Flyback Power Supply Reference Design Tips and Tricks from PI Engineers

Article 1. Ferrite Magnetic Amplifier in Flyback Power Supply

For a dual output flyback supply that provides real power on both outputs (5V 2A and 12V 3A, both regulated to ± 5%), the voltage reaches 12V and enters a zero load condition where it cannot be regulated within the 5% limit. A linear regulator is a viable solution, but still not ideal due to its high price and reduced efficiency. The solution we recommend is to use a magnetic amplifier on the 12V output, even in a flyback topology.

To reduce the cost, it is recommended to use a ferrite magnetic amplifier. However, the control circuit of the ferrite magnetic amplifier is different from the control circuit of the conventional rectangular hysteresis loop material (high permeability material). The control circuit (D1 and Q1) of the ferrite can absorb the current to maintain the output power supply.

The circuit has been fully tested. The transformer windings are designed for 5V and 13V outputs. The circuit achieves ± 5% regulation of the 12V output while even reaching input power below 1W (300 mW at 5V and zero load at 12V).

Figure 1

Article 2. Using existing arc suppression circuits to provide overcurrent protection

Consider a 5V 2A and 12V 3A flyback power supply. One of the key specifications of this power supply is to provide over-power protection (OPP) on the 5V output when the 12V output reaches no load or very light load. Both outputs have a voltage regulation requirement of ± 5%.

For the usual solution, using a sense resistor will reduce the cross regulation performance, and the fuse is expensive. However, there is now a crowbar circuit for overvoltage protection (OVP). This circuit can meet both OPP and voltage regulation requirements, and this function can be achieved by using a partial crowbar circuit.

As can be seen in Figure 1, R1 and VR1 form an active dummy load at the 12V output, which allows 12V voltage regulation when the 12V output is lightly loaded. When the 5V output is in an overload condition, the voltage on the 5V output will drop. The dummy load draws a large amount of current. The voltage drop across R1 can be used to detect this large amount of current. Q1 turns on and triggers the OPP circuit.

Figure 2

Article 3. Active Parallel Regulators and Dummy Loads

In the field of switching power supplies from line voltage AC to low voltage DC , flyback is currently the most popular topology. One of the main reasons for this is its unique cost-effectiveness, which can provide multiple output voltages by simply adding additional windings to the secondary of the transformer.

Typically, feedback is taken from the output that has the tightest output tolerance requirements. This output then defines the turns per volt for all other secondary windings. Due to leakage inductance effects, the outputs cannot always achieve the desired output voltage cross regulation, especially when a given output may be unloaded or very lightly loaded because other outputs are fully loaded.

A post regulator or dummy load can be used to prevent the output voltage from rising under such conditions. However, the added cost and reduced efficiency of post regulators or dummy loads make them less attractive, especially in recent years as regulatory requirements for no-load and/or standby input power consumption in many consumer applications have become increasingly stringent. The active shunt regulator shown in Figure 1 not only solves the regulation problem, but also minimizes the cost and efficiency impact.

Figure 3: Active shunt regulator for multiple output flyback converters.

The circuit works as follows: When both outputs are in regulation, resistor divider R14 and R13 bias transistor Q5, which in turn keeps Q4 and Q1 off. Under these operating conditions, the current flowing through Q5 acts as a small pre-load on the 5V output.

The standard difference between the 5V output and the 3.3V output is 1.7V. When the load demands additional current from the 3.3V output, and the load current drawn from the 5V output does not increase by an equal amount, its output voltage will increase compared to the voltage at the 3.3V output. Since the voltage difference is approximately more than 100 mV, Q5 will be biased off, turning on Q4 and Q1 and allowing current to flow from the 5V output to the 3.3V output. This current will lower the voltage at the 5V output, thereby reducing the voltage difference between the two outputs.

The amount of current in Q1 is determined by the voltage difference between the two outputs. Therefore, this circuit can keep both outputs in regulation regardless of their load, even in the worst case where the 3.3V output is fully loaded and the 5V output is unloaded. Q5 and Q4 in the design can provide temperature compensation, because the temperature changes in VBE in each transistor cancel each other out. Diodes D8 and D9 are not required devices, but can be used to reduce the power dissipation in Q1, eliminating the need to add a heat sink to the design.

The circuit reacts only to the relative difference between the two voltages and is essentially inactive under full and light load conditions. Because the shunt regulator is connected from the 5V output to the 3.3V output, the active dissipation of the circuit can be reduced by 66% compared to a shunt regulator connected to ground. The result is high efficiency at full load and low power dissipation from light load to no load.

Article 4. High voltage input switching power supply using StackFET.

