FET as linear mode heater

Sorry I overlooked the low temperature spec. which might rule out bipolars in the probe part. You could make a circuit with a MOSFET located in the probe, as a cascode to a bipolar darlington that is not in the probe, where the bipolar doesn't drop much voltage but the MOSFET does. I don't know which devices you can use at 77K.

Chris

Unnecessary is in the eye of the beholder. Properly designed thin film heaters are _fast_.

Cheers

Phil Hobbs

A constraint that hadn't shown up in the thread until this point.

What kind of sensor do you use use to take advantage of this speed?

Do your thin film heaters incorporate thin film temperature sensors (which could be almost as fast)?

Sometimes. But even things like diode lasers benefit from fast heaters. You can use the monitor diode (or even the laser itself) as the sensor.

Cheers

Phil Hobbs

I've done several bricks that ovenized entire PCBs, along with a crystal or a manganin current shunt buried inside the block.

The heater was usually a mosfet or two and occasionally a Minco resistor thing stuck to the bottom.

It's nice to have all the component tempcos go to almost zero.

ou can use the monitor diode (or even the laser itself) as the sensor.

The monitor diode and the laser diode are part of a common package.

The power dissipator in the package would be the laser diode, so using the laser diode as it's own temperature sensor would be pretty good, but you'd have to have a rather exact measure of the current through it, and good num bers for the ohmic resistance within the package. It's quite a way from Geo rge Herold's original post.

laser diode as it's own temperature sensor would be pretty good, but you'd have to have a rather exact measure of the current through it, and good nu mbers for the ohmic resistance within the package.

With diode lasers, you don't usually care very much about the value of the set point, just about keeping it still once it's there.

The monitor PD is brazed to the same stud as the laser, so it's pretty well identical, assuming that the laser dissipation is constant (which it norma lly is).

Wolfgang has a great page on laser stability vs temperature and current, al ong with a lot of other good stuff:

Cheers

Phil Hobbs

he laser diode as it's own temperature sensor would be pretty good, but you 'd have to have a rather exact measure of the current through it, and good numbers for the ohmic resistance within the package.

e set point, just about keeping it still once it's there.

ll identical, assuming that the laser dissipation is constant (which it nor mally is).

The monitor PD is sitting on a thermal gradient from the laser diode to the outside world. If you are going to worry about the thermal gradients aroun d a - controllable - resistor/FET heating combination, I'd imagine that you 'd be more worried about the gradients inside the brazed package, which you can only infer.

along with a lot of other good stuff:

It's been an interesting topic for a while now. My 1996 milli-degree contro ller paper cites two papers - from 1990 and 1995 - which talk about control ling the temperature of a laser diode, and the circuit described in the 199

It took a while for the people involved to understand that a temperature-co ntrolled single-mode diode laser was going to have better pointing stabilit y than an uncontrolled multi-mode laser diode. It wasn't a physicist-heavy environment.

e outside world. If you are going to worry about the thermal gradients arou nd a - controllable - resistor/FET heating combination, I'd imagine that yo u'd be more worried about the gradients inside the brazed package, which yo u can only infer.

You've got it back to front. The PD tracks the laser temperature extremely well, since it has little dissipation and is brazed to the same 1-2 mm piec e of metal, so the two track in a bandwidth of a couple of hertz. It's on t he hot side of the heater/insulation stack (or heater-TEC stack, more likel y), so forcing is much reduced, very much better than a (quadratically slow ) cold plate.

And as I said, for that application you find the right setting by looking a t laser stability, so you only care about stability and forcing rejection.

Cheers

Ph

I agree. The reference to 'an extra wire' suggests the basic concept is a two-wire constant temperature device, internally power limited to

A three-wire device would allow remote control of temperature.

A resistor is a more reliable heating element than a mosfet. If the control was linear, placing the mosfet at the heating point would allow it's own losses to contribute to heating, but would subject it to the worst spot over-temperatures developed locally during the heating phase.

RL

the outside world. If you are going to worry about the thermal gradients ar ound a - controllable - resistor/FET heating combination, I'd imagine that you'd be more worried about the gradients inside the brazed package, which you can only infer.

y well, since it has little dissipation and is brazed to the same 1-2 mm pi ece of metal, so the two track in a bandwidth of a couple of hertz. It's on the hot side of the heater/insulation stack (or heater-TEC stack, more lik ely), so forcing is much reduced, very much better than a (quadratically sl ow) cold plate.

Power dissipation implies heat flow, which in turn implies a thermal gradie nt inside the stack. Where the photodiode physically sits within the assemb ly will determine where on that thermal gradient it actually sits.

Much as you'd like to claim that I've got it back to front, the reality is that your post is essentially an exercise in posing as an authority figure, rather than an attempt to tell us anything useful

at laser stability, so you only care about stability and forcing rejection .

You did, for the sake of finding something plausible to say rather than to provide any useful information.

I really wasn't expecting you to build a diode laser controller at this stage in your career, Bill, so whether you find it useful isn't all that relevant. Other folks here are still building stuff, and high bandwidth temperature control helps.

Cheers

Phil Hobbs

Channelling John Larkin now?

As is mentioned in my 1996 millidegree temperature controller paper, but I'm sure that your target audience will be suitably impressed.

You know, Bill, my mum (and probably yours) used to say, "You catch more flies with sugar than with vinegar."

You seem rarely able to distinguish an argument from a quarrel. I have zero interest in skewering you, but you seem to find skewering other folks difficult to resist. That's why I usually don't respond to your posts these days, even the technical ones.

Years ago, in this NG, you were one of the most helpful and constructive posters, to the point that once when you were talking about quitting SED, somebody just about begged you not to. That seems like a long time ago. What happened?

Cheers

Phil Hobbs

OK like fig 9.111 in AoE3? Looks like a lot of extra parts.

George H.

Yeah it's seems conservative. At some point I'll want to crank up the temperature above 100 C. I've done this trick with a resistor and BJT. And run it up to 150 C for several hours.

George H.

That's nice Tim, I guess I see the resistor as helping the FET (BJT) share the load at the high power end.

george h.

Hi Chris, Yeah inside some loop, time constants will be in the several second range.

George H.

Sure--it's a tradeoff between gradients and complexity. The problem it's intended to address may be one you don't need to worry about, namely the way the spatial heating distribution of the FET+resistor approach moves around depending on the dissipation.

If the temperature slew is sufficiently slow, or if you aren't sensitive to the gradients changing shape with time, KISS is certainly the way to go.

Cheers

Phil Hobbs