Hello
I simulated a simple pmos current mirror circuit with a pulse current source as reference in proteus 8. But it gives the following error:
transient GMIN stepping at time = ...
Timestep too small; ...
Can anybody help?

 

Welcome to AAC!
Have you given the current source a finite resistance?
Have you tried making the two PMOS devices slightly different rather than identical?

 

The source of a timestep error in Spice can be hard to pinpoint.
Sometimes just a small change in the circuit will eliminate the problem.
Try removing part of the circuit at a time to see if you can determine, which area is the problem.

Also you might go into the Spice transient simulation values and try reducing some of the default accuracy parameters by a factor of 10 (shown below in the righthand column from LTspice).
If you can change the default integration method to Gear, that also sometimes helps.

 

Welcome to AAC!
Have you given the current source a finite resistance?
Have you tried making the two PMOS devices slightly different rather than identical?

Thanks for your reply.
Putting resistance doesn't help, but I don't know how to change transistor properties in proteus.

 

The source of a timestep error in Spice can be hard to pinpoint.
Sometimes just a small change in the circuit will eliminate the problem.
Try removing part of the circuit at a time to see if you can determine, which area is the problem.

Also you might go into the Spice transient simulation values and try reducing some of the default accuracy parameters by a factor of 10 (shown below in the righthand column from LTspice).
If you can change the default integration method to Gear, that also sometimes helps.

View attachment 129304

Thank you.
But I simulated the circuit in proteus not spice, how can we change transient simulation values in proteus?

 

And I should add that when I change reference current from pulse to dc current source, it answers properly.

 

Thank you.
But I simulated the circuit in proteus not spice, how can we change transient simulation values in proteus?

I'm not familiar with Proteus, but I would think it uses a Spice engine with Spice simulation parameters.
Most available analog simulators use Spice as the basis for the calculations.
You need to search for simulation options in Proteus.

 

I don't know how to change transistor properties in proteus.

Try adding, say, 1 milliOhm in series with the emitter of one (only one) of the transistors forming the current mirror. The intention is to avoid exact symmetry so that Spice/Proteus isn't trying to assign two values to the same node at exactly the same instant (as far as I understand the machinations of Spice ..... which could be way off ).