# Couple less thermocouple

1. Aug 12, 2011

### uncanny_man

I really need someone to tell me I'm wrong and why here or this is going to keep bothering me.

As far as I'm aware, a thermocouple works by heating the same side of two different materials, one p-type and one n-type. Since charge carriers also serve to conduct thermal energy, the charge carriers on the hot side will have a net migration over to the cold side building up a positive charge on the cold side of the p type material and a negative charge on the cold side of the negative terminal: this is a difference in potential (voltage). Generally, thermocouples connect the two materials together at the hot side and connect the cold sides of the materials together through a load, completing a circuit and generating current flow.

That all said, what happens if you do not connect the p and n type materials together on the hot side? Obviously current will not flow, but you should still be generating a difference in potential on the cold sides, right? Couldn't you still figure out the temperature from the voltage generated even without completing the circuit then? Again, please tell me I'm wrong and why.

2. Aug 13, 2011

### gsal

I think you could, if you could measure the voltage without having to draw any current, but as far as I know, ideal voltmeters are not around...to measure a voltage, you need a little bitty current and hence a closed circuit...but I could be wrong...

3. Aug 13, 2011

### uncanny_man

Typically you're right, but not necessarily. There are devices to measure static voltages by measuring the associated electric field. Not sure what kind of sensitivity they can achieve though...

4. Aug 13, 2011

### gsal

Just as so many quotes emphasize the difference between theory and practice..let's get practical, here..can you use such "electric-field"-based voltmeter in the environment where the thermo-couple is being used? is it reliable? dependable? cheap?

Are you asking a theoretical question one or a practical one?

5. Aug 13, 2011

### uncanny_man

Theoretical, but for a very specific hypothetical application. I'm thinking of making a kind of detector in which sections of a thermoelectric array are functionalized to heat under certain conditions, others are not. In this case, it is essential to maximize temperature change for a small change in heat and for a localized location (meaning I don't wan't non-essential material adding a heat sink effect), and I don't care about the power output since it is just being used for a sensory application.

6. Aug 23, 2011

### marcusl

Note that a thermocouple need not be made with semiconductors, any two dissimilar metals will do.

7. Aug 23, 2011

### K^2

I'm pretty sure the heat flows the same/opposite direction of the current, so shouldn't hot and cold sides be on the opposite sides of the junction?

8. Aug 23, 2011

### Fra

If I understand what you mean then I'd say No, the "potential difference" is even undefined.

There whole measure of the difference in electrical voltage (electron potential) between two systems, requires that they can exchange electrons, at least in principle. Otherwise it's like asking what is the difference in electrical potential between two universes - the measure is undefined.

So if the metals are strictly and fully isolated, then it's not that the difference is 0V, it's more that it's undefined (which means you have no connection, and volt meter will certainly read 0 V)

That said one of course don't need a direct connection between the measuring ends. You can have them indirectly connected by any conducting medium, but then of course you have two temperatures involved, not one so you'd be measureing something else. (One application of this is so called differential thermocoujples, where you connect two thermocouples in series, and by knowing the temp range, you can accurately measure the differential temperature. This is sometimes a way to get many differential temperature readings with less measurements channels, for example you can get 15 temperature differences in the same range using only 16 channels, which would otherwise take 30 channels if you use two sensors per difference).

/Fredrik