Temperature sensing technologies

You haven't said what kind of initial accuracy, but first pass:

1. Vbe or diode Vf
2. LM334
3. Thermistor

piglet

Measure the leakage current of a darlington transistor.

There are a few dozen good possibilities. But one point, generally measurement errors are primarily offset errors. So one simple measurement, which can be done in bulk, is sufficient to calibrate the sensors for improved accuracy.

Not an issue (reread my post)

I don't see how this would work (well) "remoted" (long cable lengths)

sb LM34?

So far, best in terms of versatility of interface (trade time for accuracy). But, a bit of concern for operating temperature range.

As with my response to piglet, I think this would be hard to do, reliably, on the end of a long wire into the field ("no signal conditioning in the field")

I'm usually not interested in accuracy. Repeatability/precision are the more important criteria. I can snapshot particular "conditions" and then use the "present value" to determine how to get to a particular (remembered) "condition".

E.g., you could tell if your car was overheating if you noted the reading when it overheated, once, and just remembered that as a future reference.

LM334 can be set up to give a _current_ proportional to temperature, making it convenient for locating on the end of twisted pair with voltage drop no issue. ...Jim Thompson

Hmmm... piglet had suggested that part and I'd assumed it to be a typo; intended to read "LM34", instead (I've encountered them in a variety of designs).

I'll have to look into it. Something similar to an AD590's behavior?

For me,it was quite reliable, producing a leakage current of several milliamps at the high end of the temp scale. Usable for a range of ~20 degrees Celcius.

Wow! I'd not have thought that! What range were you operating over and what range of currents did you encounter? Were these repeatable (long term)?

Ah, that won't handle my ~100C range. Though I can possibly reduce the range for select applications... (being lazy, I was hoping for a more "universal" approach -- one set of hardware with software tweeks, etc.)

Similar. Except the LM334 current is programmable with an external resistor. ...Jim Thompson

The sense end is a low-power voltage regulator, resistor, and diode-connected transistor. The voltage regulator makes constant voltage on the transistor and resistor in series, so the current INPUT to the voltage regulator depends on the Vbe (temperature dependent) and Vreg (temperature-independent). You measure the current in the long wire, not the voltage. Remember to average over an integer number of AC cycles, there will be pickup.

Oh, OK. So, the resistor's tempco is something else I'd have to consider (?). But, presumably, get me "out of the mud" trying to resolve 1uA (IIRC) differences...

I'll put it on tonight's "list". Have to prep the citrus trees for coming cold spell (while its still nice and comfortable outside)

I imagine you can transport milliamps over a distance of a few 100 meters. The darlington acts like current source, and you can afford to drop a few volts along the line.

OK, but there are "integrated" solutions that will do this. "No assembly required".

I'd like something that I would then not be constrained, mechanically, in terms of how I could apply it (e.g., thermistors are ideal, sizewise, in this regard).

Not a problem. I use FLL's often for integrating converters for exactly that reason.

You wish. The 'integrated' solutions still have to get the wire attached, and most are intended for PCB mount. With bit of heatshrink over it, for strain relief, it's four parts. "Some assembly required".

This guy is just a troll. He wants to stimulate conversation based on his vague post. He already indicated a thermistor is good. He knew it, but rearranged his question to pull you in.

Then don't bother with a sensor at all -- just assume that the temperature is 23C, and move on.

Oh, wait -- you DO have accuracy requirements! You're just not going to bother quantifying them!

You haven't defined "well", or, for that matter, "long", so how can we help you with this?

That depends on those accuracy requirements that you don't want to state. Put a thermistor in series with a resistor, read it ratiometrically from the supply, and you've got something that's most accurate at the temperature where the thermistor resistance equals the load resistance, and tapers off outside of that range. It'll _probably_ be good to +/- 5C with little work over your entire range -- but you refuse to say whether that's good enough, so no engineering can be done on your problem.

Where are you in AZ where you're prepping the citrus?

Creek). ...Jim Thompson