Problem CR analysis using Golden Gate

Hi,
Im new here so if my question is not related Im sorry.

In the past week I have a problem running a CR analysis. I get the following error:


I tried to solve using the error description but didnt manage.
I would be grateful for your assistance

Mike

(1) Increase Order of CR
(2) Increase Oversampling
(3) Invoke initial transient analysis.

Show me "*.gg" netlist.

aretancos said:

Hi,
Im new here so if my question is not related Im sorry.

In the past week I have a problem running a CR analysis. I get the following error:
View attachment 142900
I tried to solve using the error description but didnt manage.
I would be grateful for your assistance
Mike

Click to expand...

Convergence errors may have different sources and there is unfortunately no unique answer to such convergence problems.
Harmonics Number,Convergence Method, etc. can be sources as well as models themselves.Any kind of loops,floating nodes,very high or very low impedance nodes, etc. may also be some
reasons.If you post your circuit here, I can say something more concrete..

Hi,
I tried to do your suggestions but could not help me.
How can I see the "*.gg" netlist? Im new in virtuoso

Hi,
I dont have any kind of loops,floating nodes,very high or very low impedance nodes. I cant put the circuit because its my PhD work.
Maybe there is any suggestion?
I tried to change the parameters from these following pictures but I have no idea what they mean so it did not help me at all.

aretancos said:

Maybe there is any suggestion?

Click to expand...

Show me "*.gg" netlist.

You use "HB/FDPC(Harmonic balance with the frequency-domain Krylov preconditioner)".

Try to invoke time domain preconditioner.
And use time domain solver such as TB, TS or Hybrid.
"HB/TDPC(Harmonic balance with the time-domain preconditioner)"

Generally, CR Analysis of Goldengate is very weak for circuits which include logic gate, comparator, hysteresis, etc.

Again show me "*.gg" netlist.
https://www.edaboard.com/showthread.php?t=356735#2

Hi,
I used time domain preconditioner and time domain solver such as TB and it worked for me.
I just want to ask if the results I got are the same if i used the previous setting that did not work for me.
Because the rest of the circuit I simulated using frequency domain and HB.
Im asking because it is my PhD and really important to me.

Thanks

There is no preconditioner selection for TB.

aretancos said:

I just want to ask if the results I got are the same
if i used the previous setting that did not work for me.

Click to expand...

I can not understand what you want to mean at all.

HB/FDPC can not converge for your circuit.
So whether it is same or not is meaningless.

HB/FDPC : can not converge
Steady-state solver type = Harmonic-Balance (HB)
HB Preconditioner type = Frequency-domain (time_domain_preconditioner=0)

TB : can converge
Steady-state solver type = Time-Balance (TB)

How about HB/TDPC ?
Steady-state solver type = Harmonic-Balance (HB)
HB Preconditioner type = Time-domain (time_domain_preconditioner=1)

Thanks.
How about HB/TDPC ?
Steady-state solver type = Harmonic-Balance (HB)
HB Preconditioner type = Time-domain (time_domain_preconditioner=1)
It dost not work!
What I meant in my question is this:
Until the error I used to simulate the circuit with HB/FDPC.
Right now, I will be simulating with time domain preconditioner and solver TB.
If I go back and simulate my circuit with time domain preconditioner and solver TB. would I get the same results as simulating with HB/FDPC? (If not si it is really matter which one I use)
Im asking because I preform CR analysis in order to see the different harmonics.

- - - Updated - - -

Hi,
I tried to simulate my previous circuit with time domain preconditioner and solver TB and compared to HB/FDPC (Im talking about my first circuit that I did not have any error).
If I compare the results I see that the results are different. What should I do? I think I must use HB/FDPC but it does not converge.
Are there any solutions?

aretancos said:

Right now, I will be simulating with time domain preconditioner and solver TB.

Click to expand...

Again, there is no option to choose preconditioner for TB and TS.
We can never choose frequency domain preconditioner for TB and TS.

aretancos said:

I tried to simulate my previous circuit with time domain preconditioner and solver TB
and compared to HB/FDPC (Im talking about my first circuit that I did not have any error).
If I compare the results I see that the results are different.
What should I do?
I think I must use HB/FDPC but it does not converge.

Click to expand...

Accuracy of TB and TS is worse than HB/FDPC and HB/TDPC.

aretancos said:

Are there any solutions?

Click to expand...

Tottaly performance of Goldengate is very bad.

Currently even Keysight does not advertise Goldengate anymore.
Rather I'm very surprised that you use Goldengate.

Again CR of Goldengate is very weak for circuits which include pulse shaped operation.
Maybe your circuits require Shooting-Newton method as PSS.

Goldengate have HB, TB and TS as PSS.
However TB and TS of Goldengate are very suspicious.

Use other simulators such as Synopsys HSPICE, Cadence Spectre or Keysight ADSsim.

Both Synopsys HSPICE and Cadence Spectre have Shooting-Newton-PSS.

Keysight ADSsim has only HB-PSS.
However I think ADSsim is better than Goldengate.