The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> RF Simulators >> Weird results from PAC simulation
https://designers-guide.org/forum/YaBB.pl?num=1204016124

Message started by harmonics on Feb 26^th, 2008, 12:55am

Title: Weird results from PAC simulation
Post by harmonics on Feb 26^th, 2008, 12:55am

Recently I encountered some problems with PAC simulations so I decided to test the PAC simulation.
I built the simple RC circuit with the R replaced by a switched capacitor C_R as shown in the following figure:

Vin -------(phi 1) -------(phi 2) ----------Vout
|
|
------
------ C_R
|
|
------
gnd

phi1 and phi2 are non-overlapping clocks controlling two NMOS switches respectively. Vout is connected to the ground (gnd) through another capacitor (not shown).
To verify the simulation issues, I constructed three similar circuits as follows (all with same input Vin and clock signals):
Circuit (1): NMOS model set 1, Vout = vosci
Circuit (2): NMOS model set 2, Vout = vosc
Circuit (3): Switched capacitor CR is replaced by its equivalent resistor R, Vout = vorc.

PSS simulations are similar for vosci and vosc while vorc is simply a dc voltage as expected. However, when proceeding to perform the PAC simulation, I obtained the following results (see attached).
From the graphs, it can be seen that vosci (Circuit 1) and vorc are as expected from simple RC circuit but vosc (Circuit 2) is very odd with a gain of below -100dB.
As the only difference between Circuit (1) and Circuit (2) is the NMOS model (BSIM3), could anyone help me solve the weird PAC results mystery? Thanks!

Title: Re: Weird results from PAC simulation
Post by Frank Wiedmann on Feb 26^th, 2008, 1:32am

Simplistic VerilogA models can show such behavior, see http://www.designers-guide.org/Forum/YaBB.pl?num=1199890595. However, I have never heard of such problems with BSIM3 models.

Title: Re: Weird results from PAC simulation
Post by harmonics on Feb 26^th, 2008, 10:56pm

Hi Frank,
Thanks for replying. I have absolutely no clue which of the BSIM3 model parameter(s) cause such PAC problem. I tried to compare the two sets of NMOS BSIM3 models and found that the one in Circuit (1) uses version 3.2 and Circuit (2) uses version 3.3. Also, there are some differences between the two models but majority of the parameters are defined (although the values are slightly different).
Could it be due to some missing parameters in the BSIM3 version 3.3 in Circuit (2)?

Title: Re: Weird results from PAC simulation
Post by Frank Wiedmann on Feb 26^th, 2008, 11:29pm

Might be. You can find the official documentation on BSIM3 at http://www-device.eecs.berkeley.edu/~bsim3/. If you have access to Cadence support, you could also send them your test case as there might also be a problem with the implementation of BSIM3 in SpectreRF.

You could also try to see what happens if you use the parameters of the new version 3.3 model with model version 3.2. I don't recommend that you use such a model for your design but it might give you some further hints on the cause for the problem.

Title: Re: Weird results from PAC simulation
Post by Geoffrey_Coram on May 21^st, 2008, 12:14pm

Version 3.2 is rather old; there were some bugs fixed in intermediate versions. You could also try just changing the version number (try 3.24).

Did you try some basic dc i/v curves over the voltage range you're working with?

-Geoffrey
(Post #1000!)

Title: Re: Weird results from PAC simulation
Post by Ken Kundert on May 21^st, 2008, 1:56pm

SpectreRF is telling you that there is no path for the signal from the input to the output. That may be due to the model, but more likely there is something wrong with your circuit. Perhaps your switch transistors are not turning on? Try debugging the circuit.

-Ken