Hi all,

I have an SC amplifier which has multiple feedback loops.

I get incompatible results while simulating my amplifier with different methods. I did a PSTB analysis
on the circuit to assess its stability, and found that although stable, it had a peaking in its response in 1 of the
paths.

I tried to see the impact of this by injecting disturbances at the offending frequency in voltage/current form in this
loop, but failed to see any corresponding amplification of the disturbance. Then I tried the same but varied the frequency
of the perturbation, expecting to see some variation in the amplitude of the perturbation at different points in the loop.

However, contrary to results from PSTB which show a distinct frequency dependence, I got almost no variation in the
multiple TRAN runs. I must say I am baffled.

I use the traponly integration method in TRAN

I  have checked that the PSS analysis gives steady-state signal levels which match the steady state behavior in TRAN.

The setups are identical except that the cmdmprobe from PSTB is replaced by the perturbation sources in the TRAN setup. I add
a voltage of the order of 100 mV and current perturbation of about 100 nA, so as to avoid disturbing my circuit too much.

I have used square waves, sine waves and damped sine waves to perturb the system, all with the same result.

Any hints would be welcome on how to interpret this inconsistency between the 2 analyses. The circuit is stable as per
both methods.

Thanks
Vivek

Hi Tosei,

I do have a situation similar to what you outline. I have got 4 feedback loops (3 diff. and 1 common-mode). These do not all share
a common point where they all might be broken with cmdmprobe.

It is possible to break 2 of the loops together at 1 point, and 2 others at another. So, I did 2 different analyses.
As for individual stability of the loops, I expect that they are all stable.

I am aware that one needs to break all loops together, but I was expecting that I would atleast get a rough idea
of the behavior even when this condition was not fulfilled.

I was trying to check if the loops are well-behaved, and tried to break them 2 at a time as mentioned above.
However, the frequency-dependent peaking in 1 of the loops, which is seen with PSTB is not apparent in
transient.

If I understand you correctly, I cannot correlate PSTB results with TRAN. In that case, my question is answered.

Thanks very much for the clarification.

Regards
Vivek