Stats

  • Locked Locked
  • Replies 8
  • Subscribers 63
  • Views 16395
  • Members are here 0
This discussion has been locked.
You can no longer post new replies to this discussion. If you have a question you can start a new discussion

pnoise simulation setup questions (again?)

eeask

Hello, there

I'm simulating the jitter performance of a driven ckt (inverter chain like, e.g.). I'm confused by the simulation setup and like to ask a couple of questions regarding the setup parameters.

 1.  relharmnum

First I set sweeptype=relative and relharmnum=1 and run the simulation. Then I change  relharmnum=0 and compare the output results from 2 simulatoins. It looks like that the difference is very small (smaller than 0.01fs). Per the Q&A from Cadence's support website, if we set  relharmnum=1,we're looking at the frequency

 from f0 + 1Xfstart to f0 + 1Xfstop (f0 is the fundamental frequency and fstart/fstop the sweep frequency range).  1 is the relharmnum.

 I think this is the frequency range exactly of my interest. But I wonder why setting relharmnum=0 didn't really disable the sweep frequency range setting (fstart/fstop), if we just follow the equation as above. The small difference of simulation results show it.

 

2.  fstop

I already explored the previous discussion regarding how to set fstop to simulate a driven ckt. The suggestion is to set it to 1/2 of f0 (fundamental tone frequency). It's consistent with Nyquist sampling theorem. But I like to know if it's in conflict with up/down conversion of noise whose frequency can be N times of f0 (N set by the maxsideband). 

Thanks for clearing my questions and any comments welcomed!

 

 

 

  • I think you're just seeing the aliasing caused by the ideal sampler introduced at the output when you use pmjitter or tdnoise modes.

    Consequently it only really makes sense to sweep up to f0/2 - you'll get noise from all sidebands folded into this band. It's not in conflict with the up/down conversion of noise in the circuit itself.

    Regards,

    Andrew.

  • Hi, Andrew

    Thanks for your quick reply. 

    Also I like to know if my understanding to the "relharmnum" paramerter is correct?

     

    Best!

  • Actually, if you set relharmnum to k and the start and stop frequencies of the sweep to fstart and fstop, and your PSS fundamental is f0, then the pnoise sweep will be from:

    k*f0+fstart to k*f0+fstop

    So your "1*" were in the wrong place in your equation. Essentially the sweep is around the kth harmonic of the PSS fundamental. 

    Regards,

    Andrew.

  •  But k=0 will cause the sweep frequency from fstart to fstop (eg. 10 to f0/2). It's nowhere close to any harmonics.

    Just wonder why simulation results don't show much difference between k=0 and k=1.

  • An ideal sampler will alias the spectrum around all harmonics (including the zeroth) of the PSS fundamental. You'd expect to see the spectrum from 0 to f0/2 and then flipped from f0/2 to f0, then these two repeated between f0 and 2*f0, and so on.

    Regards,

    Andrew.

  • Hi, Andrew

    Thanks again for your reply.

    Based on what you mentioned regarding an ideal sampler introduced at the output in simulation, I have to ask another question. Let's take a simple example here:

    fundamental freq=f0, fstart=10, fstop=f0/2,

    the sampling freq of the ideal sampler=f0

    This way won't cause any aliasing for the sampler at the output. (For the time being we just ignore any other harmonics.)

    But If I'm interested the frequency at f0+f0/2+fx (fx is a freq that just makes the sampling violate Nyquist critera), the sampling at the output will NOT produce accurate result due to the aliasing.

    Please correct me if my understanding is not right.

    Best!

     

     

  • The noise which would have appeared at f0+f0/2+fx should appear at f0/2-fx

    Andrew.

  • Hi, Andrew

    I’m confused by your answer.

    In an oscillator if we just ignore harmonics of fundamental tone for simplicity, the system can be modeled by thermal/flicker noise mixing with fundamental tone (refer to SpectreRF user guide). The output at the mixer is still wideband noise. Then the wideband noise will be sampled by the ideal sampler in pss/pnoise simulation. If the wideband noise is not band-limited in advance, the aliasing will occur. I guess that there’s something I didn’t get right. Could you clarify how the noise will be calculated in pss/pnoise simulation? Thanks a lot!

    Best

Community Guidelines

The Cadence Design Communities support Cadence users and technologists interacting to exchange ideas, news, technical information, and best practices to solve problems and get the most from Cadence technology. The community is open to everyone, and to provide the most value, we require participants to follow our Community Guidelines that facilitate a quality exchange of ideas and information. By accessing, contributing, using or downloading any materials from the site, you agree to be bound by the full Community Guidelines.