Designer’s Guide Community

Ken Kundert

Global Moderator

Offline

Posts: 2386
Silicon Valley

Re: Positive values of phase noise in dBc/Hz??
Reply #15 - Jun 08^th, 2005, 12:33pm

Are you expecting a 3dB difference between L and S_ph (the PM noise), or between the PM and AM noise components?

These results seem completely wrong for a free-running oscillator. I would expect to see the L and S_ph to drop two decades for every decade of offset frequency at low offset frequencies (say less than 1MHZ). I would expect to see the AM noise to be relatively flat and much smaller than the PM noise except at high offset frequencies (say greater than 1MHz). So at low frequencies there would not be a 3dB difference between the two, it would be much greater.

I cannot explain why Artist starts the AM/PM noise plot at 10MHz, it sounds like a bug.

I also cannot explain why there would be notches in the PM noise at low frequencies unless there are resonant circuits in the bias circuits.

Finally, I cannot explain why the PM noise does not follow the traditional -2:1 slope.

All of this sounds like something is going terribly wrong. Have you checked the steady-state behavior as computed by SpectreRF? Is the oscillator really oscillating at the expected amplitude and frequency? If everything else look good, then perhaps you have encountered a bug.

-Ken

Back to top

IP Logged

Nick

Junior Member

Offline

Posts: 21

Re: Positive values of phase noise in dBc/Hz??
Reply #16 - Jun 9^th, 2005, 4:49am

Ken,

To be more clear, Cadence by default was only showing me the AM / PM waveforms only after 10MHz. So, for such big frequency offsets I was expecting to see that AM and PM have the same value and that they are 3dB less than the USB noise. My USB noise is exactly the same as phase noise of pnoise.

Instead, what I get is AM=PM=USB-3dB at the beginning of my sim (whatever frequency it is I begin the sim from), then the notch, and finally again AM=PM = constant value for the rest of the frequ.

My PSS is working. The frequency I get is the one I designed for. Also my transient response shows oscillation.

Also, I went through the examples of Cadence Documentation. Specifically I loaded the schematic oscDiff from the rfExamples of Cadence. I run AM and PM and the waveforms I got look like mine. So, I guess it is not just my design. You could check that out too. If this is a bug, what do I do to get the PM working?

Nick

Back to top

Nick

IP Logged

Ken Kundert

Global Moderator

Offline

Posts: 2386
Silicon Valley

Re: Positive values of phase noise in dBc/Hz??
Reply #17 - Jun 9^th, 2005, 6:48am

At the moment I do not have access to the Cadence software, so I cannot take a look at oscDiff. Your description of the results seems quite inconsistent with my expectation of what the results should be, and I can think of no obvious user errors that could be causing them, so I would contact Cadence if I were you and ask them for help.

-Ken

Back to top

IP Logged

Nick Junior Member Offline Posts: 21	Re: Positive values of phase noise in dBc/Hz?? Reply #18 - Jun 9^th, 2005, 9:57am Ken, Thank you for all the help. It is much appreciated. I will seek further advice from Cadence itself and I will get back to you, just for the completeness of the matter. Many thanks again, Nick
Back to top	Nick IP Logged

Andrew Beckett Senior Fellow Offline Life, don't talk to me about Life... Posts: 1742 Bracknell, UK	Re: Positive values of phase noise in dBc/Hz?? Reply #19 - Jun 9^th, 2005, 2:37pm Nick, I just looked into this - and ran through the oscDiff example. I don't see the
Back to top	IP Logged

Andrew Beckett

Senior Fellow

Offline

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

Posts: 1742
Bracknell, UK

Re: Positive values of phase noise in dBc/Hz??
Reply #20 - Jun 9^th, 2005, 2:50pm

Whoops - hit the wrong button. Sorry. I'll start again. I was messing around trying to get an image included, and made a mistake...

I just tried this (oscDiff) using both the latest subversion of IC5141 (5.10.41.500.1.18) together with spectre from MMSIM60 USR1 - and then I also tried using the Base CD version of IC5141 - with spectre from the same version.

I tried with and without highorder=yes in case that made any difference (it didn't to the phase noise results).

I got consistent results as the picture shows:

The blue line shows the traditional phase noise plot - and the yellow line (which is almost identical except at higher frequencies) as you can see. The am noise (the red line) is a fair way down compared with the pm/phase noise - exactly as you'd expect. Note, I'm plotting the am and pm noise in dBc for comparison.

