Comprehensive consideration of circuit configuration application

There are many electronic products that require low-profile and small AC-DC power supply designs, such as flat-panel displays, rack-mounted computer equipment, chassis-mounted equipment for telecommunications and aviation, etc. However, even for a fairly experienced power supply designer, it is not an easy task to maximize the efficiency of the AC-DC power supply in a flat and small device; not to mention that such equipment must provide hundreds of watts of power to the load at a given time, which brings greater design challenges.

For example, a typical 12 volt (V), 300 watt (W) power supply used in a 1U rack-mount application has size constraints, with a maximum height of no more than 1.75 inches (44.45 mm) and must include one or more fans for forced air cooling. However, for systems with height constraints less than 1U, forced air cooling may not be feasible, which means that costly and low-profile heat sinks with large surface areas must be used to achieve thermal management. Therefore, the design of AC-DC power supplies for maximum efficiency is very important because it has a direct impact on reducing the size and cost of the heat sink and improving the overall reliability of the design.

BCM/CCM PFC helps AC-DC power supply design

In most power level operation situations, AC-DC power supplies require some type of active power factor correction (PFC). However, whether PFC is needed depends on several considerations, including power level, end application, equipment type and geographical location; in addition, it is usually determined by compliance with specifications such as EN6100-3-2 or IEEE 519.

For AC-DC power supply design, a non-isolated and offline boost pre-regulator is generally used as PFC, where the DC output voltage is used as the input of the downstream isolated DC-DC converter. Since these two converters are connected in series, the overall system efficiency ηSYS will be the product of the efficiency of each converter:

¨¨¨¨Equation 1

As can be seen from Equation 1, careful and comprehensive considerations must be made when selecting the best power topology and control technology for both converters. There are two PFC control technologies. The first system solution with many high-efficiency features is a combination of interleaved dual boundary conduction mode (BCM) PFC, and the other is continuous conduction mode (CCM) PFC.

For BCM PFC mode, an asymmetric half-bridge (AHB) isolated DC-DC converter must be used, which must use a current doubler rectifier secondary with a self-driven synchronous rectifier (SR). Especially for PFC in the 300W to 1KW range, BCM PFC should be considered because it is more efficient than CCM PFC control technology at similar power levels. It is based on a variable frequency control algorithm in which the two PFC boost power stages are synchronized 180 degrees out of phase with each other.

In addition, since BCM PFC has effective inductor ripple current cancellation, the high peak current commonly seen in the electromagnetic interference (EMI) filter and PFC output capacitor is reduced, and the output PFC large capacitor benefits from the ripple current cancellation, thereby reducing the AC RMS current flowing through the equivalent series resistance (ESR). Moreover, since the boost metal oxide semiconductor field effect transistor (MOSFET) is turned off under zero voltage switching (ZVS) dependent on the AC line and turned on under zero current switching (ZCS), the efficiency can be further improved, and for a 350-watt interleaved BCM PFC design, the MOSFET heat sink can be removed, as shown in Figure 1.

Figure 1. A small, universal 12 volt, 300 watt AC-DC power supply.

On the other hand, the boost MOSFET used in CCM PFC designs is susceptible to frequency-dependent switching losses that are proportional to input current and line voltage. By turning off the interleaved BCM boost diodes at zero current, reverse recovery losses are avoided, allowing the use of inexpensive fast-recovery rectifier diodes and, in some cases, eliminating the need for a heat sink.

However, for CCM PFC design, reverse recovery loss is inevitable. To solve this problem, an RC snubber is usually used across the diode (but this will reduce efficiency), or a higher performance silicon carbide diode is used (which will increase the relevant cost).

LLC/AHB topology is widely favored in building isolated DC-DC converters

In the whole AC-DC power supply design, the isolated DC-DC converter design is an important part, and the half-bridge is a good topology choice for this design, because it has two complementary driven primary-side MOSFETs, and the maximum drain-to-source voltage is limited by the applied DC input voltage. Among them, there are two derivatives of the half-bridge topology, namely half-bridge resonant (LLC) and AHB, both of which have been widely adopted, partly because power management control ICs dedicated to these topologies are available.

First, LLC achieves ZVS switching by using variable frequency control techniques to exploit parasitic elements associated with the power stage design. However, since the regulated DC output uses only capacitor filtering, this topology is best suited for applications with low output ripple and higher output voltages. For offline DC-DC applications, the general rule is that LLC is best chosen when the output voltage is greater than 12 V DC.

