The Designer's Guide Community
Forum
Welcome, Guest. Please Login or Register. Please follow the Forum guidelines.
Jan 19th, 2020, 2:42am
Pages: 1
Send Topic Print
Sampling rate for Spectre TDPnoise analysis (Read 7195 times)
Arnd
New Member
*
Offline



Posts: 3

Sampling rate for Spectre TDPnoise analysis
Aug 13th, 2008, 3:21am
 
Hello,
I am trying to perform a TDPnoise analysis on a discrete time system (switched cap. decimation filter) in which the actual output sample rate is higher than the PSS-period (due to the use of interleaved switched capacitor banks). If I understood TD-Pnoise (timedomain Pnoise) correctly spectre automatically considers a situation in which the sample rate at the output of the system is equal to the PSS-beat-period.

In my given case the sample rate at the output of the system is 4*fbeat and Spectre TD-Pnoise would therefore undersample the output resulting in additional noise aliasing. Theses results would physically not correspond to my circuit.
My first question is now if spectre allows to define a sampling rate for the TD-Pnoise analysis other than the PSS-beat-frequency.

If that is not the case from my understanding the only alternative would be to perform at regular PNoise (source) analysis ranging
from DC.. whatever high frequency and perform the noise folding manually on the output noise spectral density spectrum. Would that be
a correct alternative to TD-Pnoise?
I would appreciate any comments on those two questions,
Thank you.
Back to top
 
 
View Profile   IP Logged
pancho_hideboo
Senior Fellow
******
Offline



Posts: 1424
Real Homeless
Re: Sampling rate for Spectre TDPnoise analysis
Reply #1 - Aug 13th, 2008, 9:00am
 
You are misunderstanding tdnoise of Cadence Spectre.

See the following.
http://www.designers-guide.org/Forum/YaBB.pl?num=1207830622
http://www.designers-guide.org/Forum/YaBB.pl?num=1059089369

Arnd wrote on Aug 13th, 2008, 3:21am:
If that is not the case from my understanding the only alternative would be to perform at regular PNoise (source) analysis ranging
from DC.. whatever high frequency and perform the noise folding manually on the output noise spectral density spectrum. Would that be
a correct alternative to TD-Pnoise?

You only consider effective sampling rate. And maxtimestep of PSS analysis is determined by this effective sampling rate.
This is same as actual digitizing sampling oscilloscope.
Back to top
 
« Last Edit: Aug 14th, 2008, 1:15am by Forum Administrator »  
View Profile WWW Top+Secret Top+Secret   IP Logged
Arnd
New Member
*
Offline



Posts: 3

Re: Sampling rate for Spectre TDPnoise analysis
Reply #2 - Aug 14th, 2008, 12:09am
 
Thank you for the reply. I read the postings you had mentioned but it somehow didn't solve my problem.
Quote:
For time-domain noise (noisetype=timedomain), SpectreRF calculates the PSD of the discrete (or sampled) noise at t0+n*T (T=1/PSSfund, 0<=t0<=T, -infinity<n<+infinity)

This corresponds to what I believe the PSD of the sampled output noise of a system would be which provides one output sample every 1/PSSfund. E.g. the switched cap. in K. Kunderts document 'Simulating Switched Capacitor Filters with SpectreRF'
The system I am looking at provides output samples at a rate 1/(4*PSSfund) does that not mean that spectre undersamples the output noise when the PSD is sampled at PSSfund (instead of 4*PSSfund)?
From my understanding the output noise in TDPnoise should be band-limited from 0..4*PSSfund/2
Thank you.
Back to top
 
 
View Profile   IP Logged
pancho_hideboo
Senior Fellow
******
Offline



Posts: 1424
Real Homeless
Re: Sampling rate for Spectre TDPnoise analysis
Reply #3 - Aug 14th, 2008, 12:19am
 
Arnd wrote on Aug 14th, 2008, 12:09am:
From my understanding the output noise in TDPnoise should be band-limited from 0..4*PSSfund/2

No.
Assume four parallel operation of SCF.
Even though four SCF are operated in an interleaved clock, each SCF is operated at fclock.
Contributions from Noise before SCF are correlated between four paths. But contributions from SCF noise itself are not correlated between four paths.
So if you observe serially synthesized output from interleaved four SCF, cyclostationality of noise has still long period of 1/fclock although signal output is generated at short period of 1/(4*fclock).

Also there is no undersampling in tdnoise.

Back to top
 
« Last Edit: Aug 14th, 2008, 1:35am by pancho_hideboo »  
View Profile WWW Top+Secret Top+Secret   IP Logged
Arnd
New Member
*
Offline



Posts: 3

Re: Sampling rate for Spectre TDPnoise analysis
Reply #4 - Aug 14th, 2008, 1:56am
 
Maybe interleaving didn't explain my problem properly. Instead the filter alternates between two banks to provide valid output samples while the input is being sampled at the same time. I attached a simplified schematic picture of a decimate by 2 FIR filter.
From my understanding TDNoise will sample the noise spectrum at Fpssfund instead of Fs_out (refer to the attachment). Do I see things wrong here?
Would be great if I could get this thing sorted out once and for all.
Back to top
 

FIR_dec2.JPG
View Profile   IP Logged
pancho_hideboo
Senior Fellow
******
Offline



Posts: 1424
Real Homeless
Re: Sampling rate for Spectre TDPnoise analysis
Reply #5 - Aug 14th, 2008, 2:27am
 
Arnd wrote on Aug 14th, 2008, 1:56am:
From my understanding TDNoise will sample the noise spectrum at Fpssfund instead of Fs_out (refer to the attachment).

What do you mean ?
Do you sample frequency spectrum ?

You can evaluate PSD(time, freq), where time can be set at far smaller time step than 1/PSSfund. time step can be small at same level of PSS time step.
But still noise is sampled at period=1/PSSfund.

Arnd wrote on Aug 14th, 2008, 1:56am:
Would be great if I could get this thing sorted out once and for all.

What is your question ?

Maybe you are thinking something like direct sampling mixer.
http://www.designers-guide.org/Forum/YaBB.pl?num=1207225157
Back to top
 
 
View Profile WWW Top+Secret Top+Secret   IP Logged
Pages: 1
Send Topic Print
Copyright 2002-2020 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.