what is p parameter cst
Another of my rather silly doubts. Suppoase you have a two port microwave device. How does CST MWS or HFSS get the S parameters. Definition of S parameters require matching of ports by characteristic impedances. How do these softwares compute the S parameters ? Does it just give the value in terms of the square root of the power ratio at the two ports or is it something else ? For example if I consider S21 I want to know if I am actually getting S21 or Gamma ( complex propagation constant which includes attenuation alpha and propagation constant beta) times the distance between the ports. Thanks for your time
-svarun
HFSS calculates the fields using the finite element method. it automatically calculates the impedance of the modelled structure at the points where the ports are placed. s-parameters are then normalized to that impedance per default. if you manually change the port impedances the s-parameters are normalized to that impedance.
does that answer your question?
D.
Short comment to Mr. D. answer :
At last version of HFSS v9.2.1 (or previous v9.x.x) the output S-matrix is ALWAYS NORMALIZED to 50 ohm (or whatever you defined) (sic!)
To obtain S-matrix normalized only to port impeadance (so-called "without normalization") I export it first to interim HFSS6+ "szg" format,
then import "szg"-file to HFSS v8.5 and export again to usual Touchstone format.
It's a stupid way, but I don't now another...
you can renormalize your S parameters that works well as Mr. D said
Svarun,
It cannot be the propagation constant times the distance or anything of that type (actually exp[-βl] ) because that would be true only for a straight portion of a waveguide/transmission line with nothing inside, while S21 is calculated for whatever you have between the ports ? like for example a filter.
It cannot be the square root of the power ratio at the ports because that would work only for single modes ports.
Yes, it is something else ?
A time domain solver like MWS actually calculates the wave that propagates inside, takes the fourier transform of the input and output field distribution at the port levels, then for each frequency decomposes those into port mode components and S21 is the ratio of the output to the input for each mode component (and whatever impedance normalization needed).
If you are worried about the port matching by characteristic impendance that is done inside without an actual impedance involved, the structure at the port includes an absorbing boundary, sometimes pml. If it absorbs most of the waves that come from inside then nothing reflects back (from the actual port) and thus it is a matched port for all frequencies and all modes ? if you change the normalization impedance it is a simple math done after the simulation without needing to repeat it.
Note that the port modes (and the complex prop. constants) are calculated separately at the beginning of the simulation by a separate 2D eigenmode solver. These characterize the waveguides assumed to be connected there to the exterior and not the structure inside.
A frequency domain solver does something similar - minus the fourier transform.
Hi,
On the same topic, could someone help to clarify: In HFSS, I have only S-parameters (gain S21 for example) if I defined 2 lumped-port. How can I convert this S21 into voltage ? is sqrt(S21 / 50 Ohm) = voltage ?
Thank you so much for this help.
The magnitudes of incident waves in HFSS are normalized to a field carrying 1 (one) watt of power.
So, the impedance calculated from values of power (P) and voltage (V):
Zpv= (V^2)/P and P = [S12]^2
---> V = sqrt([S12]^2 * Z)
Thank you navuho.
Also, if I take the square root of S21, does it give me the field ?
Thank you so much.
hi,
What MR. D said is correct.
thanx
Added after 1 minutes:
thanks navuho...i didnt even think abt this