The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> RF Simulators >> hidden state in Fixed-frequency oscillators
https://designers-guide.org/forum/YaBB.pl?num=1380726354

Message started by sashqazi on Oct 2^nd, 2013, 8:05am

Title: hidden state in Fixed-frequency oscillators
Post by sashqazi on Oct 2^nd, 2013, 8:05am

Hi,
I am trying to study the effect of clock jitter on passive mixer . When i used the Fixed-frequency oscillators with jitter (available on this site inverilog-AMS section), SpectreRF complains about the hidden state in the verilogA model. I think this is becuase of the "dist_uniform" function used in the model. Can anyone suggest an alternate (hidden states free) way to model clock jitter.

Title: Re: hidden state in Fixed-frequency oscillators
Post by Ken Kundert on Oct 2^nd, 2013, 9:17pm

The issue is more fundamental that a hiden state problem. You cannot have jitter in a PSS analysis because there is no periodic solution. Instead, you leave the noise/jitter out of the PSS analysis and only apply it during the PNoise analysis.

-Ken

Title: Re: hidden state in Fixed-frequency oscillators
Post by sashqazi on Oct 3^rd, 2013, 5:23am

Thank you for your reply. It means that i cannot analyze the effect of clock jitter on my mixer in spectreRF? However i can estimate the amount of jitter produced by an oscillator using jitter analysis in pnoise. Am i right?

Title: Re: hidden state in Fixed-frequency oscillators
Post by Ken Kundert on Oct 3^rd, 2013, 11:16pm

Both PSS and PNoise are part of SpectreRF. But yes, you would PNoise analysis to do the jitter analysis.

-Ken

Title: Re: hidden state in Fixed-frequency oscillators
Post by sashqazi on Oct 4^th, 2013, 2:12am

Thank you again. yes I understand that both PSS and PNoise are part of SpectreRF and PSS is a pre-requisite for PNoise. Probably i have not phrased my problem clearly earlier. Below is what we have discussed so far..

1) "Fixed-frequency oscillators(with jitter)" model cannot serve as a fundamental tone for PSS analysis (which is also a driving signal for my mixer simulation on LO port).
2) My idea is to drive the LO port of my mixer(passive) with jittered clock, and see its effect at the mixer output in terms of NF for example, which for the obvious reasons cannot be done directly in SpectreRF.
3) You suggested then that i should keep noise/jitter out of the PSS analysis and only apply it during the PNoise analysis. I dont know how it is possible in my scenario.

What i understand is that jitter analysis option provided in PNoise analysis is for estimating the effect of jitter present at some point in the circuit and not for modeling the certain amount of jitter at some point in the circuit??

So?? i cannot use your oscillator model "Fixed-frequency oscillators(with jitter)" in SpectreRF for purpose 2) discussed above and i have to rely on transient simulation for this purpose?? Is this correct?

Sorry if there is a repetition in my queries. I wanted to very clear this time.

Title: Re: hidden state in Fixed-frequency oscillators
Post by Ken Kundert on Oct 4^th, 2013, 10:25am

Quote:

1) "Fixed-frequency oscillators(with jitter)" model cannot serve as a fundamental tone for PSS analysis (which is also a driving signal for my mixer simulation on LO port).

Correct.

Quote:

2) My idea is to drive the LO port of my mixer(passive) with jittered clock, and see its effect at the mixer output in terms of NF for example, which for the obvious reasons cannot be done directly in SpectreRF.

Not in SpectreRF, but it can be done with transient analysis.

Quote:

3) You suggested then that i should keep noise/jitter out of the PSS analysis and only apply it during the PNoise analysis. I don't know how it is possible in my scenario.

Correct. The PNoise analysis does this naturally. During PSS all small-signal noise sources are ignored. They only become active during noise analyses. So any noise that is generated by the circuit that provides the LO will naturally be included. If you cannot include the circuitry that generates the LO, you will need to model it and the noise it generates. You can do that with the noise modeling capabilities of Verilog-AMS (white_noise(), flicker_noise() and noise_table()). You may need to include noise modulation in your model, which you can do using the techniques employed in the AM, PM and FM modulators shown on the Verilog-AMS page (http://www.designers-guide.org/VerilogAMS/functional-blocks/modulators/modulators.va).

Quote:

What i understand is that jitter analysis option provided in PNoise analysis is for estimating the effect of jitter present at some point in the circuit and not for modeling the certain amount of jitter at some point in the circuit?

Correct.

Quote:

So, I cannot use your oscillator model "Fixed-frequency oscillators(with jitter)" in SpectreRF for purpose 2) discussed above and i have to rely on transient simulation for this purpose? Is this correct?

Correct.

Title: Re: hidden state in Fixed-frequency oscillators
Post by sashqazi on Oct 5^th, 2013, 5:31am

I got your point. Thanks a lot.

/Qazi