Temp Sensor w/ low Thermal Time Constant?

There are some pretty small thermocouples and varistors out there. I suspect you've got it made in the shade if you just look.

For that matter, one of the junctions on a transistor (I think it's the BE junction, but I can't remember) makes a pretty good temperature sensor. I suspect that a SOT-70 part would have a short enough time constant if you just soldered wires onto the appropriate pins and encapsulated the thing in a minimal amount of epoxy.

At worst you could paint a black dot on your item of interest and use one of the new cheap thermal cameras to measure the temperature. You'd need to do calibration, but the time constant should be well below a second.

A small bead-type thermistor ought to do the job. I've seen quite a few which are smaller than a grain of rice - some are about the size of a sesame seed.

Are you looking for a free-air sensor, or one which will be bonded to a solid surface? The latter are likely to be able to respond faster due to the higher thermal conductivity of the material.

I've got really small thermocouples 2 mil wire thickness.

I think any diode junction can be used. (I like diode connected transistors.. bc shorted)

Even a little thermistor might have a time constant less than 1 sec.

George H.

What do you want to measue the temperature of? Air? Some object?

The thermal time constant depends on what you are measuring; a metal item's temperature can be measured in a millisecond if you weld thermocouple leads to it.

Usually (not always) the time constant is related to thermal equilibration of multiple objects. You can compensate for time delays if the system is well characterized (like fast-reading IR thermometers anticipate where their sensor readings will stabilize).

far too little info

NT

I have some data showing a (very small) metal bolometer's rise time at less than 20 ps. I agree that it all depends on what exactly the OP is trying to measure.

Cheers

Phil Hobbs

Kevin - I have used a small surface mount zener as a temp sensor in the diode mode and as a heater in the zener mode. The zener was a 060? package soldered to the edge of a pcb. In other words, one end of the diode was soldered to the top and the other to the bottom. The board shape was tapered to a minimal dimension at the contact position, so it could be waved in the air or placed in an air stream. From rough memory, the thermal response was the equivilent of a time constant in the neighborhood of 1 second.

Hul

Kev> I am looking for a temperature sensor with a low Thermal Time

One of the better temperature sensors around is the "interchangeable" therm istor - so called because parts of the same part number have resistance to temperature curves that come within (usually) 0.2C of overlapping at 25C - more expensive ones can do better.

If this is in good thermal contact with object whose temperature is being m easured - like drilling a thermistor-sized hole in an aluminium block, fill ing the hole with zinc-oxided loaded thermal grease then pushing in the the rmistor, the thermal time constant can be below a second - maybe as low as half a second.

Semiconductor temperature sensors can do better

talks about soldering a metal packaged device to the object to be measured (losing your electrical insulation), while the TO-220 packed LM35 can be b olted down with with a well-greased mica washer for electrical insulation t o do almost as well.

Thermistors are at a disadvantage - they are made by loosely sintering meta l oxide particles, and the thermal resistance inside the device is fairly h igh.

Semiconductor sensors are made on a lump of silicon and there's a whole tec hnology there for keeping silicon junctions in close thermal contact with h eat sinks.

Thin film platinum resistance sensors on alumina substrates are equally goo d - but give a factor of ten less volts per degree, so are harder to use.

Liquid.

If it's aqueous, and reasonably inert, a SS-sheathed probe with thermistor inside is a common choice. If the liquid is changing temperature rapidly, it'll have to be stirred to get best thermal transfer, and of course to get to uniform bath temperature.

This one is about 8 seconds:

If the liquid is liquid helium, or liquid glass, the answers are very different.

Well, yes...with restrictions. Use a transistor, common base, both junctions forward biased. Power into one creates the heat or analog of the heat-creating current at the input (circuit). Diode voltage of the other junction is the measuring component and does not depend on input linearity (or lack thereof). One can make a refinement with a square-wave current drive on the output junction ratio 10:1 and measure the delta V. Time constant can be less than 10mSec depending on the die size and bonding pad; easily measurable.

A SOT23 dual may be in the 100mSec region (at best is my guess) and give reasonable DC isolation (under 1KV one would think). One transistor for power in, the other for measuring; DCT configuration on both. Again, time constant depends on the die size and bonding pads; TC easily measurable.

Congratulations,you just re-invented a bunch of semiconductor temperature sensors which have been around for years.

The LM34 and LM35 do come to mind, but there are plenty of others.

So why don't you go back to concentrating on re-inventing the wheel? There's a much larger market for that.

So called "instant" thermometers work by using a regular temperature sensor of any type, followed by a filter that compensates for the slow response of the sensor.

You can add as much high frequency boost as you want, up to the point that the noise becomes a problem.

Regards, Allan

Bill S. once again going for the "Miss Congeniality" award.

George H.

Also, the AD 590 series, give a current proportional to temperature, in absolute units. 1 uA/K

These are great if you need to run long wires to the location to be sensed.

Jon

LM35 doesn't like long wires. It can oscillate, pick up noise, and latch up.

thermal time constant largely depends on how it's attached to what it's measuring the temperature of.

thermocouples can have low thermal mass.

For some solid thing the time constant goes as the length squared. Well if it's ~cubical, spherical.

(Heat capacity C ~ volume, and R goes as length/Area)

I think a cc of Al or Cu is about 1 sec. (Which leads me to conclude that Phil's