Different result between sp & hpsp frequency response

Dear Binh,

There are a lot of unknowns to your question, but I might comment on one item and suggest a second.

1. It appears you have enabled noise in your Harmonic Balance SP analysis and in your S-parameter analysis it is disabled. Is this intentional?

2. An S-parameter analysis is a linear analysis and linearizes your circuit about its DC operating point. A Harmonic Balance SP analysis linearizes the circuit about a periodic operating point. I noted you did save the initial transient of the Harmonic Balance analysis. Have you compare it to the DC operating point of your SP analysis?  I am wondering if the two operating points are placing the devices of your circuit at significantly different points on their characteristics. As a result the S parameters will differ significantly as the gain and phase of the linear network and linearized periodic network will be significantly different.

Shawn

thanks for your reply, 

so what you address out that because of saving the initial transient of HB analysis, the result of s params may be different 

thanks for your reply, 

so what you address out that because of saving the initial transient of HB analysis, the result of s params may be different 

Dear Binhngo1210,

Binhngo1210 said:

so what you address out that because of saving the initial transient of HB analysis, the result of s params may be different 

I apologize in advance as I'm not quite sure what you are asking in your response. Perhaps I need to clarify my comment. I was suggesting you compare the operating point of the waveform at the end of your initial transient of the PSS simulation to the DC operating point conditions used in your small-signal S parameter simulation. Are the operating points of the transistors in both the large-signal transient simulation and the DC simulation the same? In other words, as an example, are the gm of your differential pair transistors and their bias currents identical at the end of your transient simulation to those in your DC simulation?

If they are not, your two sets of results will not be the same - which is what I believe you are observing.

Shawn

Do also bear in mind that with the hbsp you have an additional frequency involved. The large-signal frequency input is not the same as the stimulus for the hbsp - so it's effectively computing the s-parameters in the presence of that large signal - so you can see the impact of that input saturating the circuit on the s-parameters, but the s-parameter analysis is a small-signal analysis on top of that large-signal effect. Of course, if the large-signal input is very small, then that's unlikely to be the issue here - but as Shawn said, presumably the operating point is very different - it's really hard to tell from the limited information you've given so far.

Regards,

Andrew