[time-nuts] Airflow around oscillators.

[Poul-Henning Kamp]

In message <42906DB8.80703 at onetel.net>, David Kirkby writes:

>I'm wondering if its best to try to reduce the airflow around the 
>10811-60111 and PRS10 (rubidium) to a minimum necessary to keep the 
>things within temperature,

Yes, still air is best because it has a long timeconstant for changes.

But you also have to consider that there must be enough heat transfer
to keep the kit cool.

>I'm also thinking of using a thermostatically controlled fan, in an 
>attempt to keep the temperature inside the box a constant.

Don't.

You'll just make the temperature jump up and down a lot.

Having a fan on constantly on the other hand may be a good idea, but
it should run as slow as possible to just keep the air from forming
hot pockets (which might result in inversion events).

If you want to actively steer your temperature Peltier elements
with a PID controller is the way to go.

Consider getting and old fridge (don't plug it in), put a slowly
rotating fan at the top to keep the air circulating an put your
kit on the shelves.  You should be able to get a sub-degree temperature
stability environment that way.

The cardinal rule for a steered frequency source is not that the
temperature have to be constant, as long as it changes only very 
very slowly it is no problem.

-- 
Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe    
Never attribute to malice what can adequately be explained by incompetence.