Stability of PA by Impulse Response

Stability of PA by applying a step input

Friends,

I am evaluating the stability of my PA. K_factor ( stab_fact(S) ) and B1 (stab_meas(S)) are more than 1 and 0, respectively at all frequencies. So my PA should be stable.
I have done another test too. I have applied an step signal at input and seen output voltage. As it is clear in the figure it oscillates and is not stable!!! why?!!! I have used a transient analysis for step response.

Using step response is a mistake to evaluate stability or my circuit is not really stable?!
Could you help me?

How many stages are there on your PA ?? Stability factors are valid for single stage amplifiers only.
Also, small signal stability does NOT mandatorily bring large signal stability because large signal s-parameters different than small signal s-parameters which are used to compute K ( or Mu ).

BigBoss said:

How many stages are there on your PA ?? Stability factors are valid for single stage amplifiers only.
Also, small signal stability does NOT mandatorily bring large signal stability because large signal s-parameters different than small signal s-parameters which are used to compute K ( or Mu ).

Click to expand...

It has just one stage.
"small signal s-parameters which are used to compute K" : I am using nonlinear model of the transistor. So, s- parameters calculated by ADS (the software) are not small signal.

f_t said:

It has just one stage.
"small signal s-parameters which are used to compute K" : I am using nonlinear model of the transistor. So, s- parameters calculated by ADS (the software) are not small signal.

Click to expand...

It's not correct.S-parameter simulation in ADS -by default- computes always small signal s-parameters by linearizing ALL devices in the circuit at their vicinity of DC operating points.It's essentially nothing but a kind of AC simulation technique.Using nonlinear model doesn't change this.Therefore there is another palette LSSP to do "Large Signal S Parameters" simulations.

how about an ideal stepped RF input, and look for transient phase shifts and over/undershoots at the output. An unstable amp will be very "ringy". another option would be stepped RF input, and look at the output spectrum for odd spurious frequencies showing up.

BigBoss said:

It's not correct.S-parameter simulation in ADS -by default- computes always small signal s-parameters by linearizing ALL devices in the circuit at their vicinity of DC operating points.It's essentially nothing but a kind of AC simulation technique.Using nonlinear model doesn't change this.Therefore there is another palette LSSP to do "Large Signal S Parameters" simulations.

Click to expand...

You said correctly BigBoss. I changed s-param to LSSP and again tried to see k factor and B. They did not show a stable PA. K factor was lower than 1 in some frequencies, however it was more than 1 by s-param (small signal) simulation. Thank you.

- - - Updated - - -

biff44 said:

how about an ideal stepped RF input, and look for transient phase shifts and over/undershoots at the output. An unstable amp will be very "ringy". another option would be stepped RF input, and look at the output spectrum for odd spurious frequencies showing up.

Click to expand...

Thank you biff44.
How can I see transient phase shift in ADS? When apply a step, I can just see vload not phase of vload in ADs. Do I have to add another probe for that?
Besides, what is base of your analysis? Why does an unstable amplifier have a ringy phase?

It shows a 2nd order underdamped impedance with path delay. What layout and logic diagram can you provide?