This morning, I measured and modelled various ferrite types using a VNA and the ideal ferrite should produce a decent amount of resistance across a wide bandwidth. I tried the best ferrite bead I could find in my ferrites box and the result is as below when I use the HP 478A to level the 83752A sweeper. This represents the uncorrected flatness of this 478A sensor from 40-500 MHz and this is now much better than the same test earlier in the thread. There is still a slight uptick in level at 40 MHz but by 50 MHz the response is as flat as I can measure in terms of uncertainty etc.?
?
I've learned a lot about how to model the 478A sensor in the last 24 hours and what affects the efficiency and what caused the previous blip in efficiency at 50 MHz. I suspect that the later 478A sensors will be improved compared to the earlier ones in this respect. Also, I suspect that the special option H7x sensors will show degraded efficiency once beyond VHF compared to the standard 478A sensor. This is because of the changes made to the coupling and AC grounding capacitors and I tried modelling this.
?
When tested against my Anritsu 1mW reference at 50 MHz using DC substitution the subtle dip in efficiency at 50 MHz is now gone and is now up at 99.4% (was 97.975%). This is now essentially the same efficiency at 100 MHz through to 500 MHz.
?
The blip in the return loss at 50 MHz is also gone when tested on the VNA.? So I wonder if the ferrite bead is a factory mod to later builds of the HP 478A sensor? Adding the ferrite has hugely improved the flatness of my 478A in the region from 40 MHz to about 70 MHz.?
?
?