In addition, for 300W, 12V DC-DC converters, AHB becomes a high-efficiency choice, which adopts a fixed frequency control method. Since the primary current is determined by the primary voltage of the transformer, it can provide the necessary conditions for ZVS of the two primary MOSFETs. At the same time, the premise of using AHB to achieve ZVS capability is similar to LLC, and it also depends on a thorough understanding of circuit parasitic elements, such as transformer leakage inductance, winding capacitance (Winding Capacitance) and junction capacitance of discrete power devices.

Simplify SR work with fixed frequency solution

Compared with the variable frequency control method used in LLC control, the fixed frequency scheme can greatly simplify the work of the secondary-side self-driven SR, so that its gate drive voltage can be easily calculated from the secondary side of the transformer. At this time, adding a low-side MOSFET driver, such as the dual 4 ampere (A) FAN3224 driver shown in Figure 2, can accurately give the level conversion and high peak drive current flowing through the MOSFET Miller flat region, thereby ensuring fast and efficient SR switch conversion.

Figure 2 Schematic diagram of self-driven synchronous rectification (SR) with current doubler rectifier

The current doubler rectifier of Figure 2 can be used in any two-terminal power supply topology and high DC current applications. It has several outstanding characteristics. First, its secondary side is composed of a single winding, which can simplify the structure of the transformer. Second, because the required output inductance is distributed between the two inductors, the power loss caused by the large current flowing into the secondary side will be more effectively distributed. Third, as a function of the duty cycle (D), after the two inductor ripple currents cancel each other, the two inductor currents will have an apparent frequency equivalent to twice the switching frequency, which allows higher frequencies and lower peak currents flowing into the output inductor.

Finally, in symmetrical converters (push-pull, half-bridge, full-bridge), each current-doubler inductor can carry half of the output current, which is not always the case with AHB, and the asymmetric voltage applied to the secondary-side rectifier may also be one of the disadvantages of AHB. When AHB is operated near its limit duty cycle of 0.5, the loaded SR voltage can be almost matched.

However, a more reasonable solution is to design the transformer turns ratio so that the duty cycle is kept within a specific range of 0.25 < duty cycle < 0.35 during rated operation. When the duty cycle is within this range, as shown in Figure 2, the voltage stress between Q1 and Q2 and the voltage across L1 and L2 become unbalanced, resulting in uneven current distribution between L1 and L2. The rated voltage of each SR MOSFET must be taken into account.

With this in mind, the design can be optimized by using unequal inductance values ​​for L1 and L2, and SR MOSFETs with different voltage ratings, and the transformer turns ratio can also be asymmetrical; however, using these techniques requires a deep understanding of the circuit behavior under all operating conditions.

Detailed evaluation of materials/components and efficiency/size can be taken into account

It is worth noting that the specifications shown in Table 1 illustrate the feasibility of the above solution, but an interleaved dual BCM PFC boost pre-regulator must be used to meet this design, followed by an asymmetric half-bridge DC-DC converter with a self-driven SR, as shown in Figure 1.

In fact, the specifications in Table 1 are a simple conclusion of the design requirements of AC-DC power supplies. The main design goals include obtaining the maximum efficiency in the widest range possible and achieving the smallest power supply design and heat sink size. To obtain the maximum efficiency in a wide load range, the material and component selection of each power stage must be carefully considered, especially in the magnetic design. Since the frequency of the interleaved BCM PFC can be as high as hundreds of kHz and vary by as much as 10:1, the boost inductor must be customized.

For example, using the appropriate grade of equivalent multi-stranded wire (Litz Wire) can reduce AC loss, which is the main part of copper loss in the BCM PFC boost inductor. Therefore, gapped ferrite materials suitable for high-frequency operation should be used, such as N87 material from EPCOS to make thin and flat EFD30 ferrite core groups, and the measured PFC efficiency is shown in Figure 3.

