1. Home
  2. Electronics Design
  3. Gas cell heater

Gas cell heater

So I've got this glass cell, a cylinder, 25mm by 25mm, it's got Rb and some Ne buffer gas inside. (The buffer gas stops the Rb atoms from diffusing to fast and hitting the walls.)

It needs to be heated to ~50C, the exact temperature is important, but stability is critcial. There's an exponential change in the Rb density with temperature.

Years ago we designed a heater, A plexiglass cylinder ~ 100mm x 100mm (4" x 4") Then a layer of foam, bifilar heater on a glass cylinder, and some more foam holding the 1" (25mm) cell in the center. I can post a picture tomorrow

The other night I was testing another one, and thinking that we have too much insulation. The time constant is too long. Things would be faster if I used more heat.

At 50C it runs at 12V into 50 ohms. ~3W. But I've got about 18 watts of power (28 V max)

So here's (the perhaps rather silly question) How should reduce the insulation? By a factor of two or so.

1.) Cut some holes in the foam? (Then I have air.. still a good insulator) 2.) put in worse foam. (I'm using a melamine foam.) Which I picked because it goes to ~150 C or so. 3.) Add thermal shorts..? But where... Moving in from the outside I have, plexiglas, foam, glass cylinder, foam, cell. 4.) something else.

George H.

Reply to
George Herold
Loading thread data ...

^^^^ oops, *not* important. (Ignore the other more obvious grammar mistakes)

Reply to
George Herold

If uniformity of temperature is important thermal shorts or holes in the foam might not be a good approach.

Reply to
A2

This heater has windows on the ends. (I didn't mention that.) I made another heater with open ends, on that one there was a several degree difference from top to bottom.

But I think thermal shorts (and gradients) are fine, as long as it's in steady state. A cold spot on the cell is actually a good thing, as long as the "spot" is not on the window, it gives a point for the excess Rb to collect.

George H.

Reply to
George Herold

^^^^ oops, *not* important. (Ignore the other more obvious grammar mistakes)

=========================================

Just winging it off the cuff, so it's worth what you are paying :-). I agree that holes and shorts would be bad, giving local hot and cold spots that I just think will take longer to equilibrate (I'm predicting lots of overshoot and time waiting for the hot spots to cool back down while you are still pouring heat in to get the rest of the mass up to temperature, and slow approach on the cold spots). Since the heater is on the glass cylinder I'd start by reducing the thickness of the foam layer between glass cylinder and cell by half, and if that works out ok and you want more then do the same to the outer layer. Again, I think starting by removing some of the outer layer will hurt the uniformity at the heater and cause more problems.

----- Regards, Carl Ijames

Reply to
Carl Ijames

The time constant doesn't matter. Think of the heated mass as an integrator. Adding heat leaks just wastes power and will *slow down* warmup, not speed it up.

Reply to
John Larkin

Seems to me if your prime target is stability, you need to reduce any thermal differential too, else convection or circulation is going to make its own instability. Maybe wrap the outside in highly conductive layer, like copper or aluminium to keep the outer layer as close as possible at the same temp, then insulate if necessary to keep within your power budget. Certainly you need more leakage to keep the cooling time reasonably short, but not at the expense of instability.

Reply to
Adrian Jansen

It will slow down warm-up, but also reduce the time constant, which does matter.

The heated mass isn't an integrator, but rather the capacitor in a low pass filter, with the heat leak a parallel resistance to ground (or ambient, in this case).

--
Bill Sloman, Sydney
Reply to
Bill Sloman

Yeah OK, it will take longer to get to it's set point... (when the system is just asking for full power.) (I'll have to measure that time.) But once at the set point, I think more heat leaks will make it faster. .. I'm thinking the slow time is for heat to get out of the unit.

I need some more numbers.

George H.

Reply to
George Herold

Addition, Here's a link to some pictures... if that matters.

formatting link

George H.

Reply to
George Herold

Hi Carl, I added some pictures. The foam is also the mechanical support for everything, so it hard to reduce it's thickness by 1/2.. which is why I was thinking of holes.

George H.

Reply to
George Herold

Hi Adrian, thanks for the ideas. There are a number of other constraints on the heater. There are some RF coils on the outside... (frequencies from

1kHz to 10-20 MHz.) So big pieces of metal are not so good. And anything magnetic is right out.

George H.

Reply to
George Herold

Hi Bill, yeah that's my thinking. (I'll get some numbers later today.) I don't mind calling the thermal mass an integrator. One man's integrator is an others low pass filter.

George H.

Reply to
George Herold

No. One man's integrator is another man's capacitor.

A heat leak might help if your control loop is battling an i-squared heater curve. But it won't speed up warmup and it won't improve thermal stability.

Reply to
John Larkin

The trouble with well-insulated systems using only heaters is the slew rate asymmetry. If George can get the system to settle cleanly without overshoot, the slow cooling rate doesn't hurt anything as long as the set point doesn't move.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net
Reply to
Phil Hobbs

At the set point, you want constant temperature, no? "Faster" has no meaning, unless you need to dynamically tweak the temperature.

Only if you want heat to get out.

Absolutely. You could Spice it.

Reply to
John Larkin

Yeah I think over shoot is my big problem... I had this idea that making the cooling faster, with out effecting the heating part too much would allow me to set a shorter integration time.

I use an omega controller for the loop, which is not ideal. (But that's a whole 'nother can of snakes that I don't want to open at the moment.)

George H.

Reply to
George Herold

Well I want to measure the asymmetry in the heating and cooling rate. Does it make any sense to have those two numbers close to each other?

George H.

Reply to
George Herold

Does that actually need more heat loss?

Reply to
John Larkin

The heating rate depends on how much heater power is available. The cooling rate depends on the heat leaks. They don't need to be close.

The downside of high heater power (fast warmup) is that the loop could overshoot at startup. That can be handled, especially if the controller is digital.

I've seen people measure ambient temperature and apply a feedforward signal, to essentially cancel the heat leak. That can fix loop stability problems, always tricky in thermal systems.

If you can characterize the open-loop response, heater to thermistor step response, you could make a Spice model that behaves the same. If your heater is linear square-law, include that.

PWM is best for several reasons, but it might interfere with your physics.

Reply to
John Larkin

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.