Discussion Forum

Hi

you have some small gaps in your model between the domains, this means that your domains to not couple as they have no common boundary. Check your geometry, the two domains should have a comon complex boundary split up by your strip lines. Then probably you can also better manage to select your thin Perfect electric conductor lines that are now there as nodes but not connected to the model
--
Good luck
Ivar

Hello Ivar,

thank you for your quick reply. I tried to redo the geometry, changing the way how the center conductor and the gaps at its side are defined. However I am not sure I completely understood what you mean by the gaps between the domains... how can I create a common boundary if there is the center line and the gaps in between? Which domains have to have this common boundary, is it the two sides of the metallization, separated by the center line and the gaps?

Now with the new geometry definitions the meshing finished without any errors, but when calcuating I get 'Undefined value found in the equation residual vector.' I have a Frequency Domain and two Boundary Mode Analysis', one for each port. I only tried calculating at one specific frequency (3 GHz), and as port type I have numeric (in the geometry: rectangular boudaries on top of the feed lines for the resonator, in yz plane).

Best regards,

Diana

Hi

in ACDC or RF you need air or vacuum in between the objects with a potential to allow the physics to be correctly solved. So id you have thin air gaps between your parts, be sure you ncapsule everything in "air"

Note with such large scale ratios, it's often wise to cut your air domain into several parts, particularly to fill the gaps

--
Good luck
Ivar

I have a big air-box around everything. I will try cutting it into several thinner ones. Would you leave the metallization only on the workplane or would you suggest to create an extrusion with the thin-film thickness (which is very small compared to all other scales of the model)?

Thanks,

Diana

Hi

often thin metallisations are better off by using surface layers, but you must take care that you do not "glue" other parts together, else you must set up isolation layers elsewhere when you expected to have some thickness of "air"

Such high ration structured geoemtries are sometime slightly delicate to get optimised for FEM

--
Good luck
Ivar

Hi Ivar,

many things work better know after playing around a little. Thanks again for the advice. One last thing is very confusing: I am not sure which type of port is appropriate. In the coupled_line_filter or cpw_bandpass_resonator models they use lumped ports with the option 'cable'. However, I always get errors with the lumped ports (except if I define the width and length manually). When is it wise to use lumped ports and in which situations are e.g. numeric ports better for frequency-dependent RF simulations?

Best regards,

Diana

Hi

sorry but there I'm stuck, I do not have enough experience with these different ports, I have been doing to much in other parts of the physical world lately, but there are many other skilled COMSOLers that should know RF better than I ;)

--
Good luck
Ivar