Page 1 of 1

Convergence value

Posted: Wed 15 Apr 2020, 13:33
by openems
Hey, I am designing an antenna and I have given timesteps 100000 and Endcriteria as -5(-50dB). But even after 100000 timesteps convergence value is -36 to -38dB(convergence value is saturating nearly at -35dB after 40000 time steps). Is the final results will be fine or do I have to increase time steps?

Re: Convergence value

Posted: Thu 16 Apr 2020, 00:50
by jjmaloss
My experience is that when simulation reaches a plateau around 35 dB, something is not right in the setting of the simulation. I become really doubtful when spurious oscillations occur. May be you should try to block the simulation at 1.e-3 and see if things are consistent, I may be wrong.

Re: Convergence value

Posted: Thu 16 Apr 2020, 06:41
by openems
Is -30dB convergence is good enough to get proper results or not? It is not about simulation time..

Re: Convergence value

Posted: Sat 18 Apr 2020, 16:24
by thorsten
The lower the remaining energy has decayed the better it is. But as jjmaloss said, sometimes e.g. static charges may get trapped and cannot decay or a high Q resonance may resonate forever (in both cases using more realistic losses may help).