[Test-Equipment] Need info on Noise figure meter

[Steve Harrison]

At 05:31 AM 12/28/2007 GMT, k4pf at juno.com wrote:
>
>Hi, 
>
>The ARRL 1965 Handbook, page 526, featured a home constructed 
>noise generator head with a 5722 tube.  
>Perhaps that's what the home-brewer based his design on.  
>
>Someone said gas tube noise generators are problematic when testing
>modern devices, but the 5722 is not gas-filled.  It is a vacuum tube diode
with a cathode probably of uncoated tungsten, 
>shades of 1920's technology.

That's entirely true; 5722 noise sources do NOT generate the killing spikes
that the higher frequency gas-filled tubes do. They're perfectly safe to
use with solid-state LNAs. The main problem is they only generate a little
more than 5 dB ENR of noise, and still present the impedance match problem
when they're turned off. This is reduced considerably by the series 50-ohm
resistor from the plate to the output blocking cap but then, you can
encounter stray resonances.

W6GGV and I built half a dozen 5722 sources and attempted to characterize
each of them,. We gave up when we determined we could not repeatably
measure to better than +/- 0.1 dB accuracy between each.

A 5722 is a great noise source whose noise contribution, at 220 or lower,
you can easily calculate presuming you've really built it using great
UHF-construction techniques. It's also very useful at 432 in a relative
application. But even N6CA gave up on a 5722 as any sort of calibratable
noise source. And as far as I'm concerned, if N6CA says it can't be done,
then even a pronoucement from God (otherwise known as W6XJ) would be suspect.

But a eally-well-built W1GHz solid-state homebrew noise source will easily
be flatter up through at least 2304, we found. The problem was we could
never actually charactize the actual ENR of the solid state sources without
referencing them to the calculatable 5722, at lower frequencies, and thus
assuming our construction technique held through the lower microwave bands.

Steve, K0XP