Industrial equipment that operates from three-phase AC often requires an auxiliary power stage that can provide stable low-voltage DC power to analog and digital circuits . Examples of such applications include industrial drives, UPS systems, and energy meters.

The specifications for such power supplies are much tighter than those required for standard off-the-shelf switches. Not only are the input voltages in these applications higher, but equipment designed for three-phase applications in industrial environments must also tolerate very wide fluctuations—including extended sags, surges, and the occasional loss of one or more phases. Furthermore, the specified input voltage range for such auxiliary power supplies can be as wide as 57 VAC to 580 VAC.

Designing a switching power supply with such a wide range can be a challenge, mainly due to the high cost of high-voltage MOSFETs and the limited dynamic range of traditional PWM control loops. StackFET technology allows the use of less expensive, 600V rated low-voltage MOSFETs and integrated power controllers from Power Integrations, which can design a simple and inexpensive switching power supply that can operate over a wide input voltage range.

Figure 4: Three-phase input 3W switching power supply using StackFET technology.

The circuit works as follows: The input to the circuit Current can come from a three-phase three-wire or four-wire system, or even from a single-phase system. The three-phase rectifier is formed by diodes D1-D8. Resistors R1-R4 provide inrush current limiting. If fusible resistors are used, these can be disconnected safely during a fault, eliminating the need for separate fuses. The pi filter, formed by C5, C6, C7, C8 and L1, filters the rectified DC voltage.

Resistors R13 and R15 are used to balance the voltage between the input filter capacitors .

When the MOSFET in the integrated switch (U1) turns on, the source of Q1 is pulled low, R6, R7, and R8 provide the gate current, and the junction capacitance from VR1 to VR3 turns Q1 on. Zener diode VR4 is used to limit the gate-source voltage applied to Q1. When the MOSFET in U1 turns off, the maximum drain voltage of U1 is clamped by a 450 V clamp network formed by VR1, VR2, and VR3. This limits the drain voltage of U1 to nearly 450 V.

Any additional voltage at the end of the winding connected to Q1 is applied to Q1. This design effectively distributes the total rectified input DC voltage and the flyback voltage between Q1 and U1. Resistor R9 is used to limit high frequency ringing during switching, and the clamp network VR5, D9 and R10 are used to limit the peak voltage on the primary due to the leakage inductance during the flyback interval.

Output rectification is provided by D1. C2 is the output filter. L2 and C3 form a secondary filter to reduce the switching ripple at the output.

When the output voltage exceeds the total voltage drop across the optocoupler diode and VR6, VR6 will turn on. The change in output voltage causes a change in the current through the optocoupler diode in U2, which in turn changes the current through the transistor in U2B. When this current exceeds the FB pin threshold current of U1, the next cycle is inhibited. Output regulation can be achieved by controlling the number of enable and inhibit cycles. Once a switching cycle is initiated, the cycle ends when the current rises to the internal current limit of U1. R11 is used to limit the current through the optocoupler during transient loads and to adjust the gain of the feedback loop. Resistor R12 is used to bias Zener diode VR6.

IC U1 (LNK 304) has built-in features so that the circuit is protected from loss of feedback signal, short circuit at the output, and overload. Since U1 is powered directly from its drain pin, there is no need to add an additional bias winding on the transformer. C4 is used to provide internal power supply decoupling.

Article 5. Designing a Forward Converter Using TopSwitch. - GX

This circuit ensures that the transformer is reset during each cycle, thus greatly simplifying the design of a forward converter using TopSwitch-GX.

Figure 5: Forward converter reset detection scheme.

The detection circuit is used in conjunction with the bias winding of the forward converter to detect the voltage waveform during the shutdown period. When the voltage is high during this period, a signal is applied to the TopSwitch-GX L pin, disconnecting it from the S pin, thereby inhibiting the internal MOSFET from starting another conduction cycle. When the voltage signal on the bias winding begins to decay, it indicates that the transformer has reset, the L pin is connected to the S pin, and the switch has turned on.

Article 6. Choosing a good rectifier diode can simplify EMI filter circuits in AC/DC converters and reduce their cost

This circuit can simplify the EMI filter circuit in the AC/DC converter and reduce its cost.

To make AC/DC power supplies compliant with EMI standards, a large number of EMI filter components such as X capacitors and Y capacitors are required. The standard input circuit of an AC/DC power supply includes a bridge rectifier to rectify the input voltage (usually 50-60 Hz). Since this is a low frequency AC input voltage, standard diodes such as 1N400X series diodes can be used. Another reason is that these diodes are the cheapest.

