analog temperature sensor

Millions of people did. Feel free to give us some clues, like temp range, accuracy, budget etc

:-) Temp range: -20 70 degrees celsius Accuracy : it's not defined so i guess some standard should be used Data reading : reading should be every 5 seconds Budget: max $40

thermocouple, PIC, and battery? or 24 Vac trannie?

or single chip with diode based measurement, no digital at all, just converts the Vbe into Vout?

The classic solution is a platinum resistance sensor. Typically you buy a p art that has a resistance of 100R at 0 degrees Celcius and rises - almost linearly - to 138.5R at 100C.

Typically you excite them with 1mA, so you've only got 100mV or so of signa l and you do need a good op amp to take advantage of the precision you are paying for.

NTC thermistors are an order of magnitude more sensitive and if you buy the up-market interchangeable parts, they can be almost as stable. They are ho rribly non-linear, they can't dissipate as much power as a platinum resista nce sensor, so they need even more gain in the op amp.

Semiconductor temperature sensors are a lot easier to use, if less stable a nd somewhat noisier. The National Semiconductor LM35 has been around for a while.

There are lots of alternatives.

There's a whole industry out there.

no digital at all.

Have you looked at replicating the meter that environmental test labs use to read all those thermocouples? Pure analog.

-------- no i didnt. The problem is that we are using two devices and the end device waits for some voltage level from 0 - 24V (min - max ). So i was thinking to avoid all kind of conversion and to go with pure analog sensor.

LM94022

There is also a Spice model available for it on the Device Models & Subcircuits Page of my website. ...Jim Thompson

those are old analog meter technologies. like a non-linear ohm-meter.

If you do yourself, you can use 'diode switching and resistors' around the feed back of a single OpAmp to 'shape' the output. Pure analog and fraught with difficulies in Production, but does work.

Or buy an LM94022 off-the-shelf ;-) ...Jim Thompson

LM335Z, with an opamp to provide the required output.

For $40 in parts you can use any of the main technologies except infra red. Simplest is probably a transistor diode junction plus an opamp. Trs can do upto about 200C, as long as you keep P_diss minimal.

NT

I knew there'd be a simple chip solution! Is it low enough thermal mass to be equivalent to a thermocouple? or less thermal mass?

Google "Temperature Transmitter". Most of these are 4-20mA devices, but there are some with a voltage output.

Cheers

I've not personally used it. It was part of the Apex MP118 high-voltage OpAmp Spice-modeling project. ...Jim Thompson

A cheap one here, 0-10V out -50C to +100C.

Cheers

Thermocouples can be pretty small! That, and the resulting fast response, are their main positive attributes.

All wired sensors, especially those in plastic packages, have a strong tendency to measure the temperature of their leads rather than that of the intended point. I expect an SC70 one to be better than a TO92 for that, but plastic's thermal conductivity is roughly 2000 times worse than metal and its diffusivity (the speed with which the temperature changes, i.e. alpha /( rho C_P) ) is more like 5000 times worse. So it doesn't take much plastic to trash the response speed.

The main heat conduction path into the sensor will be through whichever pin is attached to the paddle.

Cheers

Phil Hobbs

Without an accuracy spec, your question remains meaningless. In the snarky extreme, you can just use a piece of wire, reading 0V = 25 degrees C. It's not very accurate, but it sure is cheap!

Go back to your boss (or your customer) and get an accuracy requirement. Or bounce some suggested accuracies off them, and see if they go "oh, that's ridiculous" or "oh, that's fine". Then send them an email saying "this is the accuracy I'm going to achieve -- is that OK". Save the response for later, you may need it.

If it just needs to display room temperature for a consumer device or some such, then an integrated temperature sensor is probably more than good enough.

The LM94022 runs at so little current, self heating is not an issue. Thermocouples require auxiliary circuitry. ...Jim Thompson