resonances in s11 curve missing in CST
I am trying to simulate a TEM cell in CST, to compare the simulated s11 to the measured one. On the measured curve I see some resonances (the expected ones for TE10 mode but also two other frequencies which I do not know where they come from) In CST no resonance at all, regardless on how many frequency samples I used; at first I thought I miss them when using a larger frequency step; this is what happened when I did that for the VNA. I use discreet ports as feed and load. Any idea why no resonances seem to occur in the simulated model?
Thanks!
Check that your discrete ports connect to what you intend them to - it can be all too easy to have erroneous shorts/open circuits. Are you using the computational boundary conditions to define the physical model? If so, check that you are using the electric and/or magnetic boundaries you expect and haven't inadvertently selected an open boundary.
Thank you very much for your answer. I am not sure that it is very clear to me, so I attach a picture of the input of the TEM cell (including the input port) [ ../imgqa/eboard/EM/EM-22uh3y2x2u1.jpg ] The boundary conditions are set to electrical in all directions. If one of the ports were accidentaly short/open, I suppose I wouldn't get a very good s11 curve (what I get is actually not far from what i measure, it's just that I see no resonance at all on the curves (I attach an image of computed s11: ../imgqa/eboard/EM/EM-q5xmcbpzvbi.jpg )
That response looks pretty good :) ...and aye, you're right - if the port was open/short, you'd have a meaningless s11. From your first post I didn't know if you were seeing anything sensible at all.
Nothing springs to mind... Could you please post your measured response?
this is the measured curve: ../imgqa/eboard/EM/EM-1izy2ywv1xc.jpg
The computed resonances of TE10 are at about 700 and 976 MHz (I see them being at 700 and 991; I suppose that they are the expected ones) The resonances of other TE modes are at frequencies higher than what I'm interested in. I don't understand where from do they appear the other ones. And why I don't see any of them in CST...
The resonant features are extremely narrow... I suspect they're simply being "smeared' into non-existence by the finite frequency resolution arising from the finite simulation duration. The 'period' of the ripple observed in the calculated frequency response at the upper and lower frequency limits is suggestive of the finest feature able to be clearly resolved.
I've bumped against this sort of problem many times before - and if anyone knows how to arbitrarily increase CST's frequency resolution, I'd be most interested to learn!
I've had some measure of success combining settings that serve to increase the simulated excitation time (thus decreasing the frequency step size of the Fourier transformed result) - such as increasing the requested steady-state accuracy limit and setting the simulated frequency range to narrow bands around the features of interest. Perhaps try calculating from 600 - 800 MHz, with an accuracy of -50 dB?
Thank you for your suggestion. 'Till your answer arrived I have tried using the Frequency Domain solver and this is what I got (using 2e4 frequency samples in the 0.1-1.2GHz): ../imgqa/eboard/EM/EM-lg0g3osomy5.jpg It is interesting that there appear to be resonances at about 707, 780 and 978 MHz. The 1st and the 3rd is what the theory suggests. Only the one at at 780 is weired to me... I don;t see anything at 600 (in CST). Anyway it is quite fine.
As long as I have used mostly the transient solver, it's not clear to me why do I see this resonances when using the frequency domain solver and I do not see them with the transient solver. How's it "working" the frequency solver compared to the transient solver.