开云体育

Thanks Roger for pointing this out. Knowing one's measurement tools is the
basic rule here. Thanks for pointing out.
I have also checked my NanoVNA.
If I measure the Zload of port 2 (using Port 1, so using a calibration
Through connection and measuring port2's S11). I get the following |Zload|
of port 2:
[image: afbeelding.png]
This is ok-ish IMHO (variation some 0.1ohm (less than 0.2%). There is a
deviation from 50ohm (around 05%), but this could be partly due to
inaccuracy calibration Load (measured with DVM: 49.7ohm).
I don't think the Zload (|Z| of port2/load) is problematic.
If I measure metal resistors (averaged between 300 and 1800kHz) using
Reflection and Series S21 I method, I get the below |Z| values. I also give
the R values measured with DVM

[image: afbeelding.png]

[image: afbeelding.png]
So indeed |Zrefl| does not deviate much from the DVM value (<2%). The
|ZseriesS21| though differs quite a lot (~9%) from DVM value and looks to
be proportional to the resistor value.

Where is this 9% error coming from? A 9% error is large! I don't think the
deviation of the port2 load (Zload) from 50ohm is problematic (only some
0.2%). Looking at the Zsourse (the impedance of port 1): would that not be
close to Zload (impedance of port2, I hope/assume...). But that one is
compensated using the SOLTI method, or not?
I understand the missing of 12-term correction is the other problem (as
stated by Owen Duffy: ). Who has a
procedure for this? I could not find this on the web (Owen Duffy refers to
it, but I don't see a procedure).
Something must deviate some 9% from the 'my norm', but what? I tried to
simulate this by varying Zo, Zload, Zsource; but I can't find really
realistic values to cater for the variation see in ZseriesS21 (I need to
bring Zload back to some 40ohm to cater for this 9%, but then some other Z
values (|ZseriesS11|) don't match properly [and it is not measured, see
above!];-).

Any help to understand this is appreciated. Thanks.
So at this moment I agree that 2port Series method does not produce good
values.

All the best,

Victor


Op di 31 mei 2022 om 23:07 schreef Victor Reijs via groups.io
<pe1atn.victor.reijs@...>: