Learning Forum

[Scott]

I am trying to simulate a circuit that is periodic in one dimension using master/slave boundaries (now called coupled boundaries?). But when I add them the simulation hangs in one of the preliminary steps, and after a few minutes I get an error that the Com engine is non-responsive. If I replace the master/slave boundaries with radiation boundaries then the simulation completes in seconds. That leads me to believe that it's not a problem with communication with the license server. When the sim is started, before and during hanging, there is no significant usage on my CPU (it pops up to ~60% load for a tick then falls back down) and my RAM load stays constant at around 50% of my 16 GB. So it doesn't seem related to an actual CPU or RAM bottleneck.

Below is an image of my circuit, it is a very short section of CPW transmission line with a very long capacitive stub. The CPW is bisected by a H-wall symmetry boundary. The bright green panels are the wave ports. The master and slave boundaries are on the same planes as the ports. The thin outer boundaries above, to the left (across from the symmetry boundary), and below are all radiation boundaries.

Zoomed out:

[Scott]

Let me be a little more clear: As I watch the RAM load there is no change as the simulation is started or hanging or after it is self-aborted. In the analysis options I have allowed the sim to use up to 70% of my 16GB of RAM (even tried at 90% though that would really squeeze the OS if it tried to use that much). So my conclusion is that the RAM is not an issue.n

[Scott]

I just tried increasing the sim frequency from 6GHz to 600GHz and the simulation seems to get a little bit further in the Constructing and/or retrieving meshes on Local Machine step, I see a few lines of text flash on and off below the progress bar before it gets stuck on one particular step.nAlso of note:in the analysis I am only solving for the ports, not performing a heavy analysis.n

[Scott]

I just tried increasing the width of the port by 4x and the height by 2x, (back to f=6GHz) the sim still hangs on the same step.n

[Praneeth]

HiArray,nThank you for reaching to us. Please provide the following information to help us serve you better:nVersion of the AEDT tool you are using.nSolution profile data (Right mouse click on your solution setup to get it).nPlease use Upload Image option to attach the screenshots of the solution profile data.nAll the very best.nBest Regards,n

[Scott]

ANSYS Electronics Desktop 2018.1.0nn

[Praneeth]

nWe appreciate your patience on your query.nCan you try assigning Lumped port instead of wave port to your model and check if the issue still persists?nAlso please send the snapshots of your model with boundaries and H wall symmetry highlighted to help us serve you better. nAll the very best.nBest Regards,n

[Scott]

nThanks for your attention. Changing the ports from wave to lumped did not make any difference, the simulation still became nonresponsive right after starting.nAs requested, some screenshots of my boundary conditions:nH-wall symmetrynMaster boundary: (coincident with the XZ plane) (the paired slave boundary is lightly shaded in the background)nSlave boundary: (the paired master boundary is lightly shaded in the foreground)nRadiation boundary:nI am simulating a superconductor, so I've also added an impedance boundary to the surfaces of the conductors to model the added sheet inductance due to the kinetic inductance of the Cooper Pairs, as shown: (added to both the top and the bottom of the conductors, shown here is the top)nThose are all of my boundaries.nnThanks again,nScottn

[Praneeth]

Hi Scott,nLooks like your simulation model is huge (computational wise) and that is why it is hanging.nThe reason why you can solve it when you have radiation boundary is that Coupled boundaries takes lot of computational resources as you are trying to implement periodicity.nIf possible try to run the simulation in a bigger machine compared to the one you are currently using.nIf not, I will suggest you try trail and error method to see which conditions are overloading your computational resources. nAll the very best.nBest Regards,n