Hello ,
I imported ODB+ files on ADS to simulate and learn how to optimize future PCB designs. While comparing the simulation results on ADS and the measured S-parameters using the VNA , the simulation gives worse results than the measurement (supposed to be the opposite, right ?). Apart from tha layer stack, the frequency and the simulation options what could be the root cause of such a significant difference ?
Thank you for your help.
Dear Ameny Chaabani ,
I had similar problems with my designs. It is important to note a few things:
1. When measuring and characterizing the PCB material, do you calibrate the connectors out? E.g. with a multi-line TRL calibration?
2. What kind of PCB material are you using? It could be anisotropic (most fiber based substrates are anisotrop e.g FR4)
3. If it is anisotrop do the ADS files or material datasheet handle/mention it? Form what I have seen most data sheets only specify the dielectric constant in z-direction (and not x/y)
Pls keep in mind that you need different setups to measure the dielectric constant if the material is anisotrop. For x and y transmission lines are a good choice, but z- direction a resonator (or similar setup) should be used (There are of course many other methods to characterize materials published in literature).
Best regards,
Martin
I agree with Martin Obermaier ! Calibration can often be a source of error! If you want to check this you can simulated the feed structure to your DUT and then take those s-parameters and place them in a S2P de embedding block to subtract them out and see if your sim matches the measured data better.
https://literature.cdn.keysight.com/litweb/pdf/5989-9451EN.pdf
-Good Luck!
Adam
designing engineers measure first the actual pcb anisotropic permittivity; or at least an effective permittivity for the transmission lines they use. Then comes the metal surface roughness. And don't forget about the board exterior.
Dear Ameny Chaabani
Adding to the colleagues above, one normally finds differences between the simulated and measured results. In fact in order to determine the cause of the differences one has to make more investigations. But one can mention some general causes:
The physical model of the tested structure is an idealized simplification of the real structure. This concerns the geometrical descriptions and thge materials parameters. There are also the tolerances in geometrical and physical parameters of the different materials in the structure. In case of rf circuits there are the parasitic and stray elements that are normally very difficult to fix.
There are also errors associated with the measurement that may be an appreciable percentage of the nominal value.
The most critical parameter in the S-parameters is the resonance condition at high frequencies. The small deviations in the capacitance or the inductance value may shift appreciably the resonance frequency.
Best wishes
If you want an exact or similar results with the measured you need to optimize your parameters in ADS there is a way to do that
- Badges
- Science topic
- Similar topics
- Phytochemistry
- Extracts