETs integrated on the primary side have body diodes with good recovery characteristics, which can achieve shorter dead time of the upper and lower tubes. The reverse recovery is better, and many constraints in the design can be relaxed, making it easier to design. Traditional MOS tubes also have parasitic diodes, but their reverse recovery characteristics are very poor. The reverse recovery characteristics are relatively poor, in addition to causing losses, it will also limit the timing design and affect the overall power efficiency. Therefore, only if this indicator is very good, it is not necessary to strictly consider the timing design of the switch to optimize the efficiency.

The 600V withstand voltage of FREDFET can meet the bus voltage requirements. Usually the output of the front PFC is stabilized at 400V, which can provide about 20% margin and make the power supply more reliable. According to different application requirements, the center frequency of LLC resonant conversion can be selected by external setting. LLC itself is a working mode whose working frequency changes continuously with the change of load and input voltage, so its working frequency range is often wider. Of course, the higher the frequency, the smaller the volume of passive components, but the switching loss is increased if the frequency is too high, and the consumption of power devices will also increase. The specific setting of the center switching frequency depends on the specific application requirements. The HiperLCS-2 can set the center frequency at 90kHz, 120kHz, 180kHz and 240kHz through external components. No matter what the center frequency is, the half-bridge switches work in the way of ZVS soft switching.

In addition to supplying power to the LLC, the external bias supply winding can also supply power to the pre-stage PFC. The PFC is self-powered just when it starts up. After the DC-DC conversion of the latter stage works, the auxiliary winding of the transformer transformed by the latter stage supplies power to the PFC controller of the former stage. The upper and lower tubes of the LLC span the two ends of the busbar. The most feared thing is that the two switches are turned on at the same time, which will short-circuit the busbar and cause an explosion. The built-in protection function of HiperLCS-2 can prevent the driving signal from being turned on at the same time, and can also realize over-current or output short-circuit, over-power protection, and the hard switching of the power switch tube will also be detected by the chip in time, thereby avoiding the increase in loss during hard switching. large, the device is damaged.

Combined use to meet a variety of different application needs

While being driven by synchronous rectification, the HiperLCS2-SR device can transmit the output voltage and current information to the half-bridge through FluxLink to determine the switching sequence, and the secondary side detection can also improve the accuracy of the output voltage and current.

An important characteristic of LLC operation is to operate in hiccup mode at full no-load, which can cause increased output ripple and even voltage offset. The hiccup mode is specially optimized in the control engine of PI. The three hiccup modes can ensure that the output voltage can still maintain the output voltage regulation range even during the hiccup mode operation. While the load has a 0-100% jump, the stability of the output voltage can still be maintained.

According to Yan Jinguang, the PI integrated chipset has two IC devices: HiperLCS2-HB (half bridge device) and HiperLCS2-SR (isolation control device). The FluxLink used in it has a very fast feedback speed, which is beneficial to meet the application requirements of dynamic load changes. FluxLink can also feed back secondary fault conditions to half-bridge power devices, ensuring the safety and reliability of power devices during faults.

HiperLCS-2 can provide bias power for the PFC pre-stage, and the efficiency can reach 98.1%. The internally integrated 600V withstand voltage FREDFET can meet the PC application titanium (80 PLUS Titanium) standard, including light load efficiency and light load PFC requirements.

For medium power applications, the protection circuit must be well thought out or some catastrophic failure will occur. Since LLC is a variable frequency operation, the efficiency of 24V/8.5A output can reach about 98% at 205W 90K center frequency. At full load, the loss of less than 5W can be dissipated by the PCB; when the load is relatively light and when the load is heavy, the efficient efficiency curve is relatively constant.

The chipset includes two sets of devices: HiperLCS2-HB (power device) and HiperLCS2-SR (safety isolation device). There is only one model of safety isolation device, and different versions can be selected according to different frequencies. The model is LSR2000C. According to different output power ranges, there are three choices of power devices: LCS7260C, LCS7262C and LCS7265C, which integrate 600V FREDFET inside and have 80W to 220W continuous output power. Depending on the cooling conditions, the maximum power is 270W, and the peak power can reach 375W.

If the HiperPFS-5 is used in combination with the HiperLCS-2 chipset or the InnoSwitch3, it is also suitable for many different applications. In a single-stage traditional flyback power supply design, the power can only reach 65W, and the power increase needs to meet the power factor requirements. Using HiperPFS-5 and HiperLCS-2, compared to the single-stage solution for applications with lower power, although the efficiency is slightly lower, when the power reaches more than 200W, the single-stage solution can no longer maintain high efficiency. Therefore, for different applications, if the output power is relatively small, HiperPFS-5 can be used to add a flyback power supply InnoSwitch3; if the power is relatively high, a HiperLCS-2 can be added later. The combined scheme can still reach more than 95% efficiency.

The above combination has a wide range of applications, including most medium-power home appliance applications, such as televisions, USB interface monitors, game consoles, electric bicycles or printers, projectors, and PC mains. The output power of 240W and 270W has also changed the embarrassing situation that the application of 50 and 60W PD was limited to mobile electrical equipment in the past. Does it give us a feeling of vastness?

Source: PSD Power System Design

Author: Liu Hong