Interdigitated Capacitance in HFSS

[sandbo]

Hello,

I am trying to extract the capacitance of an interdigitated capacitor along a CPW waveguide.
I have been trying different ways, from using Y-parameter by modelling the the thing as a pi-network, however I didn't get something reasonable with the calculated capacitance being heavily frequency dependent, which I guess the inductance of the line played a role.

Then I turned to use surface integral method to find the capacitance,
that is by setting a voltage source of 1V between two sets of fingers, and then calculate the surface integral of one of them, multiplying with permitting, finally divided by voltage (1V) I should get the capacitance.
I am however getting a value larger than the accepted value by many other papers; I realized that I might be integrating over field that is not contributing to the capacitance.

May I know if there is a more straight forward way to find the capacitance?
I am a bit desperate as this has really spent me a long time.....

Many thanks!

 

[volker@muehlhaus]

Have you tried modelling as line (equal to the physical length) with series C?

I would simulate two lines of half the ICAP length, and use these S-parameters for de-embedding the line length from the ICAP data (cascade inverse of half length each from left and right side). Then, the remaining response should be the capacitance you are looking for.

 

[sandbo]

Hello,

Thanks a lot for your reply,
I was able to de-embed using an internal function which can be set at the excitation.

Lately I am following another HFSS simulation regarding a parallel plate capacitor, which I think might be a better starting point.
They were using just 2 pieces of metal sheets, then a lump port at the middle.
The key seems to be they tried instead to converge on a defined capacitance, which I think could be making a big difference on the end result.

I am doing the same, except using IDC model sitting at the middle of a 50Ohm transmission line, but on interpolating the frequency, I found a strong dependence of capacitance on frequency,
is that normal? I saw the same on the parallel plate cap, and I heard that fringing field is much stronger with higher frequency, which may explain a higher capacitance.

 

[sandbo]

Have you tried modelling as line (equal to the physical length) with series C?

I would simulate two lines of half the ICAP length, and use these S-parameters for de-embedding the line length from the ICAP data (cascade inverse of half length each from left and right side). Then, the remaining response should be the capacitance you are looking for.

Hello again,

May I know more about the method you are describing?
I was originally confusing it with de-embedding the two feeding lines in series with the ICAP,
however a later revisit to this thread, I realized that you were talking about CPW line with length of ICAP, may I know how exactly it should be like?

Say, for an ICAP of length 100um with feeding line 50um on each end, should I create a CPW waveguide with length 150um with the same width (keeping it to be 50Ohm), then obtain the ABCD matrix,
finally de-embedding it by inverse cascading ABCD matrices as (Line^-1)*(ICAP)*(Line^-1)?

Many thanks.

 

[volker@muehlhaus]

I was originally confusing it with de-embedding the two feeding lines in series with the ICAP,
however a later revisit to this thread, I realized that you were talking about CPW line with length of ICAP

No, this is not what I meant.

Above you wrote "I found a strong dependence of capacitance on frequency" and that is what I tried to explain.

1. For simplicity, I am discussing microstrip ICAP here, not CPW.
Picture here: https://awrcorp.com/download/faq/english/docs/Elements/images/micap_fig1.png

2. The ICAP has capacitance, but it also has some length. This length causes series inductance. This inductance is in series with the capacitance and causes an effective capacitance that changes with frequency.

3. That effective, frequency dependent capacitance is what we measure between the ICAP terminals in series configuration.

4. To calculate the intrinsic capacitance (without the inductance effect), we need to remove the series inductance from results. That is why I suggested the remove half of the length on each side, by de-embedding that length as a line.

Say, for an ICAP of length 100um with feeding line 50um on each end, should I create a CPW waveguide with length 150um with the same width (keeping it to be 50Ohm), then obtain the ABCD matrix,
finally de-embedding it by inverse cascading ABCD matrices as (Line^-1)*(ICAP)*(Line^-1)?

Yes, you would remove all feedline also. But as said, to get the "internal" capacitance you would also remove the length of the ICAP itself.