Comparison of the advantages and disadvantages of 5 classic switch power supply structures, and several common issues encountered during debugging.

2021-11-09

Source:

Author:

Comparison of the advantages and disadvantages of 5 classic switch power supply structures, and several common issues encountered during debugging.

The switch power supply primarily refers to a high-frequency switch-mode DC regulated power supply made using various new self-shutting devices and conversion technology. The switch power supply is known as an efficient and energy-saving power supply, representing the development direction of regulated power supplies, and has become the mainstream product of regulated power supplies. Below, we will introduce a comparison of the advantages and disadvantages of 5 classic switch power supply structures.

Comparison of the advantages and disadvantages of 5 classic switch power supply structures

1. Single-ended forward

Single-ended: Drives the pulse transformer unidirectionally through a single switch device.

Forward: The phase relationship between the primary and secondary sides of the pulse transformer ensures that when the switch tube is conducting and driving the primary side of the pulse transformer, the secondary side of the transformer simultaneously supplies power to the load.

The problem with this circuit is: the switch tube T operates in two states, on/off. When the switch tube is off, the pulse transformer is in an 'open-circuit' state, and the stored magnetic energy will accumulate until the next cycle, until the inductor is saturated, causing the switch device to burn out.

 

2. Single-ended flyback

The flyback circuit is the opposite of the forward circuit, where the phase relationship between the primary and secondary sides of the pulse transformer ensures that when the switch tube is conducting and driving the primary side of the pulse transformer, the secondary side does not supply power to the load, i.e., the primary and secondary sides alternate between on and off. The issue of magnetic energy accumulation in the pulse transformer is easily resolved; however, due to the leakage inductance of the transformer, voltage spikes may occur on the primary side, potentially breaking down the switch device, necessitating the addition of a voltage clamping circuit for protection in the loop formed by D3 and N3. From the circuit schematic, the flyback and forward circuits look very similar, with only the difference in the names of the transformer terminals, but the operation methods of the circuits are different, and the roles of D3 and N3 are also different.

 

3. Push-pull (transformer center tap)

The characteristic of this circuit structure is: a symmetrical structure, where the primary side of the pulse transformer consists of two symmetrical coils, and two switch tubes are connected in a symmetrical manner, alternately turning on and off, with the operating process similar to a Class B push-pull power amplifier in a linear amplification circuit.

Main advantages: High utilization of the high-frequency transformer core (compared to single-ended circuits), high power utilization (compared to the half-bridge circuit to be discussed later), large output power, and both tube bases are at low levels, making the driving circuit simple.

Main disadvantages: Low utilization of transformer windings, and higher voltage requirements for the switch tubes (at least twice the power supply voltage).

 

4. Full bridge

The characteristic of this circuit structure is: four identical switch tubes are connected in a bridge structure to drive the primary side of the pulse transformer.

T1 and T4 form a pair, driven by the same signal group, turning on/off together; T2 and T3 form another pair, driven by another signal group, turning on/off together. The two pairs of switch tubes alternately turn on/off, creating positive/negative alternating pulse currents in the primary side coil of the transformer.

Main advantages: Compared to the push-pull structure, the primary side winding is reduced by half, and the voltage requirement for the switch tubes is also halved.

Main disadvantages: A larger number of switch tubes are used, requiring good parameter consistency, and the driving circuit is complex, making synchronization difficult. This circuit structure is generally used in ultra-high power switch power supply circuits above 1KW.

 

5. Half bridge

The structure of the circuit is similar to that of the full bridge, except that the two switch tubes (T3 and T4) are replaced with two equivalent large capacitors C1 and C2.

Main advantages: It has a certain ability to resist imbalance, and the requirements for circuit symmetry are not very strict; it can accommodate a wide range of power ratings, from tens of watts to kilowatts; the voltage requirements for the switch tubes are lower; and the circuit cost is lower than that of the full bridge circuit. This circuit is often used in various non-regulated output DC converters, such as in electronic fluorescent lamp driver circuits.

 

Switch power supplies are widely used in various electronic devices today, and their unit power density is continuously improving. The definition of high power density has increased from 25w/in3 in 1991, 36w/in3 in 1994, 52w/in3 in 1999, 96w/in3 in 2001, to now reaching hundreds of watts per cubic inch. This is due to the use of many high-power semiconductor devices in switch power supplies, such as rectifier bridge stacks, high current rectifier tubes, high-power transistors, or field effect transistors.