These filter devices are used to reduce the EMI generated by the power supply in order to comply with published EMI limits. However, since the measurements used to record EMI only begin at 150 kHz and the AC line voltage frequency is only 50 or 60 Hz, the reverse recovery time of the standard diodes used in bridge rectifiers (see Figure 1) is long and usually not directly related to EMI generation.

However, input filtering circuits in the past sometimes included some capacitors in parallel with the bridge rectifier to suppress any high-frequency waveforms caused by the rectification of the low-frequency input voltage.

If fast recovery diodes are used in the bridge rectifier, these capacitors are not needed. When the voltage across these diodes begins to reverse, they recover very quickly (see Figure 2). This reduces the excitation of stray line inductance in the AC input line by reducing the subsequent high-frequency turn-off transition and EMI. Since two diodes can conduct in each half cycle, only two of the four diodes need to be fast recovery types. Also, only one of the two diodes that conduct in each half cycle needs to have fast recovery characteristics.

Figure 6: Typical input stage of an SMPS using a bridge rectifier at the AC input.

Figure 7: Input voltage and current waveforms showing the diode abrupt transition at the end of reverse recovery.

Article 7. Floating constant current source allows ultra-wide range of input voltage

For most Power Integrations products, the data sheet limits the minimum drain voltage to 50 V to ensure proper startup and function. However, if the BYPASS pin is fed with an external supply , the chip can be externally powered and start and operate even at lower input voltages.

Figure 8: Floating constant current source circuit for power controller .

The startup circuit shown in Figure 8 is a floating constant current source that provides a constant current of approximately 600 μA to the BYPASS (BP) pin of TinySwitch-III over the entire input voltage range.

The constant current value is determined by R2 and VR1:

Formula 1

This circuit is derived from a basic single transistor current source. The circuit uses a Zener diode to set a reference voltage for the base terminal of Q2 (NPN), which in turn programs a fixed voltage through resistor R2 to set the constant current value. However, due to the extremely wide input supply range, the bias current of the reference Zener diode varies over a wide range. This results in increased power dissipation and an offset in the programmed constant current.

To overcome this problem, an additional current source (formed by Q1 (PNP) and R1) is required to provide bias current. A constant voltage equal to VBE is forced across R1, which provides bias current compensation for the reference Zener diode over the entire operating range.

Transistor Q2 provides constant current at lower input voltages, while Q1 provides constant current at higher input voltages. Figure 2 shows the simulation results for current flowing through Q1 and Q2. When the input voltage reaches approximately 50 VDC, Q2 will provide constant current. When the input voltage reaches 50 VDC and above, the current through Q2 will decrease, while the current through Q1 will increase linearly. When the input voltage reaches the maximum value of 375 VDC, the constant current is mainly provided by Q1.

R3 is used to limit the input current of the entire circuit when the input voltage is maximum.

Figure 9: Transistor current vs. total BYPASS (BP) pin current when input voltage is exceeded.

The non-linear current rises due to the non-linear action of Zener diode VR1. At an input voltage of approximately 60 VDC, the Zener diode begins to have a voltage.

Article 8. Suppressing current spikes by disabling low-cost outputs with soft-start

To meet stringent standby power consumption regulations, some multi-output power supplies are designed to disconnect the outputs when the standby signal is active.

This is usually accomplished by shutting down a series bypass bipolar junction transistor (BJT) or MOSFET . For low current outputs, BJTs can be a suitable and less expensive alternative to MOSFETs if the additional voltage drop across the transistor is taken into account when designing the power transformer.

Figure 10 shows a simple BJT series bypass switch with a voltage of 12 V, an output current of 100 mA, and a very large capacitor (CLOAD). Transistor Q1 is a series bypass element, and its switch is controlled by Q2 according to the state of the standby signal. The value of resistor R1 is rated to ensure that Q1 has enough base current to operate in saturation at minimum Beta and maximum output current. PI recommends adding an additional capacitor (Cnew) to regulate the transient current at turn-on. If Cnew is not added, Q1 quickly enters the capacitive load after turning on, and thus generates a large current spike. In order to regulate this transient spike, the capacitance of Q1 needs to be increased, which leads to an increase in cost.

Cnew acts as an additional "Miller capacitor" for Q1 to smooth out current spikes. This additional capacitance limits the dv/dt value at the collector of Q1. The smaller the dv/dt value, the less charging current flows into Cload. The capacitance value for Cnew is specified so that the ideal output dv/dt value of Q1 multiplied by the value of Cnew equals the current flowing into R1.

Formula 2

Figure 10: A simple soft-start circuit disables the power supply output during standby while eliminating the current spike during turn-on, thereby keeping costs low by using a small transistor (Q1).