how to setup tstab in pss and pdisto

I am simulating LNA+Mixer with a bias circuitry and digital controls.
The total design is huge.
I realized that a correct tstab setup gives a very fast simulation
results.
I am using 2110.1 MHz LO signals, 2110 MHz and 2110.05 MHz RF inputs
to simulate IIP3.
Can anyone tell me how to properly determine tstab?

thanks.

 

There is a convergence_radius VS iteration plot in the spectreRF appendix that show what a well converging steadystate (or QPSS) looks like. My guess is that you could increase your tstab if you did not get an exponential decrease in convergence radius with your first simulation.
BTW, did anyone write a little applet that generates this plot from the spectre log ?

Unfortunately, I wouldn t know how to set tstab upfront, knowing only circuit characteristics and overall transient/startup behaviour.
You could try this question on Ken Kundert's site.