The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Analog Verification >> Analog Performance Verification >> transient noise vs pnoise (sources and timedomain noise type)
https://designers-guide.org/forum/YaBB.pl?num=1306572324

Message started by matteo.tonelli on May 28^th, 2011, 1:45am

Title: transient noise vs pnoise (sources and timedomain noise type)
Post by matteo.tonelli on May 28^th, 2011, 1:45am

Hi,

i would compare the noise performance resulting from transient noise (spectre) and pnoise (spectreRF) of a switched cap. circuit (i.e. an integrator).

Despite the exhaustive informations provided in "Simulating switched-capacitors filters with SpectreRF" (http://www.designers-guide.org/Analysis/sc-filters.pdf), some clarifications are needed.

1) the maximum noise freq. ("noisefmax") in transient noise represents the stop frequency in pnoise analysis?
2) is the minimum noise freq. ("noisefmin") in transient noise related to the transient simulation stop time?
Assuming tstop the transient stop time the minimum frequency that can be observed is 1/tstop hence none noise contributions below this limit can be observed. Is this consideration correct?
3) Assuming the minimum noise frequency in transient noise analysis to be fmin. Does fmin set the start frequency in pnoise analysis?

Simulating a simple ideal sample and hold sampling at fck the expected noise power is kT/C where C represents the holding capacitance.

What's the difference between pnoise with "source" noise type and pnoise with "timedomain" noise type?

In order to match the kT/C expected noise power, the output PSD resulting from "source" noise type must be integrated over a wide bandwidth (integration bandwidth >> fck). Otherwise, in order to obtain kT/C, the PSD resulting from "timedomain" noise type must be integrated starting from fstart to fck/2 (and the accuracy depends on the number of sidebands considered).

Is this correct? Is this the right interpretation of the spectre and spectreRF results?

Many thanks, Matteo

Title: Re: transient noise vs pnoise (sources and timedomain noise type)
Post by weber8722 on Nov 18^th, 2011, 12:10am

Hi,

matteo.tonelli wrote on May 28^th, 2011, 1:45am:

Hi,

1) the maximum noise freq. ("noisefmax") in transient noise represents the stop frequency in pnoise analysis?

2) is the minimum noise freq. ("noisefmin") in transient noise related to the transient simulation stop time?
Assuming tstop the transient stop time the minimum frequency that can be observed is 1/tstop hence none noise contributions below this limit can be observed. Is this consideration correct?

3) Assuming the minimum noise frequency in transient noise analysis to be fmin. Does fmin set the start frequency in pnoise analysis?

Many thanks, Matteo

1=NO. You can run e.g. pnoise for phase noise in MHz range for GHz VCO, but tran noise needs something like fmax>3fosc
2=Good to do so, but if no fmin is set flicker nosie will be completely ignored.
3=NO, not automatically of course.

Hope it helps a bit.

Title: Re: transient noise vs pnoise (sources and timedomain noise type)
Post by raja.cedt on Nov 20^th, 2011, 3:35am

hello,
Please read the following document. It has lot of things thing about Transient noise.

Application Notes on Direct Time-Domain
Noise Analysis using Virtuoso Spectre

Thanks,
raj.

Title: Re: transient noise vs pnoise (sources and timedomain noise type)
Post by uzzy on Mar 5^th, 2015, 5:14pm

@Raja: Could you tell me where to find the application note(Application Notes on Direct Time-Domain Noise Analysis using Virtuoso Spectre) you mentioned ?

Title: Re: transient noise vs pnoise (sources and timedomain noise type)
Post by uzzy on Mar 5^th, 2015, 5:15pm

I looked it up on the cadence website and it is not available. They may changed the name or something.

Title: Re: transient noise vs pnoise (sources and timedomain noise type)
Post by Frank Wiedmann on Mar 6^th, 2015, 12:41am

I would suggest http://support.cadence.com/wps/mypoc/cos?uri=deeplinkmin:DocumentViewer;src=wp;q=ApplicationNotes/Custom_IC_Design/Trnoi_AppNote.pdf.

Title: Re: transient noise vs pnoise (sources and timedomain noise type)
Post by weber8722 on Apr 2^nd, 2015, 3:30am

Hi,

I think it is important to recognize that pnoise is covering NOT all kind of noise effects, tran noise does - if applied correctly! :'(

One example is a free-running oscillator used as LO for a mixer. The LO phase noise will not degrade interferer behavior in SpectreRF simulations, but would be shown correcly in tran noise. This effect is well-known as reciprocal mixing, and most designers use xls to take it into account.

Also bit-error rate can hardly be adressed with pnoise, because if the bit toggles due to noise this changes the op-point too much and violates the assumption that noise changes voltages and currents only a bit.

Bye Stephan