I've deliberately chosen not to rely on scikit-rf in NanoVNA-Saver. Library
reliance is kept at a minimum, and figuring out the mathematics behind the
calibration was part of the fun of writing the software. :-)
I might implement 12-term calibration at some point, but it's not high on
my list. I'm still not sure it would be very useful. :-)
--
Rune / 5Q5R
On Mon, 9 Dec 2019 at 09:24, Gabriel Tenma White <OwOwOwOwO123@...>
wrote:
This might be unintuitive at first but long story short, a T/R VNA *can*
correct for a mismatched port 2 but *only if it knows the DUT's S22 and
S12*, which is why to do two port one path/enhanced response you must
either measure the DUT in both directions, or the DUT must be reciprocal,
even if you only care about S11 and S21.
More detailed explanation: When you measured the thru standard, you
already discovered the reflection coefficient of VNA port 2 (plus its
cable). You know S21 of the DUT, therefore you know how much power is
entering VNA port 2 and how much is reflected back into the DUT. However,
to know how the outgoing power from VNA port 2 affects the S11 (and other)
measurements, it is necessary to know in what way the DUT will scatter
power incident on DUT port 2 to port 1, etc. Therefore a reversed DUT
measurement is necessary. The math is pretty simple and it's just solving a
linear system of equations as usual.
TL;DR bug the nanovna-saver developer to implement two port one path using
scikit-rf, then do the SOLT calibration as usual, measure DUT forward and
reversed, and enjoy a 12-term full two port VNA ;)
