The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> Circuit Simulators >> stb analysis in Spectre sensitive to probe positio
https://designers-guide.org/forum/YaBB.pl?num=1187187562

Message started by vivkr on Aug 15^th, 2007, 7:19am

Title: stb analysis in Spectre sensitive to probe positio
Post by vivkr on Aug 15^th, 2007, 7:19am

Hi,

I am perplexed to find that the stability information (gain & phase margin) of my circuitry
are showing up differently depending on where I insert my probe element during the stb analysis
in Spectre.

The circuit is single-loop and I am using the loop-based algorithm, taking care to ensure that
the probe element selected is always in the loop. I use a 0V DC voltage source for the probe element.

Does anyone have any suggestion as to how I can avoid this problem?

Regards
Vivek

Title: Re: stb analysis in Spectre sensitive to probe pos
Post by sheldon on Aug 18^th, 2007, 3:30pm

Vivek,

How does the loop behave when using the current probe? Can
you publish the schematic so we can look at the circuit? Also have
you tried other methods for analyzing the stability, for example,
the two-port method, see "The Designers Guide to SPICE and
SPECTRE"? Sometimes, separating the forward and reverse
characteristics can provide clues about what the issue might be.

Best Regards,

Sheldon

Title: Re: stb analysis in Spectre sensitive to probe pos
Post by vivkr on Aug 20^th, 2007, 12:02am

Hi Sheldon,

Thanks for the tips. Let me try these different methods. Unfortunately, I cannot post the schematic.
It is just a simple two-stage amplifier, where the compensation is at the output (output load >>>> any possible on-chip cap
by about 4-5 orders of magnitude). The first stage is a simple diff. gain stage, the feedback path has an R-divider.
I was experimenting with an additional level shifting stage between the first and the second stages.

There are several capacitive paths which might act as parallel paths at high-f, such as gate cap etc. So, I placed the probe
where I was sure that there were no signal paths in parallel to it.

The only possible reason I see is that the supply is not ideal (source impedance of ~ 2 Ohms), but I would not have expected
the kind of variation that I saw just because of this (~ 5 degree variation in PM).

But let me check the methods you describe.

Regards
Vivek

Title: Re: stb analysis in Spectre sensitive to probe pos
Post by sheldon on Aug 20^th, 2007, 7:25pm

Vivek,

Is it possible that the power supply resistance is introducing an additional
feedback loop? I have seen power supply resistance, extracted-R, act as a
feedback loop around two stage amplifiers.

Best Regards,

Sheldon

Title: Re: stb analysis in Spectre sensitive to probe pos
Post by vivkr on Aug 22^nd, 2007, 7:22am

Hi Sheldon,

I am quite sure that the supply forms a second loop. The only problem I had was that I did not expect the amount of variation I got.
I was expecting more like 1-2 degrees change in phase margin, not in the range of 5-10. Perhaps my assumptions were incorrect then.

Anyway, I did AC and TRAN analyses to conclude that I had enough stability margin. In any case, STB analysis also gave me large enough
margin, even if the exact value was dependent on the probe location.

Thanks for your suggestions.

Regards
Vivek