Return Loss calibration

[krishk24]

Hello

I have S11 of DUT + Trace (Return Loss) and I have S11  and S21 (Trace Loss) of just the trace 

If I want to cal out the trace from my Return Loss for DUT, how would I go about this ? ABCD parameters are inherently designed for 2 port networks.

Would it be ok if I take magnitude of just Trace Loss and deduct it from the Return Loss of "DUT + Trace" ?

thanks

[Andy_Hinde]

Hi Krishk24,

I'm afraid I don't understand your question. You are referencing S11 and return loss like they are unrelated quantities, where return loss is the log magnitude of S11. I'm not sure how the reference to ABCD parameters fits in either.

In general, your question seems to imply you are using a vector network analyzer (VNA). Using a VNA implies vector error correction/calibration, and if you are already acquiring vector error corrected S parameters (displaying LogMag S11 as return loss, LogMag S21 as insertion lss), there is no additional calibration you should be considering.

If you are making scalar return loss measurements, say with power meters and discrerte directional couplers, you don't have phase information, and thus cannot know how the coupler's leakage is combining with the reflected waveform, and so must therefore assume complete constructive interference, which is the worst-case uncertainty condition. You can't remove the directivity of the discrete coupler as a systematic error source without phase data, which is why a VNA can but a scalar-only return loss measurement cannot.

Regards,

Andy Hinde

Senior Systems Engineer

RF and Communications

National Instruments

[krishk24]

Sorry if my question was not too clear.

i understand RL is log Mag of S11.

The reference to ABCD parameters comes in because there are 2 separate networks that are in cascade - Trace + DUT

So I know the cascaded response of the Trace + DUT (S11) in form of S parameter and I know the S parameter (S21,S11) of just the Trace itself, then response of the DUT itself can be calculated. ABCD came into the picture since A,B,C,D can be multiplied to calculate cascaded response - in this case I am expecting to perform matrix division instead of multiplication of ABCD.

But in our case we are only concerned about S11

My questions is very simple -

I have S11 (RL) of Trace + DUT. I have S21 of Trace only (Trace Loss). Can I deduct the S21 of Trace (Trace being reciprocal - S12 = S21) from S11 of DUT+Trace to cal out the Trace loss magnitude ? or would i have to perform some fancy ABCD conversion before deducing this ?

[Andy_Hinde · Accepted solution]

What you are describing is a process called de-embedding. You have two cascaded networks and are trying to isolate the response of one network from the aggregate data od the combined network.

De-embedding is quite simple, you simply convert the S parameters to T parameters, and multiply by the inverse of the network which you wish to de-embed. 

Here is a link to a tool on ni.com which can be used to perform de-embedding:

https://decibel.ni.com/content/docs/DOC-29840

To answer your question directly, you can't do simple subtraction in this case to isolate the return loss parameer in question. There are vector interactions with input and output impedance mismatches where the phases dictate how waves combine and complex de-embedding is the process used to accomplish what you are describing.

Regards,

Andy Hinde

Senior Systems Engineer

RF and Communications

National Instruments