Owen Duffy wrote:
Dave Oldridge wrote in
9:


Near as I could measure it, the NF of the receiver after my mod was
1.2db. I had to resort to boiling and freezing water and a tiny dummy
load to measure it at all.


I haven't tried hot/cold tests using ice and boiling water, I didn't
think it was practical.

You finally measured a receiver noise temperature of 50K with hot and
cold loads of 270 and 370.

That means a Y factor of 1.059dB.

If Y were just 0.1dB greater, NF would be 0.78dB, 0.1dB lower and, NF
would be 1.66dB.

With this configuration the sensitivity of NF to changes in Y are
extreme, 0.4dB change in NF per 0.1dB change in Y around that point.

If you made the Y measurements using the audio output of a narrow band
receiver, it is very hard to make high resolution measurements (eg to
0.01dB resolution) with say, a multimeter.

I have done these tests with a liquid nitrogen cooled load and room
temperature load, and that gives more practical Y ratios, 3.7dB for a
1.2dBNF, and the sensitivity in NF is 0.08dB per 0.1dB change in Y. This
still demands high resolution measurement of noise power.

Owen

Indeed, this would be a very challenging measurement, because you also
have to take into account the match of that load, and if it's just a
resistor that you're plunging into hot and cold, its resistance will
almost certainly change. At microwave frequencies, a more common
technique for radiometers is to use a flat plate absorber that has been
characterized for changes in absorption over temperature.

One might want to take a look at how NIST does this kind of thing.
Here's the slides from a talk by Jim Randa
http://www.boulder.nist.gov/div818/8...t%20Crs_06.pdf

he's a noise measurement guru at NIST.. check out the website:

http://www.boulder.nist.gov/div818/81801/Noise/



I've had a precision noise source (used to do Y factor measurements on a
precision 13.402 GHz receiver) measured in their lab over a week. The
measurement uncertainty (for a system with waveguide connections) was in
the few Kelvins range (out of a noise power of 7000K or so), and the
connect/reconnect uncertainty dominates. I doubt one could get this kind
of performance with a coaxial connector (the uncertainty in the mismatch).


By the way, a good noise diode source is probably a better standard for
the hot side than heating a resistor. They're very, very stable over
time, once calibrated, and if properly designed, have a very stable
match as the noise is turned on and off. (that's what we were using in
the above system, a temperature controlled Noise/Com style source).

http://www.boulder.nist.gov/div818/8...ability_IM.pdf
http://www.boulder.nist.gov/div818/8...ility_CPEM.pdf

describes the performance


Another useful link might be:
http://www.boulder.nist.gov/div818/8...97_Amps_IM.pdf
D.F. Wait, J. Randa, "Amplifier Noise Measurements at NIST", IEEE Trans
on Inst. and Meas., v.46, n.2, Apr 1997

They give measurement uncertainties of 0.04dB on a 0.5 dB NF for 2-4 GHz..