"Jim" wrote in message ...

As other's have said, there are ways to build a calibrated noise source.
I've even seen zener diodes and just plain old switching diodes used.

I use a 1N21 (in reverse breakdown) as an uncalibrated source up to
about 5 GHz. A friend uses a 1N23 to 10 GHz. But how can it be
calibrated without using professional test gear ? I don't know.

What frequency are you using?

Anywhere from HF to 24 GHz is of interest ! I have used the relative
noise figure measurement scheme between 3.5 & 903 MHz so far.

One thing to keep in mind is that any loss or
mismatch can affect your measurement.

Indeed...I assume that the use of a reasonable minimum attenuation in
the step attenuator will minimize the effects of receiver input
mismatch on the measurement accuracy (assuming the attenuator is
itself well matched). I rather doubt that I would be able to homebrew
an accurate measurement system at 24 GHz (or even 10 GHz). But
construction tolerance issues should not be a problem at HF or (with
care) at VHF.

73,
Steve VE3SMA