[time-nuts] Ebay FE-5680A Rb: Are they that good?
eb at telight.com
Fri Dec 9 11:51:49 EST 2011
When looking at harmonic distortion of the 10 MHz, also consider the
distortion of the SA itself. I rebuilt my filter made from LAN
components that I described a while back, and achieved about 85 dB
rejection from 30 MHz to 70 MHz, which should have made harmonics of
even a crappy sine wave all but disappear. When I put actual signals
through it, I was shocked to see that it hardly made a difference -
it seems that the SA mixer provided way more distortion than what I
was looking for, making it impossible to tell if the filter worked.
I believe the reason for the discrepancy is that when I designed the
filter I used a tracking generator, so when the SA passed through the
harmonics, there was no large fundamental present at the mixer, and
the true response was shown. When the real signal is present - even a
very clean sine wave, with or without extra filtering - the SA shows
fairly large harmonics. The harmonic levels displayed are also
affected by the mixer input level as expected, but unfortunately,
cranking down the level makes it run out of dynamic range way before
it runs out of distortion.
I gave up trying to measure the filter response to actual signals at
that point, until I review the 8566B specs. I assume that performance
limit was one of the reasons why HP offered a low-band preselector.
More information about the time-nuts