The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Modeling >> Transmission Lines and Other Distributed Devices >> Ansoft Q3D 2D solver result mismatch
https://designers-guide.org/forum/YaBB.pl?num=1265084328

Message started by neoflash on Feb 1st, 2010, 8:18pm

Title: Ansoft Q3D 2D solver result mismatch
Post by neoflash on Feb 1st, 2010, 8:18pm

I have encountered an issue of using Ansoft Q3D's 2D solver.

The problem is that I have one demo FR4 trace's geometric information and w-element model extracted by vendor. I was redoing the modeling in Q3D.

I can have exactly same sdd21 result. However, the time domain result is very different and I'm sure my own result is incorrect.

If I export hspice w element table model, the result is very different. If I use touchstone model, the result is very close.

I have no idea what is wrong with the exported w model? Any suggestion?

Title: Re: Ansoft Q3D 2D solver result mismatch
Post by pancho_hideboo on Feb 2nd, 2010, 1:51am

Please describe correctly so that eveyone can understand.


neoflash wrote on Feb 1st, 2010, 8:18pm:
I can have exactly same sdd21 result.
However, the time domain result is very different
and I'm sure my own result is incorrect.
What do you mean by "the time domain result" ?
What do you compare with ?


neoflash wrote on Feb 1st, 2010, 8:18pm:
If I export hspice w element table model, the result is very different.
What do you compare with ?


neoflash wrote on Feb 1st, 2010, 8:18pm:
If I use touchstone model, the result is very close.
What do you compare with ?

Title: Re: Ansoft Q3D 2D solver result mismatch
Post by neoflash on Feb 2nd, 2010, 6:26am

Time domain result means pulse response and PRBS output response of the transmission line.

If I use touchstone output model, the prbs response is fairly close.
If I use W element, the prbs response is very different and has much less loss.


Title: Re: Ansoft Q3D 2D solver result mismatch
Post by pancho_hideboo on Feb 2nd, 2010, 6:37am


neoflash wrote on Feb 2nd, 2010, 6:26am:
Time domain result means pulse response and PRBS output response of the transmission line.
If I use touchstone output model, the prbs response is fairly close.
What is it close to ?
Is there any reference ?


neoflash wrote on Feb 2nd, 2010, 6:26am:
If I use W element, the prbs response is very different and has much less loss.
What do you compare with ?
Is there any reference ?

Again describe correctly.

I guess followings.
 - You have some reference as correct results.
 - Results of transient analysis using s4p(4 ports s-parameters) from Q3D are cosistent with reference results.
 - But results of transient analysis using W-element from Q3D are not consistent with reference results.

My guesses are correct ?
If so, Q3D itself is correct.
Your settings for dumping W-element model file from Q3D are wrong.

Upload your netlist of HSPICE, s4p file and model-file for W-element.

Title: Re: Ansoft Q3D 2D solver result mismatch
Post by neoflash on Feb 2nd, 2010, 9:54am


pancho_hideboo wrote on Feb 2nd, 2010, 6:37am:

neoflash wrote on Feb 2nd, 2010, 6:26am:
Time domain result means pulse response and PRBS output response of the transmission line.
If I use touchstone output model, the prbs response is fairly close.
What is it close to ?
Is there any reference ?


neoflash wrote on Feb 2nd, 2010, 6:26am:
If I use W element, the prbs response is very different and has much less loss.
What do you compare with ?
Is there any reference ?

Again describe correctly.

I guess followings.
 - You have some reference as correct results.
 - Results of transient analysis using s4p(4 ports s-parameters) from Q3D are cosistent with reference results.
 - But results of transient analysis using W-element from Q3D are not consistent with reference results.

My guesses are correct ?
If so, Q3D itself is correct.
Your settings for dumping W-element model file from Q3D are wrong.

Upload your netlist of HSPICE, s4p file and model-file for W-element.


Your guess is correct. I agree that Q3D is generating consistent result with the reference. However, the dumped w model has problem.

I'm also looking into the model file dumped. One problem I see is that the C matrix doesn't change much with increased frequency. If I replace the C matrix with the one in the reference design, the transient result is much closer to the reference.


Title: Re: Ansoft Q3D 2D solver result mismatch
Post by neoflash on Feb 2nd, 2010, 9:56am

Actually, I almost see no setting for the w-element export operation.

I just right click the analyze menu, and chose "export circuit". Then in the output file selection chose hspice w element format. And click export. Am I missing any option setting?

Thanks for your inputs.

Title: Re: Ansoft Q3D 2D solver result mismatch
Post by neoflash on Feb 2nd, 2010, 5:17pm

Exported W element based on table data.

The Designer's Guide Community Forum » Powered by YaBB 2.2.2!
YaBB © 2000-2008. All Rights Reserved.