The Designer's Guide Community
Forum
Welcome, Guest. Please Login or Register. Please follow the Forum guidelines.
Oct 2nd, 2022, 3:58am
Pages: 1
Send Topic Print
PSS + PSTB for switching TIA (Read 189 times)
eshau
New Member
*
Offline



Posts: 5

PSS + PSTB for switching TIA
Aug 10th, 2022, 8:49am
 
Greetings,

I am receiving questionable results from PSS+PSTB and I am unsure how to proceed.

The circuit schematic and clock timing diagram are provided in Fig.1.
The results for Loop Gain & Phase in Fig. 2.

Please observe the abnormalities in the loop gain which do not seem realistic or even possible to achieve.

The switching frequency is ~500 kHz. I have tried increasing maxacfreq in the PSS options to 1 GHz, 2 GHz, 3 GHz... but the same behavior persists.

Any feedback is much appreciated.

Best,
Edoh
Back to top
 

figs.png
View Profile   IP Logged
Ken Kundert
Global Moderator
*****
Offline



Posts: 2349
Silicon Valley
Re: PSS + PSTB for switching TIA
Reply #1 - Aug 10th, 2022, 8:07pm
 
You did not show the location of the stability probe, or what you are using for your stability probe.  Inappropriate placement of the stability probe is a common source of problems in a stability analysis.

The things you are seeing at high frequency look like sampling effects.  You should request many more frequency points so you can resolve the frequencies and shapes of the peaks.  That can help explain them.  I expect you will find that they lie at multiples of the clock frequency.

-Ken
Back to top
 
 
View Profile WWW   IP Logged
eshau
New Member
*
Offline



Posts: 5

Re: PSS + PSTB for switching TIA
Reply #2 - Aug 11th, 2022, 1:08am
 
Hi Ken,

Thank you very much for taking the time to respond. It's a privilege.

I have uploaded my test bench where you can see the probe location.

My circuit is fully differential and so I am using a diffstbprobe, accordingly.

Also included is a higher resolution pstb. I am not seeing any correlation between those anomalies and the switching frequency of 500 kHz...

Best,
Edoh




Back to top
 

help.png
View Profile   IP Logged
Ken Kundert
Global Moderator
*****
Offline



Posts: 2349
Silicon Valley
Re: PSS + PSTB for switching TIA
Reply #3 - Aug 11th, 2022, 2:31pm
 
Okay, the probe looks good.  And use the diffstbprobe is good too, the dmcmprobe is inaccurate, never use it.

The loop gain plots are too hard to see.  And you want to avoid log sweeps when looking for peaks and nulls due to harmonics of the clock.  I recommend doing a high resolution linear frequency sweep from 0 to 2.5MHz with, say, 1000 points.  You want to make sure the simulator places frequency points at each of the harmonics.

-Ken

Back to top
 
 
View Profile WWW   IP Logged
eshau
New Member
*
Offline



Posts: 5

Re: PSS + PSTB for switching TIA
Reply #4 - Aug 13th, 2022, 6:45am
 
Hi Ken,

Thank you for following up.

Here is a another simulation with 2.5 kHz resolution.

I am not so sure what to be looking for here, although there does appear to be some correlation between those "disturbances" and the harmonics of my sampling frequency (500 kHz).

If indeed the sampling is interfering with the pstb, how is one to mitigate that?

Also, perhaps it's relevant to mention that I am generating the sampling signal with a 'vsource'.

Thanks again,
Edoh
Back to top
 

loopGainPhase.jpg
View Profile   IP Logged
eshau
New Member
*
Offline



Posts: 5

Re: PSS + PSTB for switching TIA
Reply #5 - Aug 23rd, 2022, 11:56pm
 
Hi Ken,

After further investigation I am quite confident that indeed those artifacts in the PSTB are due charge injection from the clock. I believe that the the PSTB solver sees those voltage spikes (from the PSS simulation) and assumes those spikes are just part of the transfer function.

Thank you very much for you input,

Best,
Edoh
Back to top
 
 
View Profile   IP Logged
Ken Kundert
Global Moderator
*****
Offline



Posts: 2349
Silicon Valley
Re: PSS + PSTB for switching TIA
Reply #6 - Aug 24th, 2022, 11:24pm
 
It is not clock feedthrough.  The periodic small signal analyses all employ two steps.  First, the clock is applied and periodic operating point is computed. From this a linear periodic representation is created.  Second, the small signals are applied and the small signal response is computed.  The clocks are only applied in the first step.  In the second step, only the small-signal stimulus is applied.

Thus, the ripples and peaks at the multiples of the clock signals are not clock feedthrough.  They can be one of two things.  They can be low frequency dynamics that are being mixed up to the harmonics of the clock.  Or they can be cancellations from the various paths the signals take as the get mixed up and down by the clock.

I am not exactly sure how to interpret the Nyquist criteria on clocked circuits, so I don't know what to make of your results.  However, I do know that one sure way to determine the stability of a feedback system is to simply perform a transient analysis and see if the circuit oscillates.  You should force a small time step and stimulate the circuit with a very abrupt step.

-Ken
Back to top
 
 
View Profile WWW   IP Logged
eshau
New Member
*
Offline



Posts: 5

Re: PSS + PSTB for switching TIA
Reply #7 - Aug 24th, 2022, 11:37pm
 
Hi Ken,

Thank you for the detailed response, I will give it further thought. In the meantime step/impulse response are behaving well and improved versions of the circuit also have good phase margin in PSTB (ignoring the spikes at the clock).

Thanks again for the informative troubleshooting.

Best,
Edoh
Back to top
 
 
View Profile   IP Logged
Pages: 1
Send Topic Print
Copyright 2002-2022 Designer’s Guide Consulting, Inc. Designer’s Guide® is a registered trademark of Designer’s Guide Consulting, Inc. All rights reserved. Send comments or questions to editor@designers-guide.org. Consider submitting a paper or model.