Figure 3 Measured efficiency of AC-DC power supply design with interleaved BCM PFC (100% = 330 W)

For a 300-watt flat AHB transformer, one solution is to use a two-horizontal core structure, with the primary winding connected in series and the secondary winding connected in parallel. However, this solution requires the use of two transformers because the cross-sectional area (Ae) of each core is almost half of the 150 square millimeters required to avoid saturation, and it is impossible to design a traditional shaped core with a cross-sectional area of ​​150 square millimeters on a small component less than 20 millimeters high. Therefore, similar to the BCM PFC inductor design, this solution also uses stranded wire and high-frequency ferrite core materials to maintain high efficiency.

The last important design step is to control the leakage inductance in the AHB transformer within the allowable range. For ZVS requirements, certain leakage inductance values ​​are required; for self-driven SR, the timing delay needs to be adjusted. In this design, the effective leakage generated by the transformer is optimized to 7μH, which is 1.5% of the total effective magnetic inductance. The efficiency results of the 300W AHB DC-DC converter are shown in Figure 4.

Figure 4 AHB 390V to 12V/25A DC-DC measured efficiency (100% = 300W)

Reducing conduction loss is the key, BCM/AHB controller helps

The full-load efficiency measured in Figure 4 is dominated by the conduction losses of the converter power stage, so there is little that a controller can help under these conditions. However, to maintain high light-load efficiency, there are several controller technologies that can be considered. For example, the FAN9612, an interleaved dual BCM PFC controller from Fairchild Semiconductor, uses an internally fixed maximum frequency clamp to limit the frequency-dependent output capacitor (Coss) MOSFET switching losses at light loads and near the zero crossing of the AC input voltage.

It is worth noting that during the period when the input voltage (VIN) > half of the output voltage (VOUT) in the AC line voltage part, the valley switching technique can also be used to sense the optimal MOSFET on-time, further reducing the capacitive switching loss of the output capacitor; and when VIN In addition, FAN9612 also introduces an automatic phase management function to further improve light-load efficiency. This function can reduce the dual-channel operation to single-channel operation mode, and phase management helps to improve the efficiency of light-load efficiency. As shown in Figure 3, when 10%<20%, the efficiency curve looks flatter. In addition, the single-channel operation mode can minimize the impact of switching losses on light-load efficiency. As shown in Figure 5, the interleaved PFC has the ability to maintain synchronization during phase management. The left figure records the situation when the load increases from 0 to 19% (64 watts) and the single channel is converted to dual-channel operation mode. The right figure records the situation when the load drops from full load to 12% (42 watts) and the dual channel is converted to single channel operation mode. When <20%, the efficiency curve looks flatter. In addition, the single channel operation mode can minimize the impact of switching losses on light load efficiency. As shown in Figure 5, the interleaved PFC has the ability to maintain synchronization during phase management. The left figure records the situation when the load increases from 0 to 19% (64 watts) and the single channel is converted to dual channel operation mode. The right figure records the situation when the load drops from full load to 12% (42 watts) and the dual channel is converted to single channel operation mode. >

Figure 5 PFC phase management comparison diagram

On the other hand, the implementation of AHB isolated DC-DC converter can be realized by using AHB controller FSFA2100. For example, by introducing FSFA2100 into a single nine-pin power semiconductor system package (SiP), it can integrate pulse width modulation (PWM) control, gate drive function and internal power MOSFET functions. This advanced integration allows designers to further obtain extremely high efficiency up to 420 watts with fewer external components.

It is worth noting that integrating these three key functions into a single package avoids the task of programming the dead time required for ZVS and minimizes the parasitic inductance of the gate drive between the internal driver and the MOSFET. However, most of the power dissipation in the SiP power package comes from the switching of the internal MOSFET, so a flat extruded heat sink is required, especially for a 300-watt design without forced air cooling.

The design links are closely linked to create a high-efficiency AC-DC power supply

In general, for the design case described in this article, the overall AC-DC system includes input EMI filter, bridge rectifier, interleaved BCM PFC and AHB isolated DC-DC converter, and the overall efficiency obtained is shown in Figure 6. When Vin = 120 volts AC (VAC), the peak efficiency of the design is 91%; when Vin = 230 volts AC, it is 92%; when Vin = 120VAC or 230VAC, and POUT> 38% (114 watts), it is greater than 90%.

Figure 6 Overall system efficiency of AC-DC power supply

All of these factors, including magnetics design, power semiconductor selection, printed circuit board (PCB) layout, heat sink selection, and controller characteristics, must work together to successfully achieve a flat and small AC-DC power supply design that achieves high efficiency over a wide load range.