Hi Brian,
Thanks big time for sending me the .s2p renormalized file. As a reminder to everyone your renormalized curve used my original forward and backward measurements of my ceramic filter using a 10 pf connected load and the measurement was made with a direct connection to my NanoVNA-F (NanoVNA-F calibrated using 50 ohm loads, and no external matching devices used).
My 10.7 MHz ceramic filter (SFE10.7MA5-A) has a published input and output impedance of 330 ohms.
Attached are 3 files comparing your renormalized curve using my forward and reverse measurement data versus 3 different matching methods that I used to measure the same ceramic filter with the same 10 pf load (all shown connected in the forward connected direction since their impedance matched forward and reverse connected measurements fall almost directly on top of each other). The only difference between the 3 attached files is that I used different scaling on the vertical and horizontal axis to provide some zoomed in and zoomed out views of the frequency response curves.
In the attached plots the curves are identified with the following colors:
Red = your renormalized curve
Blue = 50 to 330 ohm resistive L matching curve
Green = 280 ohm series resistor matching curve
Orange = 2 turn primary, 5 turn secondary BN-73-202 binocular core matching transformer curve (not a perfect match, provided more like 280 ohms whereas it should have been 330 ohms).
I hate to provide a firm opinion but let me make some quick observations to help folks understand what they are looking at as follows:
The renormalization process did a great job fixing my direct measurement curves to form a more realistic frequency response curve of my filter but it was not perfect if you zoom in and compare the very top of the renormalized curve with the curves I obtained using two different resistive matching techniques. The top of the renormalized curve has a premature dip and then a bump. Also if you look way down in the rejection area of the curves you will see the renormalized curve does not agree with my 3 other matching technique curves, especially in the lower frequency direction.
While my matching transformer provided an impedance that did not exactly match the published input and output impedance of the ceramic filter it did a decent job and it provided much better dynamic range which helped when looking at the filter where it provided a lot of rejection (way down in the mud so to speak). I could have added a half turn on my transformers to provide a slightly better match but had to cut my time consumption on this exercise. Not having a perfect match using the transformer appears to have had some slight impact on the shape of the top of the curve as well as a slight downward shift in the frequency response when you look at the zoomed in curve.
At some point it would be interesting to see how your renormalization routine compares with that on the NanoVNAs that have that function but unfortunately my NanoVNA-F does not have that function. Hopefully the work we have done will provide some insight for others regarding what to look for if they want to better evaluate the renormalization routine on the NanoVNAs that have that function. And there also is the big caveat regarding renormalization of nonlinear devices that folks need to be cognizant of.
Thanks again for all your help.
Don
