Designer’s Guide Community

Mehran

Junior Member

Offline

Posts: 16

Cross couple LC VCO
Feb 03^rd, 2009, 5:20am

Hi All,
I am trying to design a cross couple LC vco. The Q of my tank is about 5. I have sized my transistor to have gm~1.5/Rtank. I have read in the text books that we should use 2-3 times instead of 1.5. It does oscillate in the simulation.
Can any body tell me how is the normal way people in industry make sure that their VCO will oscillate?
If I run the corner and it oscillates in all the corners, Is it the safe method?

Thanks in advance.
Mehran

Back to top

IP Logged

pancho_hideboo

Senior Fellow

Offline

Posts: 1424
Real Homeless

Re: Cross couple LC VCO
Reply #1 - Feb 6^th, 2009, 7:32pm

Mehran wrote on Feb 3^rd, 2009, 5:20am:

If I run the corner and it oscillates in all the corners

How to confirm stable and rigid oscillation in your simulation ?

Large Signal Steady State Analysis such as Autonomous Shooting Newton or Autonomous Harmonic Balance ?
Small Signal Analysis for evaluation oscillation margin by using Loop Gain or Negative Impedance ?

Back to top

Kita━━━━━━(ﾟ∀ﾟ)━━━━━━ !!!!!
http://www7.plala.or.jp/ungeromeppa/flash/kita.html
http://www.youtube.com/watch?v=mjIxGh55bMM&feature=related

IP Logged

Mehran

Junior Member

Offline

Posts: 16

Re: Cross couple LC VCO
Reply #2 - Feb 7^th, 2009, 12:43am

Hi panchoo,
Thanks for replying.

the way I have in my mind is, running the transient analysis with small pulse of current (to kick-start)and I can see the oscillation build up and the steady amplitude, then I will run the PSS shooting and I will check if there is a steady state solution there also and I will check the amplitude with transient.

can you tell me how you experienced designer make sure that it will oscillate? �I didn't get your questions. �Are those questions the way I should go or you mean something else.
how do you measure the loop gain of your oscillator? �

Mehran

Back to top

IP Logged

pancho_hideboo

Senior Fellow

Offline

Posts: 1424
Real Homeless

Re: Cross couple LC VCO
Reply #3 - Feb 7^th, 2009, 4:29am

Mehran wrote on Feb 7^th, 2009, 12:43am:

Are those questions the way I should go.

Generally you should do both.

Mehran wrote on Feb 7^th, 2009, 12:43am:

how do you measure the loop gain of your oscillator?

Same as amplifier with feedback.

But for LC oscillator, I use Negative Impedance although I can also evaluate Loop Gain.
http://www.zen118213.zen.co.uk/RFIC_Circuits_Files/LC_Oscillator.pdf
Here you can evaluate Negative Impedances at pure small signal analysis and small signal analysis subjected to master large signal analysis.
From latter, you can also evaluate loss of oscillator at steady state.

By the way, have you been able to resolve the following issues ?
http://www.designers-guide.org/Forum/YaBB.pl?num=1229048701/8#8

Back to top

« Last Edit: Feb 8^th, 2009, 12:37am by pancho_hideboo »

Kita━━━━━━(ﾟ∀ﾟ)━━━━━━ !!!!!
http://www7.plala.or.jp/ungeromeppa/flash/kita.html
http://www.youtube.com/watch?v=mjIxGh55bMM&feature=related

IP Logged

oermens Community Member Offline Posts: 86	Re: Cross couple LC VCO Reply #4 - Feb 7^th, 2009, 2:44pm can you explain more about evaluating loop gain and negative impedance? which analysis do you use? sp? if i perform sp on the tank, the resistance should be approximately g_tank \|\| -g_active ?
Back to top	IP Logged

pancho_hideboo

Senior Fellow

Offline

Posts: 1424
Real Homeless

Re: Cross couple LC VCO
Reply #5 - Feb 7^th, 2009, 7:35pm

oermens wrote on Feb 7^th, 2009, 2:44pm:

if i perform sp on the tank, the resistance should be approximately g_tank || -g_active ?

