I am designing a flyback converter, with the following parameters:

• Input: 50-60VDC
• Output: 150VDC

I use a simple solution using UC3843, TL431, and PC817. Currently, the circuit is running and has a stable 150V output. However, there is a problem that the transformer is making a loud sound, I found out that the cause is due to the uneven gate pulse, as shown below. I think this noise comes from the Isense part. Can anyone with experience help me with this case?

 

What do the waveforms in Isense and COMP look like?

 

The problem is the timing of the error amplifier TL431 + RCs. The supply is oscillating. Like Ian0 said there is large amounts of ripple on the TL431.
This supply will not start up. Look at an offline switcher using that IC. Your 15V supply does not have to be regulated. Normally I wand the transformer for about 15V out.
C14, R19 is a low pass filter that is probably set too low a frequency. You want a ramp at pin 4.
What is the part numbers on the diodes?

This is how most startup circuits work for that IC.

 

There also may be a subharmonic oscillation going on - the first pic shows the same waveform repeated twice. Does it keep repeating like this? If so, the feedback is unstable.
I would suggest connecting the collector of the opto to COMP. Otherwise C13 and IC2 make one 90° phase shift, the UC3842 error amp and C16 make another 90° and there is a third in the TL431 itself and before long you have an oscillator.
C13 may be too small, because the C13/R14 time constant which sets the roll-off of the (real) error amplifier is only 1us, and the operating period is 13us

 

A further thought. . . what is the purpose of R17? Apart from coupling supply ripple into the current-sense circuit?

 

the UC3842 error amp and C16 make another 90°

The error amp in UC2842 is not functioning. No inversion.

 

The error amp in UC2842 is not functioning. No inversion.

Yes it is, the output from the optoisolator goes into it through a 2.4k resistor, and it has 100k in parallel with 100pF to set the gain, to 41, with a 3dB point at 16kHz, at which point it contributes a phase shift of 45°

 

Yes it is,

You are right. I normally drive the comp pin. There is a problem.

 

You are right. I normally drive the comp pin. There is a problem.

Yes, too many phase shifts.