RF Metrology

Greetings, gentlemen,

I bought one of those TinySA Ultras recently and have spent some time
testing it today. I have to say it's amazing what they've done for the
price. I mean, truly amazing. I do have two other 'proper' lab-grade
RF spectrum analyzers, but I think I'm going to be mostly using the
TinySA in future as it's just *so* convenient and doesn't weigh a ton.
Anyway, to get back to the point of this post, having checked out the
TSA and establishing I didn't buy one of the fake versions that are
out there, I turned my attention to my mid-level RF SA, an HP who's
model number escapes me (not the 8566B I've posted here about before
which is now fully working, but a newer model that's about 1/3 as
heavy. I say I can't recall the model number but it's not relevant to
this question anyway. "So what is the fucking question, CD??" I hear
you not unreasonably cry. Well, it's this:
When I'm feeding an RF signal into the SA, I'm seeing differing
amplitudes at different frequencies. So I've programmed in a sweep
from 10Mhz to 5.4Ghz at -49dBm using my Aeroflex RF signal generator
and I'm seeing the displayed amplitude vary as it sweeps through the
range. But this only happens when I'm using a cheap, Chinese N-type to
SMA adaptor at the signal generator output. Would I be right to
suspect some imperfection in the manufacturing of the adaptor could
cause such an effect? I do have a VNA I could characterize the adaptor
with but it's a bit of an effort to do. It would seem like the SA is
showing the adaptor's shortcomings in the frequency domain. But is
that a feasible hypothesis?

Your pal,

CD.