So, what version were you using - the above looks reasonable to me, and not like what you described.

Regards,

Andrew.

Back to top

IP Logged

Ken Kundert Global Moderator Offline Posts: 2386 Silicon Valley	Re: Positive values of phase noise in dBc/Hz?? Reply #21 - Jun 9^th, 2005, 4:02pm Andrew, Those results also look reasonable to me. The rise in AM noise at low frequencies is probably a simulation artifact due to limited ability to resolve very small amounts of AM noise in the face of large amounts of PM noise. -Ken
Back to top	IP Logged

Andrew Beckett Senior Fellow Offline Life, don't talk to me about Life... Posts: 1742 Bracknell, UK	Re: Positive values of phase noise in dBc/Hz?? Reply #22 - Jun 9^th, 2005, 9:38pm BTW, the subversion of IC5141 came out wrongly as I forgot to hit the button to disable smileys. It should have been 5.10.41.500.1.18. Regards, Andrew.
Back to top	IP Logged

Nick Junior Member Offline Posts: 21	Re: Positive values of phase noise in dBc/Hz?? Reply #23 - Jun 10^th, 2005, 2:36am Hi, thanks for going through this. The version I am currently using is this: CDS: icfb.exe version 5.0.0 04/30/2004 sub-version 5.0.33_USR2.34.8 Do you think that the problem I am facing is related to the fact I am using an earlier version? Thank you very much, Nick
Back to top	Nick IP Logged

Andrew Beckett Senior Fellow Offline Life, don't talk to me about Life... Posts: 1742 Bracknell, UK	Re: Positive values of phase noise in dBc/Hz?? Reply #24 - Jun 12^th, 2005, 12:26pm Nick, I'll try the same example tomorrow using IC5033 USR2 and let you know what I get. Regards, Andrew.
Back to top	IP Logged

Andrew Beckett

Senior Fellow

Offline

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

Posts: 1742
Bracknell, UK

Re: Positive values of phase noise in dBc/Hz??
Reply #25 - Jun 13^th, 2005, 1:24am

Nick,

The direct plot appears to be quite broken in IC5033 USR2. I see the problem you're having. In fact if you pan left, the axis does continue, but the results are clearly wrong...

I found this PCR:

PCR: 699011
Title: AM/PM noise has incorrect X-axis

which from the discussion in the PCR looks as if it could well be the problem. This was fixed in IC5033 USR3.

Having just tried IC5033 USR3, the problem is gone there and the results I get look the same as those I posted earlier in this thread.

So in summary, this appears to be a version related problem.

Best Regards,

Andrew.

Back to top

IP Logged

Nick Junior Member Offline Posts: 21	Re: Positive values of phase noise in dBc/Hz?? Reply #26 - Jun 14^th, 2005, 1:51am Thank you all for your help. I have much appreciated it. Best Regards, Nick
Back to top	Nick IP Logged

chlai Junior Member Offline Save the earth~ Posts: 13 Taipei, Taiwan	Re: Positive values of phase noise in dBc/Hz?? Reply #27 - Jul 6^th, 2005, 2:18am Hello all experts, In the phase noise/AM/PM noise shown by Mr. Beckett, how do you explain the PM noise larger than 0dBc/Hz ? with Best regards, Daniel Lai
Back to top	IP Logged

Ken Kundert Global Moderator Offline Posts: 2386 Silicon Valley	Re: Positive values of phase noise in dBc/Hz?? Reply #28 - Jul 6^th, 2005, 8:49am Read this thread from the beginning. -Ken
Back to top	IP Logged

chlai

Junior Member

Offline

Save the earth~

Posts: 13
Taipei, Taiwan

Re: Positive values of phase noise in dBc/Hz??
Reply #29 - Jul 6^th, 2005, 6:35pm

Hi Ken,
Thanks for your tip.

For PM noise, the unit should be radian^2/Hz, so my problem is why PM noise expressed in the figure is dBc/Hz? (Does the PM noise transfer to PSD of voltage as phase noise or just a unit mistake?) Besides, I was confused if the PM noise could be shown in the same figure with AM noise(dBc/Hz) or phase noise(dBc/Hz).

On the other hand, if the PM noise is radian^2/Hz, how can we explain the PM noise over 0dB-radian/Hz at the near offset?

Thank you very much for your interest.

Daniel Lai

Back to top

« Last Edit: Jul 6^th, 2005, 8:03pm by chlai »

IP Logged