Thermistor pressure sensor

Flow measurement using thermistors and Germanium transistors...

...Jim Thompson

Keeping the thermistor at a constant elevated temperature at least removes its non-linearities from the result, and you end up measuring a voltage whose square is related to the power needed to keep it there.

I think that's a fair use of the op-amp. It's not really open-loop -- it's more that the transistor forms an external output stage. That sort of thing is done often.

You can run into stability problems with that sort of circuit if you try to go fast, but it's not here -- the C1-R4 network in the feedback, with R5 to isolate the op-amp final stage from the FET's gate capacitance is going to really stabilize things.

OK fair enough. It's all wrapped in a loop.

Thanks, George H.

Bandwidth should be dramatically better with the feedback scheme. Same idea as keeping a PD at constant bias.

Best regards, Spehro Pefhany

o keep the element at a constant temperature.

ent source and measuring the voltage. But maybe this feedback type scheme is better?

Hi Spehro, I'm not following you. (and what's a photodiode at constant bia s?.. confused as always)

The web page talks about this being slower than a TC gauge. It's got more mass and you've got to get the heat out. (Though you might think that's a mass to surface area ratio.) I guess having it hot will help, speed wise. At the very low pressure I'll guess it cools by radiation, with some heat d own the leads. I was thinking I could stick this almost anywhere in the va cuum chamber. but if it sees surfaces that are colder than room temp (or w armer) that is going to throw off the low pressure point. (Higher temperat ure set points would reduce that effect.)

George H.

Why not just buy a thermocouple gauge? It's a bit odd that they still come with octal-tube bases (that traces back to the homebrew days when folk would hacksaw the top off metal tubes and spot-weld wires onto the exposed feedthrough wires).

Still, the concept grew old because it had significant user satisfaction.

Show a really good reason why you would use thermistors to do this, when MEMS pressure sensors are now so good. MPX2010A from Freescale, for example.

In hand-waving mode, if the temperature is constant (like the PD bias voltage across the relatively large capacitance) you don't need to actually wait for the temperature to change to measure the power required to maintain the temperature, all you need is enough change above the noise floor to compensate for the change.

The bandwidth increase in hot-wire anemometry is orders of magnitude.

I have the equations if you're interested, and could scan them and send them to you.

Even so, it may be slower than a T/C guage because you can make T/C's with *very* low heat capacity compared to a thermistor, so it's comparing apples and oranges.

Best regards, Spehro Pefhany

If this is air, is humidity irrelevant in your application ?

Depends on temperature. The partial pressure of water

is about 22mm Hg at at 25C (US room temperature), so the water molecules are - at most - 2.9% of the molecules hitting the hot sensor and cooling it.

The heat capacity of tri-atomic water molecules is actually a little higher than that of diatomic oxygen and nitrogen molecules, but you have got to be doing very accurate measurements before the potential error is big enough to worry about.

An item I saw in an Alcatel helium leak checker was a tungsten filament light bulb the same size as an NE-2, with the fill tubulation snapped off. They had a similar powered bridge scheme. I'm not sure it was run at constant temperature, may have been some sort of sliding scale. Anyway, obvious you can't run the filament over a couple hundred C in air. It seemed to be really simple and sensitive, I'm guessing the filament would be a lot more sensitive than a thermistor with thick wire leads.

Jon

On the other hand, the thermistor resistance is about ten times more sensitive to temperature change than the resistance of metal wire.

This comes up regularly in every discussion of why thermistors get used, rather than pure metal wire resistors.

Yeah I've got a TC gauge. It's OK, but I like the thermistor idea because I've got a volt level output. So it's easy to interface to other 'stuff'.

I wrote an email to Roy (the inventor, or at least first to publish) He's not involved in research anymore, but he did mention that he played around with using a light bulb as a sensor. Well an exposed filament. That's interesting too.

George H.

Hi Adrian, I guess I should have called this a vacuum sensor. I've used some of those piezo pressure sensors. For vacuum I really want to know when the pressure drops from ~100 mTorr to 1 mTorr. A heat flow type of sensor is perfect for that.

George H.

Dang... (well no this is good!) So there is a fundamental concept here that I think I know, but don't *really* know. So If I'm reading right, you're saying this is similar to bootstrapping a PD. (I don't need to get all the charge off the PD.. just see the change and fed that back.)

No, please don't bother, but thanks for the offer.

Yeah most likely. At the moment I don't think I care too much about speed. (but of course it's always silly to throw away bandwidth if you don't have too.)

Roy, (the guy whose website I found this on.) said he also played with some tungsten X-mas tree lights that had R ~ 100 ohms at 100 C and those worked too. I like the idea of using consumer items for other purposes, though Phil H. may accuse me of turd polishing :^)

George h.

Well in some way sure. The type of gas will effect the calibration of the sensor. In some sense I'm really measuring the thermal conductivity of the gas, and then just "guessing" the pressure from that.

George H.

Great. I like that idea too. In some ways tungsten is more interesting than a thermistor. (mostly because it's more fundamental.)

George H.

Great. It'd be a lot of fun to compare thermistors with little light bulbs. I can't spend any real time on this. But it would be a nice project for a high school science fair.

George h.