The Designer's Guide Community
Forum
Welcome, Guest. Please Login or Register. Please follow the Forum guidelines.
Mar 28th, 2024, 6:00am
Pages: 1
Send Topic Print
LC VCO and MOS varactors (Read 7699 times)
City
Junior Member
**
Offline



Posts: 19

LC VCO and MOS varactors
Jun 04th, 2004, 5:12am
 
Hi,

I've simulated a classical LC VCO topology with "grounded" cross-coupled NMOSTs and PMOSTs (no tail current source).
Using a linear reactive network, the phase noise contribution is dominated by the white noise of the channel charges which is normal. I replaced the (lossy)linear capacitors with NMOST varactors in inversion mode and found out that the associated tuning range was 3Ghz to 5Ghz for a 4Ghz center frequency. This high Kvco gain is associated with high AM-FM conversion as described by Abidi in a JSSC paper. What I observed is that I had at 1Mhz a slope of -30dBc/Hz instead of -20dBc/Hz so clearly flicker noise is dominating. However using SPECTRE noise summary property, I also found out that the dominating source of flicker noise was an order of magnitude higher for my PMOSTs than the NMOSTs for a similar size. (flicker noise parameter in Bsim looked correct is slightly higher for PMOST in my case) Did anybody observed this effect before ?


City
Back to top
 
 
View Profile   IP Logged
emad
Junior Member
**
Offline



Posts: 30
Cairo, Egypt
Re: LC VCO and MOS varactors
Reply #1 - Aug 29th, 2004, 5:42am
 
City,

The phenomena you are seeing is NOT AM-to-FM noise coversion. The paper by Prof. Abidi and myself illustrates that this effect (AM-to-FM) is due to flicker noise from the current source. Since you don't have one, the whole analysis is that paper is of no use in your case. Let me illustrate more, switching devices cannot upconvert flicker noise to the oscillation frequency (please read Darabi's paper on mixers to see what I mean). Flicker noise shows at the oscillation frequency through frequency modulation rather than phase modulation. Interestingly, for an RF designer, phase and frequency noise are indistinguishable.
Having that said, the amount of flicker noise in the PMOS and NMOS in your circuit has nothing to do with their relative contribution in the noise analysis.

The frequency noise modulation index is dependent to a great extent on the common mode point capacitance for example. (This was illustrated by Darabi for Mixers and hinted to by Rael for Oscillators). Therefore, if we assume an ideal current source, smaller switches will have more flicker noise but will have a much lower conversion gain. So, depending on the actual values you [u]might[/u] end up better off with smaller devices.
The analysis is far from straightforward, that's why there has been no physically-based theory for flicker noise in oscillators).

Without sounding commercial, there is a new book on oscillators coming soon in the Designer's guide series. It is co-authored by myself, Jacob Rael, and Asad Abidi. The publisher is Springer Verlag. The analysis of flicker noise is detailed there.
Back to top
 
« Last Edit: Aug 30th, 2004, 9:42pm by emad »  
View Profile   IP Logged
Frank Wiedmann
Community Fellow
*****
Offline



Posts: 677
Munich, Germany
Re: LC VCO and MOS varactors
Reply #2 - Aug 30th, 2004, 1:37am
 
There is a theory by Alper Demir regarding the effect of colored noise sources on oscillators, see http://home.ku.edu.tr/~aldemir/pubs/tcas002.pdf and the related papers available under http://home.ku.edu.tr/~aldemir/publications.html. However, the treatment is rather mathematical in nature and will probably offer only limited insight to the average designer.
Back to top
 
 
View Profile WWW   IP Logged
emad
Junior Member
**
Offline



Posts: 30
Cairo, Egypt
Re: LC VCO and MOS varactors
Reply #3 - Aug 30th, 2004, 9:27pm
 
Indeed. Demir's work on colored noise is a wonderful piece of work. It is VERY general. Just like his work on thermal noise where he does not assume cyclostationarity. In fact, there is a nice paper by Vanassche and Gielen that shows that Demir's work stands the test of QVCO.

His approach is different. His work targets the CAD world I think. It is a rigorous mathematical treatment that results in a simulator at the end. Very little design insight can be gained from his analysis. He hinted in his thermal noise paper that they built a simulator based on his equation set.

For flicker noise, he starts by redefining the flicker noise spectrum to have a power of 0 dBc at zero frequency to overcome the traditional problem "could flicker noise have infinite power at DC?"

His analyis is formidable except that it is not meant for designers. It is meant for CAD I believe.
Back to top
 
 
View Profile   IP Logged
yiping
New Member
*
Offline



Posts: 4

Re: LC VCO and MOS varactors
Reply #4 - Oct 31st, 2004, 11:31pm
 
This is for emad about no 1/f noise from SW transfer to output phase noise if no current source in City's case. Since city use MOS at inversion region as varactor wich have extreme high Kvco(not good of course), 1/f noise at gate of SW will sample at 2f will appear at VCO output at DC and 2f, although 1/f at DC not introduce Phase noise directly like your paper said, but this low freq variation definitly moudlate city's high Kvco varactor,so will produce  his high 1/f3 noise corner.
By the way, the explaination in your paper about where 1/f3 come from at VCO output if Kvco=0, kindly not convince me deeply, is your new book clearly mention about that? I will do sim to verify my explaination later.
Back to top
 
 
View Profile   IP Logged
emad
Junior Member
**
Offline



Posts: 30
Cairo, Egypt
Re: LC VCO and MOS varactors
Reply #5 - Nov 26th, 2004, 10:29am
 
YiPing,

Exactly like you say, only noise at amplitude noise at DC will create this AM-to-FM noise conversion due to the varactor. This is what we said in that paper. We focused the analysis on the case where the current source is present as the prominent source of amplitude noise but noise from the switches around zero frequency coming from the switches is downconverted to DC as per the mixer theory by Darabi and Abidi. Therefore it will create frequency noise in the presence of a varactor. This effect is so minor though and other mechanisms are in play for switches flicker noise that are more dominant.

The physical analysis in my paper is also valid for this case (just like you stated it) but the equations need to be vastly modified.

flicker noise of the switches has a different mechanism though. Looking across the tank terminals, the switching pair and the current source has an effective capacitance (negative or positive depending on different factors). Even without a varactor at all, this effective capacitance is modulated by noise. There is a hint at that effect in a paper by Jacob Rael and Prof. Abidi but the complete analysis is extended in the book (it should start shipping next week I believe).

What do you mean noise @KVCO=0? Do you mean if the varactor is removed? In that case, junction caps will provide the AM-to-FM noise conversion source. Yet the switches will play a bigger rule in flicker noise conversion.

As shown in my paper, AM-to-FM noise conversion due to inversion-type MOS caps exhibits two nulls: one at the infliction point of the varactor curve (due to its inherent symmetry) and the other at the zero bias (if the amplitude is small enough). At zero bias, the capacitance variation is very small and therefore no noise conversion is possible (this means that Kvco also drops to zero) but we regard Kvco here as a "result" of a certain capacitance curve determined by the varactor. If the amplitude goes very large then capacitance averaging will provide some tunability, thus some noise conversion as well.

Now which part was not convincing to you?


Back to top
 
 
View Profile   IP Logged
Pages: 1
Send Topic Print
Copyright 2002-2024 Designer’s Guide Consulting, Inc. Designer’s Guide® is a registered trademark of Designer’s Guide Consulting, Inc. All rights reserved. Send comments or questions to editor@designers-guide.org. Consider submitting a paper or model.