The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> Circuit Simulators >> traponly in spectre
https://designers-guide.org/forum/YaBB.pl?num=1247882391

Message started by Dipankar on Jul 17^th, 2009, 6:59pm

Title: traponly in spectre
Post by Dipankar on Jul 17^th, 2009, 6:59pm

Dear Experts,

Just an experience sharing. The topology attached below fails to converge with traponly . The voltage at some internal nodes goes to unacceptable high value and SPECTRE gives up after running for certain period. WARNINGS shown in the processes are of two types :
1. source/drain junction leakage (not at this diode but at some other mosfets in side "ckt1" or "ckt2" exceeds 'imelt' (in my model 'imelt' =100 mA) which is absurd.

2. vgd at some mosfets inside ckt1 and ckt 2 exceeds 'vbox' (=12V) - again absurd.

When I resort to other integration methods with same reltol,vabstol,iabstol values - simulation goes smoothly.

The diode in pic is not a macromodel. It's a PDK device - n-type diode, i.e., n+ in p-sub . BREAKDWN VOLTAGE ~ 11V. reverse leakage ~10-20 pA. The voltage at net1 can't be more than 5-6 V. So we are not going to te 'knee' of I_V curve . So what's the problem ? Any explanation ?

Title: Re: traponly in spectre
Post by Andrew Beckett on Jul 20^th, 2009, 5:02am

I'm assuming the convergence failure you're talking about is during the transient, rather than DC (otherwise the integration method would not affect it)?

If so, it could be that there is some instability in the circuit which gear2 is damping; this instability could be real, or it could be as a consequence of poor models (either transistor models, or sometimes things like non-linear resistor models with multiple solutions, one of which is completely non-physical).

It's virtually impossible to tell what the root cause is without seeing the netlist and models. If you can't post it here, you might want to contact Cadence customer support to let us take a look and provide more targeted advice.

Regards,

Andrew.