Discussion Forum

This is not a normal problem. I would recommend re-installation of the software from the original DVD. It only takes a few minutes which will save you a considerable amount of time in trying to fix this problem.

I faced the same problem in 4.1 too.. And when I got the 4.2 DVD I removed 4.1 and installed the new version. I still get this problem.. Once I change something in the drawing I can't change the physics settings.

Pls help me tackle this issue.

Thanks,
Sree

Hi

That one I have never experienced.

Have you tried to clear your solutions, meshes and rest the model ?, and/or restarted comsol again.

Sometimes, if you are too quick to move up and down an edit in a complex structure tree, then I wounder if some update commands do not interfere with each others.

--
Good luck
Ivar

Hi

That one I have never experienced.

Have you tried to clear your solutions, meshes and rest the model ?, and/or restarted comsol again.

Sometimes, if you are too quick to move up and down an edit in a complex structure tree, then I wounder if some update commands do not interfere with each others.

--
Good luck
Ivar

Hello Ivar,

Thanks for the reply. I tried clearing the solutions and meshes.. I still get that problem. I have created 4 models now and all of them have this problem. Any change in the goemetry after the physics are defined to the old geometry doesn't work. Surprisingly this is not the case with 2D simulations !! Only in 3D simulations this problem occurs..

What does 'Rest the model' mean ??

Thanks,
Sree

Hi

Well I ave no problem with my models ;) so it must be something specific with your install.

You have also the "File - Reset history" command that cleans up your file, particularly usefull when you want to save it in ".m" script format as its far more compacter after a "reset".

Sometimes this command cleans up issues, but some times it reports errors too, not sure exactly when and why, mostly it does its job nicly, though ;)

--
Good luck
Ivar

Hi all,

I'm experiencing the same issue. I built a geometry and could perfectly solve my problem. After a little change, I get the same "Could not trace curve" error.

Please find attached the faulty model, the respective parameter is rs (the radius of the central sphere). If you change rs to 26 micron, everything solves nicely. All other radii between 27 and 35 micron throw the error.

I could reproduce the error on three different machines, all with version 4.2.0.228, so it does not seem to be the installation itself. All solutions proposed in this thread (clearing mesh, solvers, history,...) did not help.

Thanks to everyone who takes a look at the model!

Best,

Martin

Hi

I suspect its your line that is giving you some problems, try to draw it all the way to the surface of the inner shell/sphere.

From V4 a geometry of 2 levels below max dimension must be part of a higher order type element, i.e.a line, in 3D must end on existing boundaries, or be part of an edge being the intersection of two boundaries. The same for a point must be part of an existing line, respecting the rule above

Sorry but that was not enough ;) after some time the error message reappeared. Bit if I remove the intersection it seems to become stable again, anyhow, the intersection is made by default during the finish Union

Anyhow, the comment above was an issue in 4.2, perhaps no longer in 4.2a ;)

by the way, I cannot see any GND connection, how do you define your reference voltage ?

--
Good luck
Ivar

Hi Ivar,

thanks a lot for your input. You were absolutely right - the intersection does not make any sense at all. Removing it in this model did in fact not help, but after rebuilding the model from an earlier version it did not show this behavior again.

As far as I have understood the lumped port feature, you're allowed to use it when the device dimensions are short compared to the wavelength. And because I did not find a ground connection feature in the emw-interface, I did not include one.

Actually, the results I get from the field simulations make sense without that connection.
If you can tell me how to implement GND properly, I'm nevertheless happy to try it and look forward to comparing the results.

Best,

Martin

Hi

your are right, but normally we have an input and an output for the energy, the external sphere will do the job I believe in your case, so do not bother further for my comment ;)

--
Good luck
Ivar

I've seen the "could not trace curve" problem too, in various version 4.x complicated 3D models with embedded surfaces and curves imported from version 3.5a. I've been assuming this is simply a subset of the issues that Comsol Multiphysics often has with properly resolving complicated highly-detailed geometries. My solution, to the extent I have one, is generally to completely recreate/redraw/recombine the offending parts, or even the whole geometry (since sometimes it is hard to figure out which parts are the source of the trouble) in a different (but hopefully equivalent) manner. Just remember, if it was always easy, then anyone could do it -- and then they wouldn't have to pay us so much!