Gas cell heater

Yup. Temperature controllers always slow down at the edges of their range, which is a pain.

Of course at 50C George could use a TEC and solve the problem that way, but it would take a lot of mechanical mods at this point.

Phase change cells are quite pretty as well--you control the heater with a pressure sensor. Way back in grad school I did a rough design of a CCD focal plane for an earth resources satellite that used a tank of 100 grams of n-hexane to average out the peak dissipation to something the passive radiator could cope with, and hold the FP at -90C.

There's a table of materials on Wiki,

Beeswax comes in at +60C, and is a nice benign material for student use.

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

Water freezes at - wait, don't tell me - 0C, at which point its resistance goes way up. That can make a super-accurate closed loop.

I suppose its dielectric constant changes, too.

Optical sensing should work on beeswax.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

Pressure sensors are more reliable, because you don't always know what part of the phase change material will melt first, whereas there's a 1:1 correspondence between pressure and melt fraction. Stuffing the cell with bronze wool improves the response speed of the cell by a lot.

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

Hmm, There are not many good thermal conductors which are electrical insulators. Beryllium oxide ? Sheets of that silicone material used for transistor thermal mounts ?

OK data! So I heated up the cell at full power (28 V into 50 ohms about 15W.) It took about 8 minutes to get from 23C to 60 C, with a fairly constant rate of

5 deg C/ minute.

I then shut it off and let it cool.. and took the slope at 50 C.

It's just about 1 deg C/ minute.

George H.

Re PWM: Yeah the output of the omega is PWM, which I filter and amplify to give a DC voltage with ~10% ripple at 1 Hz. I know (now) that the filtering screws up the linear gain of the PWM output. I could try getting rid of most of the filtering and see if that helps.

George H.

As John says, getting the two numbers to be the same requires that the set point be in the middle of the available temperature range.

Compensating temperature control loops isn't too hard to do approximately on a small system, especially when the set point is fixed. I usually model it as a time delay followed by an integrator, which works pretty well, and as a bonus, you can read the parameters right off the scope display.

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

OK I measured a different ratio, Full power heat rise at operating point/heat loss at operating point. About a 5:1 ratio in my case. (to be honest, that is not as bad as I thought it was.)

George H.

If I'd know about TEC's when I designed it I might have tried one. I just tried sticking a magnet on a TEC and they look non-magnetic, which is good to know.

One experiment you can do with the cell is to look at the transmission as a function of temperature, we heat the cell up to 140 C, but 100 C would be plenty. Along with the vapor pressure curve for Rb you can determine the cross section for absorption. Which turn out to be much bigger than the size of an Rb atom... and much closer to the wavelength of the light. (lamda/ pi or so.)

George H.

Well my two numbers would be the same at the maximum temperature at max power. Which I don't want.

I think the over shoot recovery will be faster if I make it leak more heat.

(Maybe I'll take it home. I always find these long term thermal things are more fun to do while watching football and having a beer on Sunday. :^)

George H.

You might try applying a smaller heater voltage and recording the step settling response. Then hack an electronic circuit with the same time response. Then wrap your control loop around that. Spice it during the commercials.

Use pumped heated/cooled air from outside and have it flow around your cylinder within an insulated outer cylinder. Constrict the air exit, maybe with more insulation. Insulate the air pipe so that there's minimal loss between the heater and the cell.

To heat the air, you could use one of those power resistors with a cooling tube down the middle. A fish tank bubbler pump may be useful. The control loop is outside and you're relying on the insulation to ensure that the cell temperature is ok, but you could monitor that anyway if you need a fine adjustment.

I've done similar for high temperature long-term testing with the DUT in a thermos flask, the air tube going down to the bottom and the top plugged with insulating wadding. Works a treat.

Cheers

--
Syd

Sadly, the environment changes, so the heat that you have to put in to maintain the set point changes. "Faster" then means how much the control loop lags the changing environment.

Sadly, you need the numbers to plug into Spice.

--
Bill Sloman, Sydney

Diamond is both a great conductor and a fine insulator. Apparently you can buy sheets of synthetic diamond (made by vapour deposition in a hydrogen-rich atmosphere) but I understand that it's fairly expensive.

--
Bill Sloman, Sydney

Hi Syd, way back when.. we though about hot air, and did a few experiments. The big problem for this setup is that changing magnetic fields are not allowed. B-fields in the milli-Gauss range are enough to screw things up. That means that whatever motor is driving the hot air has to be a few meters away. (It doesn't screw up the signal, but when I take data at home in my living room, I can see the changing B field from spinning magnet motor that pushes water through my fish tank.. about 10 feet away.)

So now I've got some ugly "umbilical" cord of insulated tubing running into the thing. We decided wires were easier. Unless you have a non-magnetic motor? (Hand crank operated by grad students perhaps. :^)

George H.

Well, the pump can be as far away as you like, it's just pumping ambient air to the heater through, say, 1/4" tubing.

The fish tank air pumps I've used don't have a rotating motor, just a solenoid operating some 'bellows' at 50Hz from the AC mains - still magnetic, of course, but at least at a known frequency and with no commutation.

Cheers

--
Syd

Yeah something like that... You don't have to wait for commercials when watching football, I only need to watch a few seconds every minute of so.

I'd like to redo the electronics sometime soon.

I'm running out of the plexiglass assemblies too, so if I wanted to change something now would be the time.

For most of the experiments that the kids will do with this the stability is not an issue. (Wanna see some data?) But I was playing around the other day, and found these small, but maybe interesting signals. (Hmm I'd have to try and explain some of the physics.. I don't mind doing that, but it might be boring to everyone.)

OK here is the money shot! (If I don't get this we don't sell it :^) This is the change in light transmission as I sweep the B field through the Zeeman resonances.. they get split at high B-fields because of interaction of the electron spin with the nuclear spin.

The F=1 transitions are very small, and sometimes go the "wrong way". (F is a measure of the total angular momentum of the state.)

Change in the signal level of the biggest absorption is about 2V/(7V *100) ~0.3%.

Tilting the interference filter I pump with the other Rubidium line.

The signal is about 10 times smaller, (Gain = 1000) (I'm not sure why it's so much smaller.) The tilted nature of the signal is due to the thermal drift in the cell.

George H.

Yeah but they still have a magnet flapping around to drive the bellows. Vibrations are bad too, so it would need an expansion chamber or something.

I appreciate the other ideas, but a major redesign is not in the cards... Any changes I make I'd like to be backwards compatible to what we've already sold.

George H.

I cost were no object, you could make it out of aluminum nitride and have serious heat conduction.

Could the case be partly or mostly aluminum? Start with aluminum tubing maybe. Is RF involved?

Somebody makes a machinable ceramic, which may conduct heat better than plastic.

(Those BMP files are huge!)