Calibration of SMC channels using 4-port N5225A, CPM and de-embedding.
We are building a system for automated measurements on "mixers-devices" (each device contain 2 mixer stages each having 2 I/P and 2 O/P) with I/P frequencies up to 45GHz.
We perform a "CalAll" @ Tier-1 points which we can access via transfer switches @ each PNA port. Then we perform Tier-2 calibrations @ each DUT I/P and O/P port.
We tested this calibration and it worked as expected when we did both Tier-1 and Tier-2 calibrations "at the same time" (the hole process toke mayby 4 hours). Looking at the s2p files generated in CPM we see a 15 degree phase shift over 25 MHz which indicates that we have approximately 0.5 m "extra cable" @ Tier-2 compared to Tier-1.
The problem is that when we next day perormed a new CalAll the calibrations "did not match" anymore (ripple occured at VSWR measurements). We want to be able to do CalAll quite often (we also meaure noise figure) and Tier 2 we expect to do every 6 month or so.
Could this be caused by frequency drift in PNA? Could this be improved if we have approximately same electrical length to Tier-1 and Tier-2?