Gas cell heater

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

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

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.

^^^^ 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

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.

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.

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).

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.

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

George H.

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.

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

George H.

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.

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.

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

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.

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.

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.

Does that actually need more heat loss?

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.