Hi Andy -
am I to understand that it's worth trying to
calibrate with a load other than 50R
then doing some simple back tracking calcs ?
The W0QE paper indicates that, seemingly because of errors introduced
by parabolic curve fitting in AIM4170 software,
calibrating with higher impedance yielded results that he liked better.
Some issues are:
* At least for the nanoVNA resistance bridge, sensitivity drops away from 50 Ohms.
Sensitivity is related to curve slope steepness here:
...so e.g. errors and noise have more impact.
Since a reflection bridge can use a reference load higher than 50 Ohms,
it might work better for higher impedances, but requires correcting and converting nanoVNA S21 values.
* Where to find / how to verify high frequency impedance calibration references?
With good high impedance calibration references, a 9:1 transformer may be characterized and used.
As Kurt pointed out, W0QE S21 result depends on ideal (or at least precisely known) CH0 and CH1 impedances.
Oristo