Transitor as heater

Using it as a temperature sensor would be complicated by the fact that, due to its recent service as a heater, it would be warmer than its surroundings.

Since transistors are cheaper than relays or analog switches, why not use one for heating and another for temperature sensing?

And since, watt-for-watt, resistors are cheaper than transistors, why not use a transistor to hard-switch a resistor for your "heater"?

Not crazy at all--you can get better loop bandwidth that way. It's thermal conduction that's slow. Put a voltage divider from collector to emitter, with the base connected to the middle. That makes your transistor an n*V_VBE voltage drop, for some convenient value of n. Use a resistor in parallel with a high side switch, and measure the collector-emitter voltage between pulses.

You do have to use enough sense current that the sensor/heater device doesn't run out of gain.

Cheers

Phil Hobbs

Yeah well I'm going to have to pulse the heat and then wait for it to diffuse (or whatever heat does) into the sample. So I've got a pulse and then a wait time for the temperature to settle down. No matter what I use I need good thermal contact between the sample and my heater/ temp sensor, gizmo (sometimes called the addendum those that do heat capacity measurements.)

Yeah I could do that too. Cheap is not really the issue here. I want low mass/ easy to mount/ good themal contact. (yeah I know, pick two of the three :^)

George H.

Using the transistor as sensor and heater sounds like a fine idea-- you can't beat the thermal coupling.

V+ -+- | Q1 --+ | | | | |/ '-| Q2 |>. | | ===

Forcing high current into the transistor might cause undesirable heating elsewhere--caveat emptor.

Ye Olde d(Vbe)/d(Ic) temperature trick (above) might be handy.

Cheers, James Arthur

,snip>

surroundings.

Yeah I've done that in the past. (put both the pass element and resistor on the thing to be heated.) This is all about small and low mass. I've got a little clamp sketeched out. A couple of 2-56 screws (0.4g) and what looks to be about 1.5 grams of copper on the bottom part and maybe 1/2 gram of G-10 for the flexible top of the clamp.

George H.

ter.

ntially)

g of controlling the current

ransistor as a temp sensor with the c-b shorted. (diode connected transist or.) Could I just push a bunch more current through it for a heater.

g from heater to temp sensor. Do I use relays or analog switches?

Ahh, that's the Vbe multiplier that you see in push-pull stages. A very interesting idea! Thanks.

One issue I have with using the transistor in this way (as well as the simp le diode connected transistor) is that as I pulse the transistor it's going to warm up. Vbe will change, and so the amount of heat delivered to the sample will change. (during the pulse) I'd have to monitor the Vce voltage and integrate (or something) to get the energy. That's not impossible, bu t it's not as clean as I*V*time for a resistor, or wrapping an opamp around the transistor to keep the current constant. It does make another part of life easier though...

George H.

how about a dual fet, available in anything from d-pack to to-220

one fet as heater, other fet body diode as temp sensor?

-Lasse

Use a separate heater and sensor. The problem is that you're trying to regulate the temperature of some attached device, such as a crystal, and NOT the temperature of the heater. If you simply attached a crystal to your proposed stabilized heater, the transistor junction temperature would be stabilized, but the temperature of the crystal would be susceptible to ambient changes. In effect, you would have a temperature gradient starting with the stabilized heater, and ending with room temperature, with the crystal somewhere in between. You could insulate the entire affair to minimize this, but I think that will burn up all the alleged cost savings of having everything in one package.

If you look at what's inside a typical OCXO, you'll usually find the crystal sandwiched between a TO-3 transistor heater and a thermistor. Sometimes, there's a heater wire wrapped around the crystal can. The important point is that the thermistor is on the other side of the heater. The idea is to make the crystal can part of the temp sensor and thus stabilize the crystal can temperature, not the heater or junction temperature.

You didn't mention what you were heating. If it has a large mass, you're also going to have thermal lag and stabilization issues. If it's truly massive, it can take a long time to stabilize. If you need extreme stability, mechanical considerations tend to be more important than electronics.

That's true. It's also not such a nice accurate sensor for that use--I was thinking you were building a temperature controller.

The transistor-as-heater-and sensor trick is great for inner loops in temperature controllers, because it can improve the forcing rejection out of all recognition. You don't care much about its calibration in that application, because there's a slower but more accurate sensor someplace else.

