Crystal oven theory

Rick Karlquist used to post here from time to time--a very smart guy. (I know him slightly from a side gig some years back, before he retired.)

He also did some interesting ham radio work on direct frequency synthesizers based on cheap 455 kHz ceramic IF filters, and got pretty impressive results.

Cheers

Phil Hobbs

Am 22.05.22 um 03:46 schrieb Phil Hobbs:

Gerhard

Here's my EO modulator oven, with the big top cover removed.

The temp sensors are thermistor wheatstone bridges on the small blue board on the bottom of the modulator, with a 24-bit ADC. The modulator is mounted on spacers inside the main oven block, not very thermally conductive, so thermally it's a second-order system. It's stable to way better than a millikelvin.

Note the SMA feedthrus and long squiggly cables, which reduce cold sneaking into the EOM over the coaxes.

He didn't survey the literature all that well.

Conduction. Foam pretty much stops convection.

People have been managing that since the 1960's. It's not easy, though good inter-changeable thermistors and 20-bit A/D converter chips have made it easier.

Probably more helpful to say that they increase the thermal time constant of modulator whose temperature is being controlled.

Right, thanks. Cool gizmo, anyway.

Cheers

Phil Hobbsn

It was mentioned on another thread that the thermal resistance of radiation between two concentric blackbodies at 300K is about 6mm, interestingly that's very close to the same thickness of foam-filled negative space between the outer and inner shells of the OCXO module on my 5334B.

Don't know if that's a coincidence or not, obviously the two aren't at exactly 300K. The interior I think runs about 355K, while the exterior (I haven't measured it) seems cool enough to keep your hand on a while.

Equivalent to 6mm of air, rather.

In a reasonably close-fitting box, without a lot of space for convection, air is a better thermal insulator than foam or fiberglass.

That depends very much on the thermal emissivity of the surfaces and on the thickness of the gap. The JWST doesn't have to deal with a lot of convection, but its sun shield has many layers on account of the radiation issue.

Cheers

Phil Hobbs

You add foam or fibre-glass to stop air convecting.Both the walls of the foam cells and the glass in the glass-fibre have much higher thermal conductivities than air, but you don't put in enough to short-circuit the air.

Convection does depend on the space available. 6mm isn't much.

Radiation goes as the fourth power of temperature, and the sun is at 6000K. It's not really a relevant example.

Here's a glamour shot of the internals:

Don't know what the foam is. It doesn't seem to conduct heat too good.

My design for a foyer quartz clock back in the early 1960s included a crystal in a thermostatic oven made from Perspex. I have a sketch and isometric of the oven and a note that there were five thermistors in a bridge, but no schematic. I do remember that on first testing the control circuit initially ran at full power heating, then cut off, then gradually settled on steady warmth!

JSTW is dealing with a huge delta-t, and vacuum is a poor thermal conductor. I've experimented with heated gadgets inside a deep-drawn aluminum box, which has a low emissivity. My testing usually showed that I got less heater power dissipation with air, as opposed to styrafoam or fiberglas.

It's not a difficult experiment. I think most people just assume that insulation always helps.

The interesting test would be to measure heater power with and without the foam.

The delta-T is only huge in the first gap. After that it's much much smaller. The temperature of the first shield is near the subsolar temperature, but the amount it radiates into the first gap is much lower than a black body's, on account of the very low emissivity (i. e. high reflectivity) of its surfaces.

The subsolar temperature of a flat plate at a radial distance r from a star of radius a is

T_ss = T_star * sqrt(a/( r sqrt(2) ) ),

i.e. the temperature at which the thermal emission of the star (sigma T_star**4 * a**2) balances that of a notional sphere of unit emissivity (sigma T_ss**4 * r**2). (The factor of sqrt(2) is there because the plate radiates from both sides--for a spherical planet, whose area is 4 pi R**2, it's a full factor of 2.)

The temperature of the solar photosphere is around 6000K, and its radius is 7e5 km. The mean radius of the Earth's orbit is 1.5e8 km. Thus the subsolar temperature for a flat plate is

T_ss = 6000 K * sqrt(7e5 km / 1.5e8 km / sqrt(2)) = 334 K.

(For a spherical object such as the Earth and Moon, it would be sqrt(sqrt(2)) less than that, i.e. 281 K.)

There are two main contributors to the effective thermal resistance of a multilayer stack of spaced reflectors. One is just breaking up the gap into N subgaps, which does what you'd expect--M layers gets you about M times the thermal resistance. (The effect is bigger at larger deltaT, where the emission from inner layers is much less.)

The other contributor is that, since the layers of JWST's heat shield are spaced wide apart, the great majority of the thermal emission of each surface bounces around until it escapes into space. Of course the ambient radiation comes in as well, but since its temperature is very low, that's not a big worry.

Plain air is better than insulation for narrow spaces and small delta-T, and insulation is better otherwise. Just where the break-even point happens depends a lot on the situation.

Cheers

Phil Hobbs

Insulation in the sense of low-conductivity material is NOT the intended function of fiberglass; it is used to reduce air convection, which (thermal convection is delta-T squared heat transfer) is a tiny effect in a low-temperature-gradient system. In a house-on-fire, fiberglass in the walls is not tiny, but a big help in getting out alive.

In a very real sense, fiberglass filled cavities are air-insulated.

On a sunny day (Sun, 22 May 2022 22:51:09 +0100) it happened Mike Coon snipped-for-privacy@mjcoon.plus.com> wrote in snipped-for-privacy@usenet.plus.net>:

Yes that is overshot, maybe over-compensated

This was pretty good too:

temperature within a few ADC steps over many years.. Took about half an hour to get the PID working correctly in PIC asm ..

For the rest the 1 ppm TCXOs as in my RTL-SDR sticks are very good and very small.

Air at room temperature has an alpha of about 0.026 W/m/K. High-density styrofoam (the kind that crunches when you poke it with your finger) is about 0.040, and low-density styrofoam (the kind that squeaks) is around

Cheers

Phil Hobbs

Or else windup, which is pretty likely if the sensors and heaters aren't well coupled.

Cheers

Phil Hobbs