Thermocouple cable

[paulandrews000]

Hi

I have a CR1000 with 8 thermocouples TypeT. Thermocouples are homemade twisted and then isolated thermocouple cable. The cables hang down from a buoy and are used to measure watertemperature down to 8m depth. This has been working pretty good for 5 years now.

But from time to time individual thermocouple-channels show weird data. For example jumping randomly betwee

http://www.eng-tips.com/postedit.cfm?id=8190795&action=editn

the actual value that is for example 6 degC and much higher vales e.g. 40 deg C.
Sometimes I replaced the particular thermocouple with a new one to solve the problem but I recognized several times that values got back to normal without any action been taken.

I am wondering if that can be a weakness in the wiring panel. I always checked that the screws at the wiring panel are tightened firmly.

Any ideas?

I have been checking Aussie sites Pyrosales & temperature controls for information.

 

[Compositepro]

Twisting wires together is not a good way to make a thermocouple, particularly for long term use. There will be surface corrosion on the wires with time and the contact between the wires will be lost. Commercial thermocouples are welded.

 

[paulandrews000]

Thanks Compsitepro.

 

[EdStainless]

There are small commercially available devices that you use to spot weld TC junctions.
Are you leads all TC wire (not copper, except of course for the leg that is Cu)? Are you junction block links also TC alloy and not just Cu, brass, or Ni plated? (they should be marked, and are different for the + and - legs)
Every dissimilar metal connection can generate a voltage. You need to be careful.

= = = = = = = = = = = = = = = = = = = =
P.E. Metallurgy, Plymouth Tube

 

[btrueblood]

Find a local weld shop that does tig welding, and ask them to spotweld the twisted ends of the wire for you.

 

[danw2]

Those Campbell Scientific boxes use a thermistor for cold junction (reference junction) measurement. It's probably a PTC, positive temp coefficient, which would report a higher resistance if its own connections were flakey, resulting in a higher than expected cold junction temperature, which would result in a reported temperature value higher than expected.

I don't know how the reference junction sensor attaches, whether it is standard and soldered in place or optional and connects to screw terminals somewhere. If it connects to screw terminals, then check it for clean connections because any resistance in the connections would result in positive temperature error.