The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> Circuit Simulators >> ELDO .noisetran fmin/fmax
https://designers-guide.org/forum/YaBB.pl?num=1129917993

Message started by Visjnoe on Oct 21^st, 2005, 11:06am

Title: ELDO .noisetran fmin/fmax
Post by Visjnoe on Oct 21^st, 2005, 11:06am

Hello

I have a rather practical question about the .noisetran analysis offered by ELDO, namely concerning the fmin and fmax parameters.

I believe they correspond with the frequency band of noise sources you want to take into account.

So, when simulating (transient noise) a SC filter, fmax should be set to e.g. 500MHz, right?My confusion now lies in the fact that for a (SC) filter, I'm wondering if fmax should not be set to the cut-off frequency of the filter?

Furthermore, anyone has had some (good) experience with the results predicted by the ELDO .noisetran analysis?

Thanks a lot!

Peter

Title: Re: ELDO .noisetran fmin/fmax
Post by Paul on Oct 21^st, 2005, 2:55pm

Hi Peter,

personally, when using ELDO, I tend to use the AC noise simulation and get the total noise through calculations. I find the noisetran result difficult to exploit.

In an SC circuit, I believe (at least) part of the noise contributions at frequencies above the filter cut-off may be folded back due to the sampling process. For this reason, I would set fmax to something (one decade?) above the amplifier cut-off frequency.

Paul

PS: there's a whole chapter on transient noise simulation in the Eldo user's manual.

Title: Re: ELDO .noisetran fmin/fmax
Post by Visjnoe on Oct 23^rd, 2005, 8:00am

Hi Paul,

thanks for your answer.

I'm aware of the chapter on .noisetran in the Eldo manual, but I find it a bit confusing at some points.

A clear disadvantage of .noisetran is that you have to simulate for a long time to make sure you take into account the 1/f noise. Typically, this shows by an increasing RMS noise value with simulation time.

Kind Regards

Peter

Title: Re: ELDO .noisetran fmin/fmax
Post by Andrew Beckett on Oct 23^rd, 2005, 11:30am

Do you have access to SpectreRF? That would be a much better bet for simulating the noise in an SC filter. Transient noise analyses (and spectre can do this as well, since MMSIM60) suffer from the problems you've described if you want to make any kind of quantative measurement - you need to set the maximum frequency to something sensible (which results in rather short timesteps) and you need to simulate for a long time to get the averaging and to include flicker noise properly. So it's best for qualitative measurements (in my opinion) (i.e. did the change I made in my circuit improve things or make things worse) - and it's best to use something else (like pnoise, or tdnoise) if you can (i.e. if the circuit is periodic in nature, and the noise can be treated as small signal).

Regards,

Andrew.