[HFSS] Simulation of S21 parameter of CPW line

[mys ruby]

Hello,

I would like to simulate a simple CPW line on a large band of frequency 0-20 GHz. I have created the excitation waveport and the boundary as the same as the CPW tutorial.
https://obrazki.elektroda.pl/8201766200_1432220823.jpg
The E-field on waveport is ok.
https://obrazki.elektroda.pl/5336836600_1432220824.jpg
The S11 parameter have a good agreement between the measurement and the simulation.
https://obrazki.elektroda.pl/7835222900_1432220825.jpg
But with S21 parameter, the level of S21 simulation is inferior to the S21 measurement (while S21 simulation level must be superior to measurement because of connector losses in calibration OSLT). S21 Momentum simulation is very good even though it doesn't include metal losses. (Measurement: red, simulation HFSS: blue, simulation Momentum: green)
https://obrazki.elektroda.pl/3860663900_1432220826.jpg
I would like to find my errors in HFSS simulation, could you help me, please. Thanks you very much.

 

[volker@muehlhaus]

simulation level must be superior to measurement because of connector losses in calibration OSLT

From experience, I do not agree with this part. Contact resistance exists in calibration standards and in the device under test (DUT). It can be the same for CAL and DUT, or larger in the DUT resulting in too high measured DUT resistance, or larger in the CAL resulting in too small measured DT resistance. You really don't know.

 

[mys ruby]

From experience, I do not agree with this part. Contact resistance exists in calibration standards and in the device under test (DUT). It can be the same for CAL and DUT, or larger in the DUT resulting in too high measured DUT resistance, or larger in the CAL resulting in too small measured DT resistance. You really don't know.

Dear volker@muehlhaus,

Could you explain in detail, I really don't know this part. I have measured with an OSLT calibration that includes the connectors losses due to the transitions between the input-output coaxial cables and the cpw line. I think S21 simulation level must be slightly superior to measurement even if I can't estimate exactly these connectors losses and the S21 Momentum simulation seems more reasonable than HFSS. I don't know where are my errors in HFSS.

 

[volker@muehlhaus]

I have measured with an OSLT calibration that includes the connectors losses due to the transitions between the input-output coaxial cables and the cpw line. I think S21 simulation level must be slightly superior to measurement even if I can't estimate exactly these connectors losses and the S21 Momentum simulation seems more reasonable than HFSS. I don't know where are my errors in HFSS.

Ok, if you have calibrated with coax cal standards you are right. I expected that you did wafer prober measurement with coplanar cal standards.

For HFSS, you should describe your model in more detail. How did you model the conductors (surface impedance or solve inside)?

 

[mys ruby]

I have assigned the conductors (Copper adhesive tape) as Finite Conductivity Boundary with the conductivity 2e7 S/m. The substrate is synthetic paper and there is a adhesive layer between the substrate and the conductors. The waveport is renormalized to 50 ohm.

 

[volker@muehlhaus]

conductivity 2e7 S/m.

2e7 S/m is very pessimistic for copper. The usual value is 5.8e7 S/m.

 

[mys ruby]

2e7 S/m is very pessimistic for copper. The usual value is 5.8e7 S/m.

I have tried with the usual value 5.8e7 S/m: The level is a little superior, but it is also inferior to the measurement. I don't think it is a principal reason.