Question regarding thermocouples and a voltage terminal block

Are you going to parallel the tc's into one signal pair? If so, *any* terminal block will work fine. Just find one big enough to squish all five wires together. Or crimp them into a butt splice, any wires in either or both ends (3 in one end, two in the other maybe), doesn't matter.

John

Interesting. I am wondering, since tc's basically convey the temperature as the potential difference between the two conductors, using a different metal at the junction would not alter this potential that we will be measuring across the copper and the constantan. And just out of curiousity, using a metal doesn't typically change the pot. difference does it?

Thanks aga> >

google for :-/ thermocouple "cold junction"

Geo

As long as the junctions are isothermal, they can't generate any potential. As a practical matter, scrunching some number of wires under a screw, or crimping them into a butt splice, will be isothermal.

And oh, we like to bottom-post here.

John

I don't understand- are you paralleling the T/Cs? If so, for type T, just solder the leads together and be done with it.

Best regards, Spehro Pefhany

--
"it\'s the network..."                          "The Journey is the reward"
speff@interlog.com             Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog  Info for designers:  http://www.speff.com

...Well, that's a bit too simplified, I believe.

Different conductors in contact generate a potential difference.

Differences in temperature influence this potential difference in particular ways depending upon the specific conductors connected.

The key is to know the temperature of all other connections (called junctions in thermocouple language) to different conductors so that you can infer the temperature of the test junction based on the known temperatures and the known junction relations.

So, you can use just two different metals, join them at one junction, and connect them to a measuring device. BUT, this device has to be at a specific temperature if there's any other metal involved -- for instance, your voltmeter certainly has only copper in its circuitry, so where it connects to the constantan wire, you have a second junction. You have to know the temperature of that junction before you can infer the test junction's temperature from the thermocouple relation.

This means you have to either control the voltmeter's terminal temperature, or measure it in some independent way. For decades, the usual method was to chill the second junction in ice water (thus the standard term cold junction). Later, it was controlled at a higher temperature. Now, we simply measure the temperature of the second junction and calculate the test junction's temperature from the temperature difference between the two junctions.

But the key is to know the temperature and the materials of every junction in the circuit.

And we ought to edit irrelevant text from our messages.

John Perry

The OP wasn't entirely clear, but it sounded to me that all 5 wires at each junction were the same tc alloy. That being true, the temp of those junctions doesn't matter, since each junction is isothermal and mostly all the same stuff.

If that's not true, well, the temps do matter. As usual, the exact situation is unclear. A diagram would help.

John

...Well, that's just plain wrong. Different conductors in contact do not generate any potential difference unless a poor connection is made, and that effect has nothing to do with proper thermocouple operation. The voltage in a properly assembled thermocouple is generated entirely by the temperature gradient along the conductors. The junctions serve only to complete the circuit so that the difference in voltage produced by the same temperature gradient along two different wire materials can be measured. Of course the junction temperatures are the temperatures of the ends of the two different wires and so determine the total temperature gradient (and therefore voltage) along the length of both wires, so the junction temperatures are used in all thermocouple calculations even though the junctions do not generate any voltage.

So in order to make a proper temperature measurement with a thermocouple the connections to the copper voltmeter terminals must be at the same temperature, in order that the temperature gradient along both thermocouple wires is the same, and one end (either one) must be at a known temperature so that the other temperature can be calculated.

Yes, I'm familiar with all that theoretical verbage. And, as you point out later, it's effectively indistinguishable from the practical reality I spelled out. Which is much easier to introduce to one who doesn't already know it all.

jp

The TC's are being paralleled together.

Do you think by soldering the leads together of the type T thermocouples would give as accurate a measurement as they could with a terminal block of the correct metals?

Attempting to get as accurate as possible paralleling those 4 TC's into the input and having 1 output is our main objective.

Thanks

The TC's are being paralleled together.

Do you think by soldering the leads together of the type T thermocouples would give as accurate a measurement as they could with a terminal block of the correct metals?

Attempting to get as accurate as possible paralleling those 4 TC's into the input and having 1 output is our main objective.

Thanks

The solder adds no error, as long as all of it is at the same temperature (no thermal gradient across the soldered section, since temperature gradient is responsible for the thermocouple voltage).

I assume you are paralleling thermocouples to average many temperature measurements. If so, keep in mind that the resistance of each of those thermocouples plays an important part in how each one contributes to the average. Lower resistance couples will contribute a proportionately larger part of the average result.

If all metals were placed in a box at the same temperature, is there any reason to believe the soldier section would have a thermal gradient?

Thanks,

Pete

Not unless one side of the box were exposed to heat and one side exposed to cold. Wrap the box in thermal insulation and the internal gradients will be very low.

If you are paralleling identical-alloy leads, scrunching them together in a clamp-type terminal block or under one screw, the terminal block material *doesn't matter*. And a crimped butt splice would be just as good. Or twist, solder, and tape. All good to micro-kelvins. The biggest source of error will be alloy differences between the t/c leads and the extension wire, so keep both of the 5-wire junctions close together; but that's way second-order, still a minute error.

Besides, thermocouples aren't super accurate to start with, and I can't imagine the solar cell measurement has to be ultra-precise. So just do it.

John

Really, none of that sort of thing is necessary. Connect the 5 wires together at one point, any way you can, and it will be a negligable contribution to system error. This is *not* an issue.

What kind of signal conditioning will you be doing on the other end? That error will probably swamp the junction thing by, literally,

1000:1. Most commercial t/c acquisition stuff has really mediocre reference junction compensation.

John

I can't agree with that view at all. The verbiage "thermocouple voltage produced by the junction" is complete nonsense; how can it possibly help anyone's understanding? It does not take much investigation to actually understand the physical basis for the function of thermocouples, which in my opinion is far better than relying on a completely fictitious explanation even if that fiction can be used to make correct calculations.

Sure. As someone else said, you're actually going to be getting a weighted average of the temperatures, based on the thermocouple resistances, but if they're all similar materials and similar lengths it will be good enough for your purposes.

Just don't do anything silly like putting the solder blobs near a heat source. You can tack them down by putting them into a terminal strip or whatever. You certainly do not need T/C material in the strips.

Best regards, Spehro Pefhany

--
"it\'s the network..."                          "The Journey is the reward"
speff@interlog.com             Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog  Info for designers:  http://www.speff.com

Late at night, by candle light, John Larkin penned this immortal opus:

At best the actual reading will be the average of all four TCs. No way to tell which one is out of whack if the readings look suspicious.

TCs are common in industrial applications and eg. Devicenet devices can have several inputs and transmit the readings to a central receiver over a single wire. Try taking it up in a controls and instrumentation forum. Go the whole hog and get a PLC and some SCADA for a few kilobucks.

- YD.

--
Remove HAT if replying by mail.