Designer’s Guide Community

mohammadreza

Junior Member

Offline

Posts: 19

convergence error in corner analysis
Sep 09^th, 2010, 11:59am

I am simulating a 2.45 GHz power amplifier and I got approproate results in typical conditions. However, when I want to run corner analysis for worst speed, the simulation quickly aborts for convergence at the begining. my transient analysis also stops aftera few nano secons and shows my bias voltage has suddenly increased so much. But, I can get reasonable result from transient analysis if I set the errpreset to liberal.

I have checked for instability possibilities and broke my feedback control loops, but no difference.

can any body help me please?

Back to top

IP Logged

mohammadreza Junior Member Offline Posts: 19	Re: convergence error in corner analysis Reply #1 - Sep 9^th, 2010, 12:41pm I have to add that when I set cmin to 2fF, it works, but the performance degrades very much! how can get rid of that convergence error without adding capacitance?
Back to top	IP Logged

pancho_hideboo

Senior Fellow

Offline

Posts: 1424
Real Homeless

Re: convergence error in corner analysis
Reply #2 - Sep 9^th, 2010, 8:25pm

I can't find out any design issue in your post.
Your question is no more than very specific vendor's simulator's issue.

The followings are general notes for you.

- Always describe correct tool's name and vendor's name which you use as tool or simulator.
- Don't do multiple posts which are same content.
- Don't request source code or behavioral model without any efforts.
- There are many simulators which have analyses called as PSS, PAC and Pnoise.
- Describe in detail with using correct terminologies.
- Warnigns are different from Errors.
- ADS is not name of simulator.
- There is no tool which name is Cadence.
- Don't use Direct Plot of Cadence ADE blindly without knowing definition.
- All gains in Direct Plot of Cadence ADE are "right", "true" and "practical" voltage gain.
- Don't mix up Simulation with Post Processing. They are completely different phase.
- MATLAB are different from Simulink.
- Learn measurements using actual instruments. Not "EDA Tool Play

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

mohammadreza

Junior Member

Offline

Posts: 19

Re: convergence error in corner analysis
Reply #3 - Sep 10^th, 2010, 2:52am

Well, Thanks, I was in hurry!

I am using Vadence spectre for simulating my RFPA.

My power amplifier works properly when the typical mean model is chosen in the library and I now I need to run a corner test for it. But, when I choose the worst speed type for the transistors corner, the simulations breake with the convergense error on stating that didn't converge with 1ps step.

Then I tried to find the reasone. I ran transient analysis with the worst speed type. It didn't converge when I set the errpreset to moderate. But it continued for a few nano seconds (my PA input is 2.4GHz). The problem was one of my bias voltages which suddenly fell for several volts (my supply is 1.8V). and that's the final point for the simulation.

But, when I run the transient anlysis with liberal errprest, it shows a stable reasonable output.

Then I played with the PSS options, setting the cmin to 2fF, the simulation converged. But, the performance had fell too much (almost 5 dB in output power).

I wanted to know if I can do a simulation without setting the cmin, which probably degrades my performance!

Back to top

IP Logged

pancho_hideboo

Senior Fellow

Offline

Posts: 1424
Real Homeless

Re: convergence error in corner analysis
Reply #4 - Sep 10^th, 2010, 7:16am

mohammadreza wrote on Sep 10^th, 2010, 2:52am:

I am using CVadence spectre for simulating my RFPA.
................................................................................
...............................................................
I wanted to know if I can do a simulation without setting the cmin, which probably degrades my performance!

Maybe you are facing "Blowup" which we very often face in Cadence Spectre especially for Shooting-Newton.

I have a similar case as you.
PSS analysis of my PA circuit for slow corner model where Vth is high can not converge except for Agilent GoldenGate Simulator.
� � - Agilent ADSsim HB ; Can not converge.
� � - Cadence Spectre HB ; Can not converge.
� � - Cadence Spectre Shooting-Newton ; Can not converge after very disgusted long time iterations which I've often faced in Cadence Spectre's Shooting-Newton.
� ��- Agilent GoldenGate HB(CR) ; Can converge quickly and give reasonable results.

Try to use other vendor's simulator such as HSPICE which is absolutely Golden Standard Simulator.

But being different from your case, Transient analysis of my PA can converge in any simulator even for slow corner model.

I recommend you to check DC operation points of your circuits.
Generally Vth of MOSFET is high for worst case corner, so some MOSFET could be cutoff region.

Is there any "Self-Bias" circuitry which requires some "Startup mechanism" included ?

Back to top

« Last Edit: Sep 11^th, 2010, 3:31am 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

mohammadreza Junior Member Offline Posts: 19	Re: convergence error in corner analysis Reply #5 - Sep 10^th, 2010, 12:26pm Yes, the operating point is affected.
Back to top	IP Logged