I can't understand what is done here, and why de-embedding is needed? And how to decide upto what distance inside should we de-embed
I can't read your file, as I don't have HFSS, but I expect de-embedding is used to move the reference plane, from the port that was used to the component being modelled . For example, if the port is a quarter wavelength away from a lumped capacitor, the capacitor would look like an inductor at the port. De-embedding takes out the quarter wavelength and makes it look like a capacitor. You often can't have zero distance between a component and a port, particularly if you want to know what the component looks like at its centre-plane, so de-embedding from a port further away is used. I think that de-embedding can be used to take out other components, too, if you know what they are or what they do to the signal, and is a similar process to the calibration of a vector network analyser (VNA), which takes out the effect of all the VNA components and lines, up to the calibration point.
Thanks for the answer. But I wanted to know in the above design, why not just delete the feed (lambda/4) line and then simulate to extract s11, s21 etc and dispersion diagram? Isn't that correct?
Or just use some PEC PMC condition and get the dispersion diagram?
I want to know in order to get dispersion diagram, when should we use 2 waveport with de-embeeding and when we use PEC PMC boundary?
P.S. I am a newbie to this :)
If the feed line is a quarter wavelength then it may well be a transformer from the 50 ohm port to the impedance at the edge of the antenna, so is a necessary part of the feed network. Is it 50 ohms? If the feed port is 50 ohms and the quarter-wave line isn't, then look up quarter-wave impedance transformers to find out what they do. A quarter-wave transformer like this is often used to feed microstrip patches.
I believe that the intent of this is to model only what that stub and via would look like when placed into the middle of a similar line. The de-embedding exercise is the obtain that result, absent the trace leading up to it.
The point of doing this, I believe, is to have all of the fields be in the arrangement they'd be in when the stub is reached. if the waveports were attached directly to the stub, the assumption would be perfect TEM orientation of the waves right at the point that the stub is encountered, which is not really correct.
In this manner, if the user of the stub model combined the single port s-parameter model obtained with two traces from, lets say, a cross-sectional or BEM solver, the right overall answer is obtained.
I hope I explained this well enough.
- Badges
- Science topic
- Similar topics
- Philosophy
- Philosophy Of Science
- Reasoning