You don't seem to understand measurement of interstage impedance.
But AC analysis is more suitable than SP analysis.
Port needed in SP analysis always perturbs oscillator since it has impedance.

oermens wrote on Feb 7^th, 2009, 2:44pm:

which analysis do you use? sp??

Small Signal AC Analysis and Small Signal AC Analysis subjected to Master Large Signal Analysis such as HB-SS and PSS/PAC.

http://www.designers-guide.org/Forum/YaBB.pl?num=1221012939
In Agilent ADS, such impedance probe(short probe) is provided as standard probe.
But you can create it using combination of vsources and isources by yourself.
In the above example, I use my own impedance probe for Cadence Spectre.
You can create differential impedance probe if you understand diff/com mode copncept correctly.
Also you can create diff/com Loop Gain Probe by yourself.

But If you are Cadence Spectre Player, STB are avaialble for evaluation Loop Gain.
http://www.designers-guide.org/Forum/YaBB.pl?num=1215501569/2#2

By the way have you been able to resolve the following issues ?
http://www.designers-guide.org/Forum/YaBB.pl?num=1228240587/26#26
http://www.designers-guide.org/Forum/YaBB.pl?num=1231617402

Back to top

« Last Edit: Feb 8^th, 2009, 12:37am by pancho_hideboo »

Kita━━━━━━(ﾟ∀ﾟ)━━━━━━ !!!!!
http://www7.plala.or.jp/ungeromeppa/flash/kita.html
http://www.youtube.com/watch?v=mjIxGh55bMM&feature=related

IP Logged

emad

Junior Member

Offline

Posts: 30
Cairo, Egypt

Re: Cross couple LC VCO
Reply #6 - Feb 8^th, 2009, 5:47am

Mehran

Obviously the 2-3 rule is meant as a guideline rather than a rigid number. It stems from the fact that gain compresses as oscillation grows as well as a PVT margin. It depends on the allowed signal swing. A larger swing requires higher loop gain to fully switch the pair. However, higher loop gain degrades flicker noise ( Can't capture that in equations yet but perhaps I'll do that one day).

In a nutshell, if your worst case, PVT, RLC-extracted, design has a gain of 1.5 then it's probably ok. If this is your typical, schematic level, 27C design then 1.5 won't do in the real world.

In my experience, 1.5 is a bit too small, 1.8-2.0 would give me better sleep in the nights between tapeout and measurement.

Emad

Back to top

IP Logged

oermens

Community Member

Offline

Posts: 86

Re: Cross couple LC VCO
Reply #7 - Feb 8^th, 2009, 9:38am

thanks for the reply, your insight is always appreciated.

pancho_hideboo wrote on Feb 7^th, 2009, 7:35pm:

By the way have you been able to resolve the following issues ?
http://www.designers-guide.org/Forum/YaBB.pl?num=1228240587/26#26

Yes we switched to MMSIM stream, but I didn't do much related to the ocean scripting.

pancho_hideboo wrote on Feb 7^th, 2009, 7:35pm:

http://www.designers-guide.org/Forum/YaBB.pl?num=1231617402

Yes.

Back to top

IP Logged

Mehran Junior Member Offline Posts: 16	Re: Cross couple LC VCO Reply #8 - Feb 9^th, 2009, 3:53am Hi Panchoo, Thanks for your help. I still have a problem with the spectrum on previous issue.
Back to top	IP Logged

Mehran

Junior Member

Offline

Posts: 16

Re: Cross couple LC VCO
Reply #9 - Feb 9^th, 2009, 4:43am

Hi Emad,
Thanks for replying.
May I know how do you measure the loop gain? Do you measure the loop gain of your oscillator as Panchoo said,in spectre, �putting a differential probe between the output of the VCO and the gates? or you first characterize your tank resistance and then use the DC operating point gm to calculate gm*Rtank?
The problem is, if you look at the attached file and results, loop gain in stability analysis is around 1.7 while if I use gm*Rtank=15.8m*130=2.054. �

Can you tell me how do you do it?

(PS: I have learn a lot from your papers)
[img][/img]

Back to top

Question.JPG

IP Logged