Andrew Beckett wrote on Sep 17th, 2019, 4:06am:You can do this by running a stability analysis (stb) in Spectre, and then using the Results Browser go to the "stb" results and picking that the graph type will be either "Polar" or "RealvsImag" rather than "Default" (which is "Rectangular") and then double clicking on the "loopGain" signal within the stb results.
It's probably polar that you want rather than real vs imaginary because that then shows both the unit circle and also the angles - so it's easier to assess the stability that way.
Regards,
Andrew
Thanks Andrew !
It's strange that I could not find the RealvsImag in the Result Browser, only the Polar is available.
The case I have is strange, the phase margin is >60deg, Gain margin > 15dB and yet I have more than 3 rings in the transient. I don't think it is due to large signal instability i.e. it tripped some comparator. I notice that the Q factor due to the complex pole can be a factor, i.e. the higher the gain-peaking, the worse the ringing it will be. Maybe there's another better way to assess the stability due to the gain-peaking that can be recommended ?