Thermocouple Errors - What to Expect?

Thanks - confirms my suspicions (and that I'm not going crazy). I'll head to the Omega site - thanks for the link.

At boiling point I'm getting variations between thermocouples of around 2 degrees C... so 2% error I can live with... it's the 16% error at low temps that's annoying as I was trying to catch some small variations. Never mind - I've redesigned the experiment to run over longer periods and I'm pumping the water round and round... so I don't have to try to capture 1 or 2 degree rises any more.

Just twisting the wires is not sufficient. A proper junction requires welding, but I know some people get reasonable results with soldering. However, the dissimilar metals used for thermocouples do not cooperate well with soldering. Google "making thermocouples" and you will find many suggestions.

FYI: Here's a good "how to" guide on using thermocouples, including error compensation: http://zone.ni.com/devzone/cda/tut/p/id/7108

Thanks for the links. And over the past few days I have worked out by experiment that it's a real bugger to solder thermocouple wires together!! Can't do it with my equipment... hence the twists. I'll see how the links help me!

Thanks

Robert

Understanding the material composition of thermocouple wires will enable the user to understand that it cannot be soldered. Even if it is possible it would not be wise. The essence of the measuring capability of thermocouples lies in the fusion of the two wires. The junction created between the two wires should prefferably be free from carbon buildup due to too much oxygen in the fusion process if a gas flame is used. For the laymen like you and me specialised equipment for fusing the wires is not available.

I suggest you use a gas flame, something like oxy-accetaleen or similar. Twist the wires as tight as possible for approximately 4 turns and then melt together. You should have perfect results.

Believe this will solve your questions

Regards

LJMarais

Thank you LJ - good thinking - the oxy torch is definitely worth a try.

Robert

>Also if you weld the two wires to a piece of metal that metal will affect the reading also.

This is true to the extent that thermal mass at the measurement point is changed. The transfer of heat is affected by the mass being measured and the mass of the measuring element. It is not true that the 3rd metal affects the thermocouple output because the chemistry of the junction is altered.

>A company called Omega has a great handbook on temperature

Omega does indeed have good material.

Their "Thermocouples" fundamentals app note (document #021-032), available here:
http://www.omega.com/temperature/z/pdf/z021-032.pdf
clarifies the 3rd metal issue.

Parts of page 3 of that document are shown below; those pertaining to the Thermocouple Law of

Intermediate Metals:

Check it out.

How many decimal points in the readout? If you are displaying degrees, no decimals, the reading can be +/- one unit (one degree) out also.

For school electronics projects, I provided crimp ferrules normally used to put stranded wire into terminals. They come in various sizes but the ones suitable for 0.25mm - 0.5mm wire would be best for you. Cut back the insulation on both wires leaving bare wire longer than the ferrule. Insert both (untwisted) wires fully and crimp the ferrule. (There are specialist tools to do the crimp but squeezing in a vice, trying to apply the same pressure each time, will do) Cut back any surplus wire protruding from the end. It is not perfect but as all the ferrules are the same length the joints are both durable and consistent. Typical results are +/- 0.5°C For better accuracy, make more than you need and compare them against a purchased/borrowed (known standard) thermocouple. Use the ones that are nearest to the standard.

Twisting wires together is not the best idea because you don't know where along the twisted length the actual electrical contact is made, or not. This gives spacial uncertainty. It is better to make a weld at the end of the wires so you know the exact source location of your signal. You can buy or build a small arc welder at very low cost. Many other type of welding (gas, plasma, laser,etc.) work well also.