The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> RF Simulators >> PSS results not matching with transient
https://designers-guide.org/forum/YaBB.pl?num=1425281918

Message started by nazgul on Mar 1^st, 2015, 11:38pm

Title: PSS results not matching with transient
Post by nazgul on Mar 1^st, 2015, 11:38pm

Hi guys,

I am using the following Cadence Version
"sub-version IC6.1.6-64b.500.3"

I was simulating an RF CMOS power amplifier intended for WLAN applications. It's a typical cascode architecture with both input and output interfaced to two different baluns. Both the baluns were designed using VeloceRF tool and the corresponding output spice model was used in my simulation. With the above set-up, I tried running PSS and transient analysis. PSS is converging but the result was not matching with transient. I tried decreasing the tolerance of PSS, even then I am facing the same issue.
What could be the reason for this mismatch??
Is it because of the balun or something else.
Normally which analysis is reliable PSS or transient??
If you are looking for any intrinsic details, Please comment down below

regards,
nazgul

Title: Re: PSS results not matching with transient
Post by sheldon on Mar 3^rd, 2015, 3:05am

Nazgul,

One note, ICXYX is the environment version number and
unrelated to the simulator version. Spectre RF is in the MMSIM
release stream.

Here some ideas to try:

1) You could try Harmonic Balance and see what the results are

2) Starting with a sine wave to test the circuit, does it reach
steady state by the end of the transient simulation?
if the transient simulation has not settled then allowing more time
to settle may resolve the issue

3) If you tighten the transient tolerances do the result match the
PSS results better?

Sheldon

Title: Re: PSS results not matching with transient
Post by nazgul on Mar 4^th, 2015, 8:46pm

Hi Sheldon,

Thanks for the reply.. :)
How to find out which spectreRF version I am using??
1.During my initial stages of design I have used harmonic balance, which later turned out to be inaccurate. I did some digging and found out PSS is more accurate than HB.
I want to mention one more observation at this point, When I was operating my PA at slightly lower input powers all the results were matching. The issue is rising only, when I was operating at slightly higher powers. There is a good of chance of my PA going into non-linear region at higher powers. Could it possibly be the reason??
2. When I tried to decrease the tolerance in PSS, the simulation is not converging. I don't know why this is happening.

regards,
nazgul

Title: Re: PSS results not matching with transient
Post by Ken Kundert on Mar 5^th, 2015, 12:02pm

SpectreRF is simply part of Spectre. To find out Spectre's version, use 'spectre -h', or 'spectre -V' or 'spectre -W'.

Many people cause problems for themselves by using absurdly tight tolerance settings. If you are using absurdly tight tolerance settings, stop. If you are not, and PSS does not converge despite using a reasonable tstab, then your circuit might not be periodic.

If it is a driven circuit, use strobing at the clock frequency and perform a long transient analysis. If the strobed waveforms become constant valued, then your circuit is periodic and SpectreRF should work. Use the settling time to determine a reasonable value of tstab. If they do not become constant valued, then you need to determine why your circuit is not periodic. I recommend that you observe the power supply currents when running this test.

-Ken

Title: Re: PSS results not matching with transient
Post by Frank Wiedmann on Mar 6^th, 2015, 1:17am

The Spectre version is also reported in the first lines of the Spectre log file (usually named spectre.out). Starting with MMSIM 14.1, Spectre can determine a suitable tstab value automatically when you set autotstab=yes (see http://support.cadence.com/wps/mypoc/cos?uri=deeplinkmin:ViewSolution;solutionNumber=20308900).

Title: Re: PSS results not matching with transient
Post by nazgul on Mar 9^th, 2015, 8:19am

Hi Ken,

Thank you for the reply..
The reason why I went with tight tolerances was, at normal tolerances my PSS results were not matching with transient results. I thought slightly tightening my tolerances will help me, but that didn't help me either.
During the normal transient I can see that my circuit is periodic, But the output is not a proper sine wave. Its slightly irregular with lot of higher harmonics, but its periodic.

regards,
nazgul

Title: Re: PSS results not matching with transient
Post by Ken Kundert on Mar 9^th, 2015, 4:59pm

Quote:

If it is a driven circuit, use strobing at the clock frequency and perform a long transient analysis. If the strobed waveforms become constant valued, then your circuit is periodic and SpectreRF should work. Use the settling time to determine a reasonable value of tstab. If they do not become constant valued, then you need to determine why your circuit is not periodic. I recommend that you observe the power supply currents when running this test.