I am currently looking at performing de-embedding using open and short test structures. i try out this method in ADS (using microstrip element) but found that the de-embedded results shows significant difference from the actual results. Can anyone explain the reason for this. Attached are ADS files and ppt slides of my design.
I think you are not considering the open end and short end effects contained in MLOC and MLSC models. Check in ADS help for these parts.
In fact your deembed work for low freqs, where these effects are negligible.
Try to use normal MLIN instead of these two blocks.
I hope it can help.
Furthermore do you know that exist in ADS a block (Deembed2) that is designed exactly to avoid the construction of all these equations?
Mazz
Hello
The short-open calibration You applied is working if there is no intrinsic parasitics.
I have appload the example.
Your test ckt (ADS_deembeding.dsn) includes two parasitic tline
Tline is Cshunt+Lser+Cshunt (simplified).
So Your test ckt includs intrinsic parasitics and method is not applicable at high freq