Simulating Parallel Plate Capacitor in HFSS
I am trying to create a very basic simulation of a parallel plate capacitor to make sure I understand how to use HFSS. I created it with
two 150mm x 150mm PEC sheets, .1mm distance between the plates, LumpPort excited at the center of the plates
I don't know what the best frequency to use for this simulation is, my advisor suggestetd on the order of 100KHz, but from what I've read that means the airbox needs to be ≈.75Km (3e8/100e3/2 for total width, with model in center) which fails to sim.
I've been plotting the Z(im) parameter and find C from Z=1/(2*pi*f*C) -> C=1/(2*pi*f*Z).
εA/d = 8.84e-12*0.0225/.1e-3 = 1.989nF
C from Sim = 2.14nF
My advisor said HFSS should provide results that are spot on, so, what am I doing wrong here?
I attached my model to try and save the time and posts of a request for the model I'm using.
Thanks in advance.
FindCap.zip
I guess, that HFSS considers fringing fields, which suggests a more exact solution than your simple 1-D capacitor model.
Your results look very reasonable to me. If you increase the distance of 0.1 mm to a higher value, you will see even more difference of the values which with HFSS, the fringing capacitor was included.
At 150mm x 150mm sheet size and only 0.1mm (!) distance between the plates, the fringings fields should be small.
An estimate of the fringing fields is to increase the plate size by 1/2 dielectric height on each size, so that it would change from 150x150 to an effective 150.1 x 150.1
---------- Post added at 20:51 ---------- Previous post was at 20:41 ----------
The value that I get from Sonnet EM is 2.000 nF
I agree, that fringing fields can't explain a deviation of 7.6%. 2.0 nF is much more in accordance to the 1-D plate capacitor calculation, pacticularly with correct ε0 value of 8.854e-12. So there's apparently a problem with your simulation setup. A deviation of about 8 % isn't very high in practical EM simulation however.
Thank You for all the responses. I was also wondering how to determine the best solution frequency to use. I created a very broad frequency sweep to see how what happens as freq increases (10k - 20M). I found that I get results that are pretty close(1.4% off) to my calculated at around 1MHz-5MHz, but my solution frequency was still set to 100KHz. If I change my solution frequency to 2MHz, the results get slightly closer still. I'm curious if there is a good way to know where to set the Solution Frequency.
Thanks.
What is the sheet thickness on each plate?
They are 2D sheets that are perfect electric conductors
The key to this simulation is the mesh and the quantity of convergence. Converging on the default 'delta S = 0.02' does not make much sense here. The port is always looking into an open so the S parameters do not change much at all from pass to pass causing premature convergence.
To solve this... First, seed mesh to 5mm on the plates. This needs to be done as this structure is nowhere near a resonance and the mesh will not be fine enough based upon lambda refinement. Second, Create an output variable for capacitance and converge on this quantity...now we are converging apples to apples. I get a capacitance of ~2.04 nF. If I want to get closer, I can by increasing the mesh seeding...
I attached the modified file...it is in HFSS13 and NO I will not make it available in any other version.
Have Fun
Tallface65,
Thank you for the reply. I have taken your comments into consideration and changed the meshing of my model to improve the accuracy of the simulation. I was previously unaware of the ability to change the meshing.
Also, HFSSv13 works fine as that is the version I am using
Christopherdean,
Sorry, I know your file is HFSS13 (I always check the ASCII file to ensure ;) ) I was just stating that only one version will be made as on several occasions when I post an HFSS file, I get several PMs about having it translated to another version... usually by kids using hacked versions of HFSS10 or 11...really annoying. I was just trying to circumvent that for this one :)
Have Fun
What if the space is filled with dielectric?
How should be excite then