Measuring (DIY) active probe performance

[fozzyvis]

I'm trying to make an active probe, based on https://content.instructables.com/ORIG/FQZ/1QZP/IPJTFO82/FQZ1QZPIPJTFO82.pdf . My repo is on https://github.com/dietervansteenwegen/FETprobe, but that does not represent the current state of my prototype as it is a work in progress.
As this is my first time trying something like this, I am somewhat surprised by my measurement results as they seem rather promising. I am wondering if my measurement methods and assumptions (what I should be measuring and interpretation of the results) are correct.

I'd appreciate some remarks/critisism on my thought process and/or measurements.

I first tried a bode plot on my scope, which gave me the following (forgot to make a screenshot, but have this picture I made):



As my scope signal generator is limited to 25MHz, I started up my old HP network analyser which goes to 1.3GHz. I'd be happy if the probe is useable up to 200MHz, which is probably way beyond any use I'll have for it.

My measurement:

• First I normalized the reflection S11 with a short connected to the RF output (resulting in a flat line around 0dB),
• Then I normalized the transmission S21 by connecting the output directly to the input. With a direct cable between the two connectors, I get a flat line on 0dB.
• Then soldered a coax directly to the input of the active probe and connected that to the VNA output. The output of the probe connects to the input of the VNA.

This gave me the following plot:



What I think I see and can conclude:

• The measurement on the scope and VNA seem to match ok (both around -18dB at 10MHz)
• It seems useable from around 10MHz (can make another trace 0.3 -> 10MHz), before that, the gain is too unstable.
• From 10 to, say, 300MHz the gain is fairly stable around -17dB. This makes it useable in this range, as long as I do the conversion to account for the 17dB loss, correct?
• S21 up to 300MHz is reasonable as well, I think. I don't know what kind of numbers I should aim for. Since this is a hobby project I am happy with "decent" and mostly want to learn. I figure that since you want as little loading on your measured signal, a high reflection is what you want, correct? The higher the reflection, the higher the impedance, which is exactly what you aim for, no?
• Below 10MHz the transmission/gain is very low. Not sure how I'd go about improving this.
• It would be easier if I could get the gain to -20dB as this makes the amplitude conversion easier. I can likely doing this by playing with the supply voltage of the FET.

Are my assumptions correct?