[Getdp] u-strip RF

michael.asam at infineon.com michael.asam at infineon.com
Wed Dec 3 17:01:54 CET 2014

Hi Dan,

I used the GetDP version compiled with complex PETSc, which also includes the SLEPc eigensolver.
This eigensolver is actually quite fast. Please run this example with the following options:
-solve -bin -pos -slepc
or even simpler:
Open Microstripline.pro (not .geo) with Gmsh. Then the right command line options are applied
by default. On my notebook the simulation takes about 18 seconds.

Of course you can use the symmetry here to reduce the calculation effort here. (I haven’t used it
here because I’m also simulating non-symmetric cases.)


From: Danny Holstein [mailto:dgholstein at embarqmail.com]
Sent: Wednesday, December 03, 2014 4:41 PM
To: Asam Michael (IFAG ATV BP D PD DES1)
Cc: getdp at geuz.org
Subject: Re: [Getdp] u-strip RF


Thank you so much, I especially like the comments you added.

I did try unsuccessfully running the problem and it appears the solution doesn't converge on my machine with the following messages:

Info    : GetDP -   998  2.3177217e+05  8.5152840e-01
Info    : GetDP -   999  2.3162576e+05  8.5099048e-01
Warning : GetDP - Iterative solver has iterated too many times
Info    : GetDP -  1000  2.2645535e+05  8.3199446e-01
Info    : GetDP - 1000 Iterations / Residual: 275968
Info    : GetDP - Arpack required 50 iterations
Also, I was curious with regards two issues:

  *   You don't apply symmetry along the center line, why not?
  *   Wouldn't seeding the solution with the solution to the DC case (freq = 0.) speed up resolution significantly?  I imagine e sub t could be specified, with h sub t similarly based on the DC current into a 50 ohm load.
Another consideration with regards seeding, the non-transverse fields will be very small (zero) starting at freq=0; if we swept the frequencies the field solutions to the lower frequency should always work well for the next, with the non-transverse increasing beyond when higher order modes become possible (the point at which we wouldn't be interested in using u-strip anyway).


From: "michael asam" <michael.asam at infineon.com<mailto:michael.asam at infineon.com>>
To: dgholstein at embarqmail.com<mailto:dgholstein at embarqmail.com>, getdp at geuz.org<mailto:getdp at geuz.org>
Sent: Wednesday, December 3, 2014 3:28:23 AM
Subject: RE: [Getdp] u-strip RF
Hi Dan,

here is a small example for simulating the impedance and propagation constant
of a simpe microstrip line (2D cross-section).


From: getdp [mailto:getdp-bounces at ace20.montefiore.ulg.ac.be] On Behalf Of Danny Holstein
Sent: Tuesday, December 02, 2014 9:26 PM
To: getdp at geuz.org<mailto:getdp at geuz.org>
Subject: [Getdp] u-strip RF


Has anyone done the formulation for the RF characterization of microstrip?  DC is clearly in the examples, but I don't see a microstrip formulation with determination of impedance and phase velocity (beta) in the frequency domain.


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.geuz.org/pipermail/getdp/attachments/20141203/454ed06a/attachment.html>