Re: A Simple Lab Demo of Resistor Noise EMF.

["John S. Denker" <jsd@MONMOUTH.COM>]

At 09:17 AM 3/19/01 -0600, Gary Karshner wrote:
> You might notice the low input impedances of the amplifiers you listed.
> This is of course to limit the effects of thermal noise.

Well, let's be careful here.

Thermal noise is a certain amount of *power* per unit bandwidth, and you
can't change that by fooling with the impedance.  To say the same thing in
another way, there is a noise *current* as well as a noise voltage.  For a
high-impedance source, the noise looks "mostly" like a noise voltage, but
for a low-impedance source you have to deal with the noise current.

If your preamp input impedance is too low relative to your source
impedance, you will "short out" useful gain at least as fast as you "short
out" the noise voltage.

You can often match the source impedance to the instrument impedance using
a transformer.

I've always thought that measuring noise in nV per root Hz unduly
over-emphasized the noise voltage at the expense of the noise current.  The
professionals talk about noise figure or noise temperature instead.

For the example instrument I mentioned, described at
  http://www.thinksrs.com/html/sr560technote.html

the noise figure bottoms out for source impedance somewhere around 300 kOhm
-- which is not what I would call an unreasonably low impedance.


> They do have impressive noise figures.

Yup.