T-spice DC analysis question

t-spice+example

With T-Spice, I am having trouble doing dc sweeps because non-convergence of data keeps happening. So I get errors such as:

"Conventional DC operating point computation failed.
Gmin stepping failed
Final gmin value = 1e-008"

and etc. etc.. I imagine the errors are happening because I may have subcircuits operating in non-linear regions, and I think I know where those blocks are (for example, in bias networks/current sources). However I'm not sure how to get around those. As of now I'm substituting current sources with ideal sources, but I'm worried that simulation data will not accurately reflect actual performance. Does anyone have any tips?

P.S.: Maybe this should've been in "elementary questions" section?

pspice dc analysis

Hi,
I also had the same situation like you.But My circuit at MOS level,so I checked the Lenght of transistor,and I found it not valid with Model. I think you should check your circuit again(it maybe run in Winspice but not in T-spice)

can you post your circuit? what circuit that you simulate?

whatever anhtuan said is correct. You need to see your model file in which matrix will be given for maximum and minimum possible length and width of transistor. You should have the W/L value under given limit.

I'll keep in mind the tips just given.

Unfortunately, I can't post the exact circuit due to it being a research project, so it is not material I can post online. I'll try to post something similar though if I have time so I have something more concrete to ask.

Thanks for the replies.

EDIT: I can't post the exact schematic, but I can describe what I'm trying to simulate. I'm trying to simulate an analog (not digital) comparator which is more or less a diff amp with an output follower stage for current gain. The two stages are biased by identically sized current mirrors connected to a single bias network.

I'm primarily interested in simulating the diff-amp. The bias network compensates for temperature variations (although so far I'm not running simulations with variations in temperature). For now, I've simply replaced the bias mirrors with identical current sources for each stage.