Flyback - Stability, Oscillations, Buzzing

After extensive comments and tentative fixing of this and that I believe my original thought about the control loop becoming unstable was correct. The original two op-amp circuits used in the current control loop were this: -

enter image description here

And the spectrum is this: -

enter image description here

Things to note about this spectrum: -

  • Peak gain (67 Hz) is circa 64 dB
  • At 67 Hz, the phase response is where we want it to be i.e. about 180 degrees.
  • Unity gain happens at circa 600 kHz
  • At 600 kHz the phase angle has shifted a further 215.7 degrees from 180 degrees (orange box)

Hence, due to the massive shift in phase whilst still operating with greater than unity gain there has to be significant cause for concern that this add-in circuit to an already established control loop (as made by the TOP267 chip) is going to bring problems of instability.

If we look at the transient response with a small 1 kHz square wave input (just to get a feel for the delays involved) we see this: -

enter image description here

The signal delays for when Vin drops is about 80 us.

My strong conclusion is that this add-in circuit to the control loop feedback will cause instability.

Later on, through discussion in comments, the CC control-loop was got working by removing the first op-amp stage and lowering the integrator's series feedback resistor to 1 kohm (R6 in my diagram and R5 in the OP diagram). There are still problems because the op-amp in either the original circuit or the modified circuit (that is stable) doesn't have sufficiently low input offset voltage and, as a result, DC accuracy is problematic. This is a different problem and might be solved by using a precision op-amp with 10 uV input offset voltage like the ADA4522.


One possibility is that magnetostriction in your lab-wound transformer is causing instability in the control loop. The audible buzz hints at an audio range frequency (perhaps a sub-harmonic of 66 kHz). The vibrations in turn cause the transformer's impedance to fluctuate, and that could be interfering with control loops inside the switcher. Buzzing can be damped by potting, as a commenter suggested, but in addition it can also prevent instability.

https://en.wikipedia.org/wiki/Magnetostriction