Designer’s Guide Community

kumar.g

Community Member

Offline

Posts: 51
Frankfurt, Germany

Transient simulation of Verilog-A model - Explanation required
Sep 17^th, 2010, 2:40am

Hallo All,

I have simulated a D-latch using OFET. When running a transient simulation it gave me the following warnings.

"Matrix is singular (detected at `NQ')"

"Zero diagonal at net1 and net1"
Then the simulation ended prematurely with convergence problem. Now I ran the simulation using a cmin of 1fF and then the zero diagonal warning disappeared and the circuit also converged. But the first warning is still there. I am giving a pulsed input to DIN but the output of the latch (Q and NQ) doesn't change. Is this error due to cmin or the first warning? I know that cmin is used at all floating nodes to avoid convergence problem. But does cmin affect the functionality of the circuit? I have attached the model(with parameters input file) and schematic. The inputs used are as follows.

DIN - pulse (-0.2 , -14 V); CLK - pulse (-0.2, -10.2V, freq = 100Hz) ,NCL- negated CLK

Back to top

ofet.zip

IP Logged

Geoffrey_Coram

Senior Fellow

Offline

Posts: 1999
Massachusetts, USA

Re: Transient simulation of Verilog-A model - Explanation required
Reply #1 - Sep 22^nd, 2010, 5:39am

Your code appears to set I(d,s) <+ 0 if (vds >= 0).

This seems completely wrong to me. I don't know what an "OFET" is, but the way you've drawn your schematic leads me to believe vds>=0 is the "normal" way the device is connected, and thus both devices will be completely off, presenting no conduction path for the node inv_op of inv_sch.png. Model writers will often put in a GMIN term to provide some minimal conductance.

Back to top

If at first you do succeed, STOP, raise your standards, and stop wasting your time.

IP Logged

kumar.g

Community Member

Offline

Posts: 51
Frankfurt, Germany

Re: Transient simulation of Verilog-A model - Explanation required
Reply #2 - Sep 22^nd, 2010, 6:36am

Hello Coram,
Thank you for your reply. Please note that the Vdd is -15V and the OFET is normally ON transistor. i.e it is ON even for Vgs = 0V... The output of INV switch between few mV and -14.75V.. The D terminal of load is -15V , S is few mV to -14.75V. So the Vds is not >= 0.. Similarly the S terminal of driver is 0V, D is mV to -14.75V. So teh Vds is not >= 0. Anyhow I will try to chk the schematic thoroughly..

Back to top

IP Logged

Geoffrey_Coram Senior Fellow Offline Posts: 1999 Massachusetts, USA	Re: Transient simulation of Verilog-A model - Explanation required Reply #3 - Sep 24^th, 2010, 1:49pm Well, OK, but still, if the simulator somehow gets to the point where it thinks vds > 0 (perhaps Newton's method ends up overshooting the origin), then your model provides no information about how to get back to a reasonable solution.
Back to top	If at first you do succeed, STOP, raise your standards, and stop wasting your time. IP Logged