 

The 10 most common problems during switch power supply debugging

1. Transformer saturation

Transformer saturation phenomenon: When powered on under high or low voltage input (including light load, heavy load, capacitive load), output short circuit, dynamic load, high temperature, etc., the current through the transformer (and switch tube) increases non-linearly. When this phenomenon occurs, the peak value of the current cannot be predicted or controlled, which may lead to over-stress on the current and thus cause over-voltage damage to the switch tube.

1. Situations that easily lead to saturation:

(1) Transformer inductance is too large;

(2) Number of turns is too few;

(3) The saturation current point of the transformer is smaller than the current limit point of the IC;

(4) No soft start.

2. Solutions:

(1) Reduce the current limit point of the IC;

(2) Strengthen the soft start to make the current through the transformer rise more slowly.

 

2. Vds too high

1. Vds stress requirements:

Under the worst conditions (high input voltage, large load, high ambient temperature, power supply startup or short circuit test), Vds should not exceed 90% of the rated specification.

2. Methods to reduce Vds:

(1) Reduce the path voltage: reduce the turns ratio of the transformer primary and secondary;

(2) Reduce the spike voltage:

a. Reduce leakage inductance:

The leakage inductance of the transformer is the primary reason for the occurrence of this peak voltage during the switching tube registration, and reducing leakage inductance can decrease the peak voltage.

b. Adjust the absorption circuit:

① Use TVS tubes;

② Use slower diodes, which can absorb a certain amount of energy (spikes);

③ Inserting damping resistors can make the waveform smoother, which helps reduce EMI.

3. IC temperature too high

Causes and solutions:

1. Internal MOSFET losses are too large:

Excessive switching losses and large parasitic capacitance of the transformer result in a large overlap area between the MOSFET's registration and turn-off current with Vds. Solution: Increase the distance between transformer windings to reduce inter-layer capacitance, such as adding a layer of insulating tape (inter-layer insulation) when winding in multiple layers.

2. Poor heat dissipation:

A significant portion of the IC's heat relies on the pins to conduct to the PCB and its copper foil, so it is advisable to increase the area of the copper foil and apply more solder.

3. Ambient air temperature around the IC is too high:

The IC should be placed in an area with good air circulation and kept away from components that generate excessive heat.

4. Cannot start under no load or light load

1. Phenomenon:

Cannot start under no load or light load, Vcc repeatedly fluctuates between the start voltage and the turn-off voltage.

2. Cause:

Under no load or light load, the induced voltage of the Vcc winding is too low, leading to repeated restart conditions.

3. Solution:

Increase the number of turns in the Vcc winding, reduce the Vcc current limiting resistor, and appropriately add a dummy load. If increasing the number of turns in the Vcc winding and reducing the Vcc current limiting resistor causes Vcc to become too high under heavy load, refer to methods for stabilizing Vcc.

5. Cannot add heavy load after starting

Causes and solutions:

1. Vcc is too high under heavy load

Under heavy load, the induced voltage of the Vcc winding is relatively high, causing Vcc to exceed the OVP point of the IC, triggering the over-voltage protection of the IC, resulting in no output. If the voltage further increases beyond the IC's tolerance, the IC will be damaged.

2. Internal current limiting is triggered

a. Current limiting point is too low

Under heavy load and capacitive load, if the current limiting point is too low, the current flowing through the MOSFET is restricted, leading to a lack of output. The solution is to increase the resistance of the current limiting pin to raise the current limiting point.

b. Current rise rate is too large

If the rise rate is too large, the peak current will be greater, easily triggering internal current limiting protection. The solution is to increase the inductance without saturating the transformer.

6. Standby input power is too high

1. Phenomenon:

Vcc is insufficient under no load or light load. This situation can cause excessive input power and large output ripple under no load or light load.

2. Cause:

The reason for excessive input power is that when Vcc is insufficient, the IC enters a repeated restart condition, frequently requiring high voltage to charge the Vcc capacitor, resulting in losses in the startup circuit. If there is a resistor in series between the startup pin and high voltage, the power loss across this resistor will be significant, so the power rating of the startup resistor must be adequate. The power supply IC has not entered Burst Mode or has already entered Burst Mode, but the Burst frequency is too high, leading to excessive switching losses.

