S-parameter measurements

I have been trying to measure the S-parameters of GaN HEMTs and LDMOS. but some reason I haven't been able to get expected values. I am not really sure of the reason. For example: in cold FET or below pinch-off measurement(Vgs<below pinch-off and Vds=0V), S11 parameter should be purely capacitive and if converted into Y-parameter, then Y11 should be linear. but after de-embedding, the S11 i get goes into the upper half of the smith chart and corresponding Y11 is very strange (kind of zig-zag) shape. is it because of the fixture (i designed it), calibration (i use mechanical SOTL calibration kit). does the s-parameter measurement depends on the network analyzer? If anyone here has done this before and help me out, then I would be really grateful.

Watch your drive level if it is a small signal device. Most network analyzers default to 0 dbm output.

You did not state frequency. At higher freq package inductance comes into play.

Thanks a lot for your reply.

The device is an unmatched and packaged device. The gain given in the datasheet is small signal gain. So I believe it is a small signal device. The RF input power for the transistor (or the output from network analyser) is indeed 0dBm. Do I need any higher value for S-parameter measurement?

The devices operate up 6GHz or 12GHz, but the network analyser I use can go from 300kHz to 3GHz. When I take the measurements, I usually sweep for the whole range of the network analyser. Do you have to specify a certain range of frequency while doing this kind of measurement? and if yes, could you please explain a bit?

(I am very new to RF)

Have you checked for self oscillations using a spectrum analyzer?

0 dBm is 0.77 V into a 50 Ohm load, and up to 1.54 V into a high impedance load. You should measure with less power.

actually I haven't. I am completely new to RF, specially have no idea about how to do s-parameter measurement successfully. If the transistor does have self oscillations, then is there any way to avoid it?

- - - Updated - - -

thanks. I will try this out.

The best way to remove oscillation is to insert a resister (2 ohm probably will work depends upon transistor) in the base/ gate of the transistor... it will make it stable... on the cost of a very low power dissipation