s-parameter measurement and matching reference planes

I am using a microX package infineon MESFET device
for designing an amplifier at Ku band.
I am using s-parameter data provided in the data sheet for matching at operating frequencies.
I am not very sure,how this s-par DATA is extracted by the manufacturer?
Whether it is extracted at lead ends or device body ,as shown in fig1 and fig2.
I want to know ,what is the general practice of choice of reference planes in such measurements?
Though the leads are only 1mm long,this will be enough to vary the phase at 12-15 GHz.
This error in phase will affect the matching .

If some one can help and share the experience in such a matching,I would be very grateful.

Depending upon your advice,I would be choosing the matching network starting at device lead end or device body ,as shown in fig3 and fig4.

Certainly ,this will be taken care while making PCB layout,from which part the matching circuit should start,lead end or device body.

For most of these devives, the ref plane for manufacturer S-parameters is at the ceramic/plastic body.

you can clarify with infineon by emailing them. or can do matching for both cases. you might end up with same topology for both cases with different component values/types. PCB fabrication will not be any problem if its true. you can optimize the amplifier on PCB board based on what the impedance really is!

Whether ICM test jigs measures s-param at lead ends or metal body.

Measuring with S-param reference at the end of the lead makes no sense.

And not only it makes no sense, it would also require different test substrates when the ref plane moves around for different lead length. Ref plane is at the device body, not lead end.

You are right Volker,
But its important to know the reference plane location chosen ,while extracting s-param of a device by the vendor,its not always the device body.
For example ICM test fixture extracts s parameters at lead ends'which makes it substrate dependent and need to be de-embedded to be useful for all substrates.

Sure?

Really sure?

I have used similar fixtures, and the ref plane is where I place it with my TRL calibration standards.

Of course, you can not use coax cal standards for this type of measurements, because then all the transition etc would be included in the results. Proper calibration removes the fixture and feedline.