Alright, I'm modelling some amplifier designs in LTspice and would like to take a look at loop stability. I used to just do an AC analysis of the output and look for peaking in the frequency and phase response, but that doesn't tell you much about the loop stability.
A fellow forum member suggested that the loop gain can be modeled by shorting the amplifier's input and moving the voltage source into the feedback loop and then divide the voltage at the output by the voltage after the voltage source. This should then give you the loop gain and loop phase.
Let me illustrate this with a few images of the simulation of a simple MOSFET class-AB amplifier:
As can be seen the voltage source V3 is inserted between the output and the feedback resistor and the relation between V(out) and V(A) will give us the loop gain and phase.
This is accomplished simply by dividing V(out) by -V(A) after having done a AC analysis of the amplifier. For the amplifier we're simulating here the following plot is then generated:
The Nyquist thereom specifies that for an amplifier to be stable the phase shift at the output when the amplifier's loop gain has dropped to 0dB must not be greater than -180 degrees. In the above plot it is clear that when gain=0dB the phase is -105 degrees. Hence the amplifier will be stable.
The phase margin is the difference between the phase at gain=0dB and -180 degrees, thus in this case the phase margin is 180-105=75 degrees.
However when simulating a different amplifier design the following plot is generated:
The phase at gain=0dB is -100 degrees, which would indicate stability, however unlike in the previous plot the phase doesn't start at 0 degrees, but at -90 degrees. Hence the net phase shift is 100-90=10. Thus the phase margin is a very healthy 170 degrees.
Does the above make sense? Or are there different (better? simpler?) ways to model loop gain and phase with LTspice?
A fellow forum member suggested that the loop gain can be modeled by shorting the amplifier's input and moving the voltage source into the feedback loop and then divide the voltage at the output by the voltage after the voltage source. This should then give you the loop gain and loop phase.
Let me illustrate this with a few images of the simulation of a simple MOSFET class-AB amplifier:
As can be seen the voltage source V3 is inserted between the output and the feedback resistor and the relation between V(out) and V(A) will give us the loop gain and phase.
This is accomplished simply by dividing V(out) by -V(A) after having done a AC analysis of the amplifier. For the amplifier we're simulating here the following plot is then generated:
The Nyquist thereom specifies that for an amplifier to be stable the phase shift at the output when the amplifier's loop gain has dropped to 0dB must not be greater than -180 degrees. In the above plot it is clear that when gain=0dB the phase is -105 degrees. Hence the amplifier will be stable.
The phase margin is the difference between the phase at gain=0dB and -180 degrees, thus in this case the phase margin is 180-105=75 degrees.
However when simulating a different amplifier design the following plot is generated:
The phase at gain=0dB is -100 degrees, which would indicate stability, however unlike in the previous plot the phase doesn't start at 0 degrees, but at -90 degrees. Hence the net phase shift is 100-90=10. Thus the phase margin is a very healthy 170 degrees.
Does the above make sense? Or are there different (better? simpler?) ways to model loop gain and phase with LTspice?
I'm not experienced in such matters, but I've seen other engineers at work using a low value resistor added into the feedback loop somehow to use real equipment (i.e. not a sim) to measure the response of switch mode PSUs so as to determine compensation values for stability. You may have more luck searching for your desired method with regard to SMPSUs and adapting it to audio amps.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.