Designer’s Guide Community

Emre

Junior Member

Offline

Posts: 10

PSS doesn't converge or not accurate for pnoise
Mar 29^th, 2006, 1:21am

I designed a frequncy detector (frequency-to-voltage converter), but it has ripples on the output voltage and after the filter this result in some fluctiations at the input of the VCO, so the VCO output frequency doesn't settle at one frequency, but have perodic ripples on it. When I run pss on the system, pss doesn't converge.

I changed several parameters that are written in SpectreRF documentation and at the end find out that increasing steadyratio upto 10 help it to converge. �However; this time simulator gives a warning that pss analysis may be insufficiently accurate for pnoise analysis and The Floquet eigenspace computed by PSS appears to be inaccurate, so PNOISE can be inaccurate and asks me to run with smallet reltol.

Then I modeled the VCO in voltage domain with jitter as it's described in Kenn's paper and the codes in this webpage. After fixing the hidden state problem, when I run PSS and PNOISE again, I realized the same problem and after increasing the steadyratio, I couldn't get any phase noise, even if there is quite a lot jitter.

Two problems can be because of the same reason and if it's what is the best way to handle these kind of situations in SpectreRF.

Emre

Back to top

IP Logged

Emre

Junior Member

Offline

Posts: 10

Re: PSS doesn't converge or not accurate for pnois
Reply #1 - Mar 29^th, 2006, 5:50am

I found out the problem in the second paragraph. The model was not supposed to work in pnoise simulations to give related phase noise.

But the first problem still goes on. Even if I filter the frequency detector output, there stays some ripples in the control voltage of the VCO as a uV range sinus wave at the carrier frequency. This results in sinus wave ripples at the output frequency of the VCO. As my VCO is supposed to work around 10GHz, this sinus has amplitude of 100s of kHz.

As I said in the previous thread, changing steadyratio in pss options helps me, but now I found out that pnoise simulation after such a relaxed pss using moderate gives much more noise than normal (it begins from +75dBc/Hz instead of -40dBc/Hz)

So what is the trade-off between accuracy and convergence. Maybe it's not a good idea to increase steadtratio that much. How can I look at the phase noise perormance of my system, while these fluctuations exist?

Emre

Back to top

IP Logged

Andrew Beckett

Senior Fellow

Offline

Life, don't talk to
me about Life...

Posts: 1742
Bracknell, UK

Re: PSS doesn't converge or not accurate for pnois
Reply #2 - Mar 29^th, 2006, 1:11pm

A steadyratio of above 1 is not a good idea.

It sounds as if something is not right - I'm presuming from your description that you have two independent frequencies in your circuit - the primary oscillation, and a secondary oscillation at a different (non-harmonically related) frequency? I assume that's not what you wanted...

Andrew.

Back to top

IP Logged

Emre

Junior Member

Offline

Posts: 10

Re: PSS doesn't converge or not accurate for pnois
Reply #3 - Apr 3^rd, 2006, 8:47am

Thank you for your response Andrew...

Yes, you are right because of the frequency detector, there is a small amplitude oscillation on top of the control voltage of VCO at the same oscillation frequency as the VCO. That's not what I would like to have in an ideal world. However; because of the nature of frequency detector, I have this secondary oscillation on top of the first one.

I still would like to see the phase noise performance of my system and that's why I need to run pss and pnoise. There can always be ripples in the control voltage of the VCO, so I guess I should still be able to simulate the system somehow.

So the question in general is, how to make pss converge but still accurate enough for a pnoise analysis, in unideal conditions like having a very small amplitude oscillation in the control voltage.

Thanks.

Emre

Back to top

IP Logged