One of the Kelvin clips on my Agilent 16089C test fixture just snapped and I'll be replacing the Kelvin clips with probes connected via very small coax cable. My configuration will be a four terminal-to-two terminal setup. Is it best to connect Lcurr and Hcurr to the center conductors of the coax wire and Lpot and Hpot to the shields or vice versa? Or instead of going from four pairs to two, should I just solder all four wires as close as possible to my probe tips so that all four terminals are shielded up to the probe tips? This seems ideal, but the I'm wondering why the 16089C came from the factory with an in-line adapter box that converts from four terminals to two if it is so critical. Why didn't Agilent extend all four wires up to the Kelvin clips?

 

FYI, the LCR meter's signal generator only goes up to 100 kHz so signal loss isn't a concern. The concern is ambient noise coupling to the shield of the coax thereby inducing error in the measurement.

Update: I connected the new probes just like the original ones and everything appears to be fine. Now I just need to determine why the inductance is changing with frequency and signal amplitude. LCR meters are frustrating!