Cheers

Phil Hobbs

You inevitably have that anyway, unless the crystal itself is the sensor. And combining the actuator and sensor can give you a good two orders of magnitude better bandwidth, which translates to 40 dB better forcing rejection at all frequencies. You just put the sensor/heater between the mounting hardware and the controlled volume, and insulate around it.

Multiple-point heater/sensors can be used to control gradients if necessary.

Cheers

Phil Hobbs

All that have already been said.

One pb with in circuit monitoring BJT temeprature is the Rbb.Ib drop, particularly while heating, since Rbb.Ib doubly depends on temperature...

Since you're having the transistor in a feedback loop, you could use a MOSFET instead of a BJT, and monitor its VGS.

- have your heating current pulse;

- set current to a constant low measurement current;

- measure VGS

- repeat.

If you're PWMing a constant high heating current, you could even monitor VGS while heating...

Sensitivity of a few mV/K.

The main drawback is that temperature dependency isn't as well defined as for a BJT so you'll need calibration. But if you're going to digital aren't too picky about accuracy, you could have just one ambient offset calibration point, right at beginning of your first heating pulse, which will compensate for the MOSFET "VGSth" and rely on the better defined (but still not as good as for BJTs) VGS-temp slope.

Hi Guys, (Fred, Jeff, Phil) I'm just going to make one global response.

So this gizmo is to measure the heat capacity of something. (How much does the temperature rise when you add a joule of heat.) It will be in a probe down a LN2 dewar. And in vacuum with a radiation shield around it. The r adiation shield will be servoed to stay at the same temp as the sample insi de*. So I'm not trying to stabilize the temperature. I give it a pulse of heat and then measure how much the temperature rises. I need to know the a mount of heat in the pulse. And then the final temperature of the sample. (There will be an initial temperature spike as the heater warms mostly the temp sensor, but then the heat will diffuse into the sample.) That needs t o be as fast as I can make it.. which means small samples. But I then get pulled in the other direction because the temp sensor/ heater and bits of m etal that hold all that also have heat capacity. This (addendum) heat capa city has to be measured and then subtracted from the total to get the contr ibution from the sample. Which means that large samples would be better.

OK the more I think about it the less I like the transistor as both heater and sensor. It's cute (which makes it appealing), but I'm thinking it's ea sier, both to build and understand, if I keep the functions separate.

George H.

• Hmm OK I'm seeing some problems with having my temp sensor "off line" for some of the time. It's going to confuse my servo. I also haven't made th e servoed radiatin shield yet so...

For really a low thermal mass heater with no lag due to contact problems, could you plate the outside of your heated object with a conductive film and heat it by means of induction currents or microwaves?

Not sure how you would measure its temperature.

Grin, sure or bath it in IR, send a laser onto the sample. The problem with all of these 'types' of ideas is that I need to know how much heat I'm adding. And that's hard to measure for an indcution/ absorption type of heat transfer.

George

Yeah, I wouldn't use it for that either. A nice fibre-coupled diode laser would be one approach, but you probably don't have the budget for that.

Cheers

Phil Hobbs

So your real problem isn't how to heat it - it's how to measure its temperature. That is quite difficult because different bits might be at different temperatures, especially if you are heating it quickly.

Induction might be neat though. You can measure the Q of the resonator; if you can do it accurately enough (and keep the geometry constant enough), you can calibrate it to temperature.

Tim

I went looking on the web for a disscussion of this tehcnique.. but I could n't find any.. (in a quick search) part of the problem is I don't know what it's called.

So the sample is sitting in a vacuum (held by a few strings). Surrounding it is a metal can that is at the same temperature as the sample. The metal can also thermally anchors the electrical leads before they pass through t o the sample. (so no heat load from either radiation or thermal conductivit y of the leads... in theory) The sample is just sitting there and (hopeful ly) it's temperature is not changing. I then give it a pulse of heat I*V*t ime = X Joules. The temperture of the heater and temp sensor shoot up, and then relax back down as heat moves into the sample. After a time (hope fully short...10's of seconds I hope.) the temperature reaches a new (high er) equilibrium. I record it. Heat capacity = Energy (in pulse)/ (delta temperature) Then repeat. Kinda boring really. (Unless there happens to be a phase transistion.)

George H.