Some Views on Switching Power Supply Loop Design

P adjustment. It is pure resistance , without C, L. This adjustment is attenuation or amplification. It makes the system have static error. The open-loop gain increases, the steady-state error decreases, fc increases, the transition process shortens, and the system stability deteriorates.

This is rarely used.

Improve it, PI regulation: eliminate static error. For example, place two components between R and K of 431, R in series with C.

The advantage is that it provides a negative phase angle because there is a pole and a zero point. The pole is at 0. The

phase margin is reduced .

Therefore, the relative stability of the system is reduced.

However, the crossover frequency fc is increased.

PD regulation. This is not used much. PD regulation increases the fc of the system, resulting in faster system response and increased phase margin. There is noise at high frequencies.

PID regulation: PI at low frequencies, PD regulation at higher frequencies.

Improve static performance at low frequencies, and improve stability and response speed at high frequencies.

Flyback. The most commonly used is the improved PI, that is, type II and III. So, what should the ideal transfer function look like: 1. Low frequency band: high gain to reduce static error 2. Mid-frequency band: near fc, -20db, to ensure sufficient phase margin 3. High frequency band: small gain to reduce the impact of switching harmonics and noise. If the -40db drop cannot solve the problem, then add a low-pass filter. If TYPE II is not enough to provide enough phase margin, then try TYPE III. To summarize: Low frequency band: steady-state performance Mid-frequency band: dynamic performance High frequency band: anti-interference performance The larger the fc, the better the rapidity, but the anti-interference ability decreases The mid-frequency band can best reflect the stability and rapidity of the system . P: coarse adjustment, that is, DC gain. If it is too large, it may oscillate. It is to make a difference between the current value and the given value and amplify it. I: fine adjustment, integrate the error . D: prediction function, this can be seen in the automatic control book. If D is large, burrs will be generated. Determine the trend of the current value change, make adjustments in time, reduce the adjustment time, and improve the response speed. There are many adjustment methods, but the soul is P, which must be there. Whether there is I and D depends on the actual situation. In fact, we use the improved PI in the switching power supply , that is, type II, type II. D is rarely used. D, that is, at the output of the power supply , RC is connected to the 2.5V reference pin in series. We usually don't do this. As for the improved PI regulation, it is explained in the automatic control book, so I won't go into details. There are a lot of information about type II and type III on GOOGLE. The calculations in this regard have also been completely formulated. Switching power supplies mainly use these two compensations. Type III is relatively rarely used. When we adjust the loop, we mainly adjust this compensation circuit . I have a question: We output DC and collect DC. So why do we still use op amps to amplify? What is the purpose of adding RC? What is the purpose of compensation? What do these things related to AC frequency have to do with DC? Isn't DC separated by capacitors ? So how to answer the above questions? The model of the switching power supply has three entries: 1. Reference input vref 2. Input bus voltage vin 3. Load disturbance Io The changes in 2 and 3 can be considered as AC. The purpose of feedback is to make the output voltage still stable under these disturbances. Let's talk about the role of PC817: PC817 is a linear optocoupler, the dynamic resistance of the collector-emitter is determined by the primary current iF and the collector current iC. iF is controlled by the three-terminal adjustable voltage regulator TL431 for feedback . The output voltage rises . The output sampling resistor, the voltage of the lower one rises, the VAK of TL431 decreases, iF increases, the secondary VCE of the optocoupler decreases, if 2 is grounded, 1 feedback is connected to pin 1 , then the voltage of pin 1 decreases, the duty cycle D decreases, and the output voltage decreases. So it is stable. In fact, VCE and iF form negative feedback. It is easy to understand. At this time, TL431 is connected to RC compensation, that is, TYPEII for optocouplers and 3842, pin 2 is grounded, and the emitter of the optocoupler is also grounded. Pull a 1K or slightly larger resistor from pin 8 to the collector of the optocoupler. Pull this feedback signal directly to pin 1. Feedback is completed. As for between pins 1 and 2, a pf-level capacitor can be connected. It cannot be too large.