3. Solution:

Adjust the feedback parameters to slow down the response speed.

7. Short circuit power is too high

1. Phenomenon:

When outputting a short circuit, the input power is too high, and Vds is too high.

2. Cause:

During an output short circuit, there are many repeated pulses, and the peak current of the switching tube is very large, causing excessive input power. The excessive current in the switching tube stores too much energy in the leakage inductance, leading to high Vds when the switching tube turns off. There are two possible causes for the switching tube to stop working during an output short circuit:

(1) Triggering OCP can cause the switching action to stop immediately.

a. Triggering the OCP of the feedback pin;

b. Switching action stops;

c. Vcc drops to the IC's turn-off voltage;

d. Vcc rises back to the IC's start voltage, and restarts.

(2) Triggering internal current limiting

When this method occurs, it limits the duty cycle, relying on Vcc to drop to the UVLO lower limit to stop the switching action, and the time for Vcc to drop is relatively long, meaning the switching action persists for a longer time, resulting in larger input power.

a. Triggering internal current limiting, duty cycle is limited;

b. Vcc drops to the IC's turn-off voltage;

c. Switching action stops;

d. Vcc rises back to the IC's start voltage, and restarts.

3. Solution:

(1) Reduce the number of current pulses, so that when a short circuit occurs, triggering the OCP of the feedback pin can cause the switching action to stop quickly, reducing the number of current pulses. This means that when a short circuit occurs, the voltage at the feedback pin should rise faster. Therefore, the capacitance at the feedback pin should not be too large;

(2) Reduce peak current.

8. Output ripple is too large under no load or light load

1. Phenomenon:

Vcc is insufficient under no load or light load.

(1) Reason:

When Vcc is insufficient, the oscillating IC operates intermittently for a longer period between the startup voltage (e.g., 12V) and the shutdown voltage (e.g., 8V), supplying energy to the output for a short time, followed by a longer period of intermittent operation. This results in insufficient energy stored in the capacitor to maintain stable output, causing the output voltage to drop.

(2) Handling method:

Ensure that Vcc can supply stably under any load conditions.

2. Phenomenon:

In Burst Mode, the frequency of intermittent operation is too low, and this low frequency prevents the output capacitor's energy from maintaining stability.

Handling method:

Slightly increase the frequency of intermittent operation while meeting standby power requirements, and increase the output capacitance.

9. Heavy load and capacitive load cannot start

1. Phenomenon:

It can start under light load, and can also take on a heavy load after starting, but cannot start under heavy load or large capacitive load conditions.

2. General design requirements:

Regardless of heavy load or capacitive load (e.g., 10000uF), the input voltage remains low, and within 20mS, the output voltage must rise to a stable value.

3. Causes and handling methods (ensuring Vcc is within normal operating design):

Taking a capacitive load of C=10000uF as an example, according to specifications, sufficient energy must be available to raise the output to a stable output voltage (e.g., 5V) within 20mS.

E=0.5*C*V^2

The larger the capacitance C, the greater the energy required to be transferred from input to output within 20mS.

Taking the chip FSQ0170RNA as an example, the total area S of the shadow part represents the required energy. To increase area S, the methods are:

(1) Increase the peak current limit point I_limit, allowing a larger inductor current Id to flow: increase the resistance connected to Pin4, reducing the shunt from the internal current source Ifb, which will raise the voltage at the positive input of the PWM comparator that references the current limit voltage, thus allowing a larger current to pass through the MOSFET/transformer, providing more energy.

(2) At startup, increase the time for energy transfer, i.e., extend the rise time of Vfb (before reaching the OCP protection point).

For the FSQ0170RNA chip, the inductor current control is referenced to Vfb, and the waveform of Vfb is proportional to the envelope of the inductor current. Controlling the rise time of Vfb can control the rise time of the inductor envelope, thus increasing the time for energy transfer. The IC's OCP function detects when Vfb reaches Vsd (e.g., 6V) to complete. Therefore, to reduce the slope of Vfb, the rise time of Vfb can be extended. When the output voltage has not reached the normal value, if the feedback pin voltage Vfb has already risen to the protection point, the energy transfer time is insufficient. When starting under heavy load or capacitive load, the output voltage builds up slowly, resulting in a lower voltage applied to the optocoupler, causing a small current through the optocoupler diode, and the phototransistor of the optocoupler remains in a high-resistance state (tending to turn off) for a longer time. The internal current source of the IC charges the capacitor connected to the feedback pin quickly, and if Vfb rises to the protection point (e.g., 6V) during this time, the MOSFET will turn off. The output cannot reach the normal value, and the startup fails.

