Home > Community > Forums > RF Design > half harmonic frequency in PSS analysis

## Email

Recipients email * (separate multiple addresses with commas)

Message *

 Send yourself a copy

## Subscribe

Intro copy of the newsletter section here, some intro copy of the newsletter. Instruction of how to subscribe to this newsletter.

First Name *

Last Name *

Email *

Company / Institution *

 Send Yourself A Copy

# half harmonic frequency in PSS analysis

Last post Thu, Jun 9 2011 3:34 AM by Andrew Beckett. 8 replies.
 Started by Andy Liu 30 Apr 2009 08:31 AM. Topic has 8 replies and 6458 views
• #### Thu, Apr 30 2009 8:31 AM

• Andy Liu
• Joined on Thu, Apr 30 2009
• Posts 4
• Points 110
half harmonic frequency in PSS analysis
 I am designing an LC based voltage control oscillator. I want to simulate phase noise of my oscillator, but, I ALWAYS got a very bad phase noise performance (for example, -60 dBc/Hz @ 1 MHz offset, compared to my expectation -110 dBc/Hz). The reason is that in the periodic steady state analysis (PSS), the fundamental frequency is simulated wrong. I estimate the fundamental frequency is 5 GHz from the transient waveform, whilst the fundamental frequency given by PSS simulation is 2.5 GHz. Has anyone encountered or does anyone have any knowledge of this problem?  I figure out a method to get around this half harmonic frequency problem. In the pnoise analysis, people usually choose “sweep type” as “relative”, and “relative harmonic” as 1. I wonder if I can choose “relative harmonic” as 2 to compensate the half harmonic in PSS simulation.   Thank you.
• Post Points: 20
• #### Thu, Apr 30 2009 10:26 AM

Re: half harmonic frequency in PSS analysis
 It's best not to post in two separate forums, otherwise nobody knows which to respond to...Anyway, usually the reason for this is because you have a subharmonic in your circuit. Is there a divider in there, even though you're actually setting the oscillator output (for PSS and PNoise) to be the undivided output? If so, PSS still needs a periodic solution for the whole circuit, and consequently must find the subharmonics.The solution is indeed to use a relative harmonic 2 in the Pnoise analysis.Andrew.
• Post Points: 20
• #### Mon, Jun 6 2011 8:44 AM

• aamar
• Joined on Wed, Sep 29 2010
• Posts 3
• Points 60
Re: half harmonic frequency in PSS analysis
 Hallo,I actually have a similar problem without having subharmonics in the VCO, I am only simulating the LC VCO. The estimated frequency from the PSS analysis is the same like what I expect and the same which I get from the dft of the tstab transient signal. However when the PNOISE runs I get a lower frequency with an offset of arround 400MHz just as an example. The VCO is including an AGC circuit and the inductor is modeled using a nport device. What could be the cause actually?Thanks and best regards, aamar
Filed under: ,
• Post Points: 20
• #### Tue, Jun 7 2011 3:11 PM

Re: half harmonic frequency in PSS analysis
 If the PSS is reporting the right frequency, I see no reason why (or how) the pnoise analysis can be using a different frequency. Perhaps you can post the output log you get, as well as the analysis statements from your input netlist (input.scs)?Regards,Andrew.
• Post Points: 20
• #### Wed, Jun 8 2011 7:40 AM

• aamar
• Joined on Wed, Sep 29 2010
• Posts 3
• Points 60
Re: half harmonic frequency in PSS analysis
• Post Points: 20
• #### Wed, Jun 8 2011 9:30 AM

Re: half harmonic frequency in PSS analysis
 Look at the end of the harmonic balance simulation - you'll see that the frequency matches that used in the pnoise. The whole point of the frequency printed out at the beginning of the PSS is that it is an estimated frequency. It's used to help the harmonic balance start with a reasonably close estimate of the actual settled frequency - the harmonic balance iterations will take care of converging on the final, settled, steady state frequency - which is  6.4471e+09 as shown in the 47th (i.e. last) iteration of the harmonic balance. Since it seems to be taking a while to converge, I'm wondering whether you have enough harmonics, or whether you need a bit more tstab time to help things along. But either way, there is no consistency between the PSS and pnoise analyses.Andrew.
• Post Points: 20
• #### Thu, Jun 9 2011 12:50 AM

• aamar
• Joined on Wed, Sep 29 2010
• Posts 3
• Points 60
Re: half harmonic frequency in PSS analysis
 Dear Andrew,  Yes that is the problem, no consistency between the PSS and the PNOISE analyses.The PSS results (i.e. tstab output signal and its estimated frequency) are the same like what I get from the transient analysis. But what the HB reaches is something else. What is the problem then? I am using 10 sidebands and even using 20 it gets the same results but through less iterations. Also the tstab time in my case is long enough, i don't think it would be the problem. What shall I do then? Best regards, aamar
• Post Points: 20
• #### Thu, Jun 9 2011 3:34 AM

Re: half harmonic frequency in PSS analysis
 aamar,The PSS results and PNOISE results ARE consistent. The PSS is finding the settled steady-state behaviour of the oscillator. Quite likely your transient and the initial transient used for tstab are not long enough to reach the settled steady-state behaviour. If you ran a much longer transient, it wouldn't surprise me if you found that it eventually reached the frequency that the PSS is solving to. Bear in mind as well that you may not have tight enough tolerances in your simulations - try tightening reltol to 1e-4 or maybe even 1e-5.Andrew.
• Post Points: 5
• #### Thu, Jun 9 2011 3:34 AM

Re: half harmonic frequency in PSS analysis
 Note by this I'm not suggesting that having a longer tstab will fix it; I'm just saying that you may not be comparing like with like.Andrew.
• Post Points: 5