Voltage is proportional to the product of resistance and absolute
temperature.  As an experiment place a voltage across a high value
resister like say one 1M  raise the volts until you are near the limit of
the resister and connect it via a coupler cadaster to an audio amplifier.
You will hear white noise in the speakers.

There is a similar effect in semi conductors.  The best example of this is
visual noise in digital camera, when you tern the gain way up (set the ISO
high) you can see it in the photo.

All of this is proportion; to absolute.

As I remember we ran the "oven" at -20C There was a valve used to flash out
the air inside with inert welding gas to reduce ice.

Some people use vacuum and get to cryogenic temperatures.  But that is
expensive and way-hard without an institutional size budget.

--

Chris Albertson
Redondo Beach, California

On Sun, 4 Jun 2017 22:01:45 -0700
Chris Albertson albertson.chris@gmail.com wrote:

Yes, Johnson noise increases by about 20% (or about 0.8dB)
when going from 25°C to 90°C. But Johnson noise is usually
not the first limiting noise one runs into when designing
a crystal oscillator. I would rather use a simpler,
"high temperature" oven and invest the time saved into a
desgining better oscillator structure.

If the -20°C was used on the camera, then this is not to reduce Johnson
noise, but the dark current noise. While the former is due to thermal
vibration of the atom lattice, the later is due to spontaneous forming
of electron-hole pairs in the p-n junction region. Thus also their
temperature characteristics differ: While Johnson noise is linear
in temperature, dark current noise is (almost) exponential in temperature.
Going down from room temperature to 0°C is something like a factor 100
(IIRC, from memory, could be wrong) in noise for a CCD. While for
a resistor, the difference is almost negligible.

		Attila Kinali

--
You know, the very powerful and the very stupid have one thing in common.
They don't alters their views to fit the facts, they alter the facts to
fit the views, which can be uncomfortable if you happen to be one of the
facts that needs altering.  -- The Doctor

On 6/4/17 10:01 PM, Chris Albertson wrote:

As in Johnson noise - vrms = sqrt(4kTR)

I believe that's a different mechanism, shot noise.

cryogenic (as in LN2 temps) is very hard with TEC it takes many stacked
devices that wind up looking like a ziggurat.

near vacuum isn't that tough at home - refrigeration vacuum pumps are
inexpensive.  If it's a one time thing, a sorption pump and one time Ln2
or Dry Ice can get you pretty empty as a one shot. Then you seal it off
and it's done.
the Bell Jar - http://www.belljar.net/  is a handy reference of vacuum
hacking stuff

Yes,  Agree.  I wrote tin the first post that "for your use a resistive
heater would be better"

But everything else, I'd do over.  Drill the aluminum block and use thermal
epoxy to hold the sensor in place.  Use a vacuum insulated mug or bowl for
a cover and let a micro controller run a PID loop to control temperature..
You can keep it within 0.1C without the need to be a control theory
expert.    Another order of magnitude would really hard and expensive

The last temperature controller I made was not as good but keeps a rubidium
oscillator within about a degree.  I put a thruster in the Rb's heat sink
and a micro controller adjusts the fan speed.  The self heat from the Rb
unit is enough.    The Rb is about the size and shape of an old 3.5 inch
disk drive so I put it n an old disk drive enclosure that already had a
place for a fan.    As crude as this is, it is much better then what I had
which was not control.  The Rb output is better then I have means to
measure now.

So same with the poor man's oven.  It can be very good with very little
effort.    What has happened is that now we have nearly free micro
controllers and can use daily sophisticated control algorithms that HP
could not use back in the day.    We can actually model non linearities in
the sensor.  We can use a full on PID solution rather then the simple
on/off thermostat.

But no you don't want to cool a crystal, I was not suggesting that.  We
happened to have a high gain amplifier that was in effect counting
electrons.

On Mon, Jun 5, 2017 at 2:15 AM, Attila Kinali attila@kinali.ch wrote:

--

Chris Albertson
Redondo Beach, California