[time-nuts] Frequency referenced temperature regulator

[jimlux]

Poul-Henning Kamp wrote:
> I'm contemplating building a small temperature control enclosure for
> testing various electronics.
> 
> I have a handful of peltiers suitable for the purpose, and was
> pondering the right control mechanism.
> 
> Most people would reach for a NTC, put it in a wien-brige etc etc.
> 
> But since I happen to have access to much more stable frequencies
> than voltages, I thought of a different way:
> 
> 1. Mount a X-tal-osc with really lousy tempco inside the enclosure.
> 
> 2. Compare its output to a stable reference frequency.
> 
> 3. Use the output of the phase comparator to drive the Peltier.
> 
> It is basically a PLL where temperature is used as EFC...
> 
> Has anybody tried that ?
> 
I used to work at a place where we used SAW resonators as sensors for 
just about everything (including temperature, pressure, force, chemical 
presence), and pretty much as you describe.. output of resonator 
affected by what you wanted to measure was mixed with a similar 
resonator which was not, and a  microcontroller would count the 
difference frequency and do what ever else was needed (run control loop, 
apply cal curve, etc.)


There's also the whole MCXO scheme, where you compare the fundamental 
and 3rd overtone, and measure the temperature of the crystal that way 
(in order to adjust the output frequency of a DDS driven by the oscillator)