Handling method:

When the output voltage reaches the normal value, the feedback pin voltage Vfb is still less than the protection point. Keep Vfb away from the protection point and allow it to rise slowly, or extend the time for the feedback pin Vfb to rise to the protection point, i.e., reduce the rise slope of Vfb, allowing the output sufficient time to rise to the normal value.

A. Increase the feedback capacitor (C9), which can reduce the rise slope of Vfb, as shown in the diagram, changing from line D to line A. However, if the feedback capacitor is too large, it will affect normal operation, reduce response speed, and increase output ripple. Therefore, this capacitor should not be changed too much.

B. Due to the shortcomings of method A, connect a capacitor (C7) in series with the voltage regulator (D6, 3.3V) in parallel to the feedback pin. This method will not affect normal operation, as shown by line B. When Vfb < 3.3V, the voltage regulator will not conduct and shunt. When it rises to 3.3V, the voltage regulator enters a regulation state, and capacitor C7 begins to charge and shunt, reducing the subsequent rise slope of Vfb. C. Connect a capacitor (C11) in parallel with the K-A terminal of the 431. When the power is started, the voltage of C11 is low and charges through the optocoupler diode and the bias resistor R10 of the 431. This allows a larger current to pass through the optocoupler, reducing the impedance of the phototransistor and shunting, causing Vfb to rise slowly, as shown by line C. R10×C11 affects the charging time, which in turn affects the rise time of the output.

Points to note:

① Increasing the feedback pin capacitor (including the capacitor in series with the voltage regulator) has little effect on solving the problem of extremely large capacitive loads;

② Increasing the peak current limit point I_limit also raises the OCP point under steady state. It is necessary to check whether the transformer will saturate under capacitive load and input conditions;

③ If the current limit point is to be maintained, R10×C11 must be made larger, but under extremely large capacitive loads (10000uF), this may increase the rise time of 5Vsb beyond 20mS, and this method needs to check whether the dynamic response is too affected;

④ The bias resistor R10 of the 431 is too small, and the capacitor C11 in parallel with the 431 should be larger;

⑤ To ensure rise time, increasing the OCP point and the method of increasing R10×C11 may need to be used together.

10. Output bounce under no load or light load

1. Phenomenon:

When the output is unloaded or lightly loaded, turning off the input voltage may result in a voltage bounce waveform as shown in the figure below.

2. Cause:

When the input is turned off, the 5V output will drop, and Vcc will also drop, causing the IC to operate intermittently. However, under no load or light load, the large capacitance voltage of the PC power supply does not drop quickly, still providing a large current to the high-voltage startup pin, causing the IC to restart, and the 5V output will bounce back.

3. Solution:

When initiating the pin with a larger current-limiting resistor, even when the voltage of the large capacitor drops to a still relatively high level, there is a lack of sufficient starting current to supply to the IC. By connecting the start to the rectifier bridge, the start is not affected by the voltage of the large capacitor. When the input voltage is turned off, the voltage at the start pin can drop rapidly.

 

The above is a comparison of the advantages and disadvantages of five classic switch power supply structures, as well as the ten most common problems encountered during switch power supply debugging. Nowadays, switch power supplies are becoming increasingly widespread, and high frequency is one of the main directions for the development of switch power supply technology, as well as one of the main trends in the development of high-frequency switch rectifiers. However, as the switching frequency increases, the switching losses of power devices will increase proportionally. Therefore, when the switching frequency is relatively high, it is necessary to adopt very effective 'softening' methods to minimize the switching losses of the devices. Currently, a popular method is to use active soft-switching technology, such as resonant technology, quasi-resonant (or multi-resonant) technology, ZCS-PWM (or ZVS-PWM) technology, and ZCT-PWM (or ZVT-PWM) technology. Another more practical method is to use passive lossless soft-switching technology, which involves using passive components (L, C, D, etc.) to form a unique (patented) circuit network to achieve lossless switching for power switches.

Previous: