Thermocouple spacing

Inteificio · Oct 3, 2011

Hi,

I am getting issues with thermocouple readings getting messed up by magnetic interference from live wires.

What kind of spacing is needed between an unshielded thermocouple wire and 2.5mm cable to minimise these problems?

Thanks,

Peter

Marvo · Oct 4, 2011

What type of thermocouple arrangement do you have. Is the thermocouple driving something directly or is the thermocouple connected to a pc board as an input sensor? Is it an RTD such as pt100, if so is it a three wire system with reference or just 2 wire?

Inteificio · Oct 4, 2011

That is part of the issue, I am very light on information!

It is a two wire system and polarised. It goes in to small PCB control unit. The temp range we use it in is 50-90. (We wire an apolar thermocouple to the same PCB for our cold stuff).

I would assume that they have filters on this circuit as the interference problem is VERY rare, but still comes up occasionally.

My obvious next step is to speak to the PCB manufacturer, but was wondering if there were any regs or relevant advice!

Thanks,

Peter

Marvo · Oct 4, 2011

I'm not based in the UK but I doubt this would be covered by general electrical or building regs.

I would suggest either you'll need to re-run the CAT5's with a greater degree of separation or replace them with shielded and bond the braid to the services earth. Depending on the existing earth integrity maybe use a dedicated earth spike if necessary to earth the braid.

You could also try converting it to a three wire system which from experience is less prone to interference but this would depend on whether your control board supports the three wire system.

Apologies in advance, it's difficult to be more specific without more info I'm afraid.

netblindpaul · Oct 4, 2011

It would come under the EMC directive and the associated A, B & C standards which are associated with compliance with the requirements of said EMC directive.

Inteificio · Oct 5, 2011

Ok, more info (but still not much).

Two wire 'k' type thermocouple, board only takes two wires.

I could calculate the distance that a current would be induced, and put it outside the thermocouples sensitivity range, but as it would induce AC in to a DC circuit I am clueless.

Thanks,

p.s. This is not CAT5. Just two wires twisted at the end.
p.p.s. I know it is annoying when someone asks a question with limited info, but someone must use thermocouples and know from experience how far to space them!

Marvo · Oct 5, 2011

It's not so much annoying as just difficult to help. Can you even tell us what type of machine it is? How long is the wiring between the thermocouple and the pc board? How far does the thermocouple wiring and the 2.5 electrical cable run together? Are the two cables running side by side in straight lines or do they vary in distance and criss cross each other in trunking for example? What are the original symptoms that led you to believe that induced interference is the problem?

Inteificio · Oct 6, 2011

We use this thermocouple in varying devices from large industrial fridges to ovens.

Cable is at most 3 metres long.

We had a MAJOR issue recently where we were getting a false reading from one of the couples (device kept on heating, temp stayed same, not good).

When I was called in to assist, the other sparky had tried everything but magnetic induction, so that was my first step.
Wires running outside device, no issue; wires running in the device, fixed reading.

I want to go back to our designers saying we need spacing of Xcm between the live and thermocouple wires, give us enough room!

Marvo · Oct 6, 2011

I ran a few searches through my collection of ISO regs and I didn't find anything relevant. Did you check out the ENC directives suggested by Netblindpaul?

Peter Scowcroft said:
We had a MAJOR issue recently where we were getting a false reading from one of the couples (device kept on heating, temp stayed same, not good).

This type of problem is far more likely to be an intermittently faulty thermocouple or control board. I really don't see induced voltages over a length of a few meters being your problem. I've seen SCADA systems where unshielded sensor wiring is actually cable strapped to electrical cables for over 50 meters and there's still been accuracy within 0.5%.
If you're still convinced that induced voltages are the problem then the first thing to do before you start retrofitting a machine that's in production would be to prove it. To start with I would use a working machine and extend the thermocouple wiring to at least double its normal length and experiment with running it alongside a 2.5 electrical cable for varying lengths and monitor the accuracy of the control board. If the results remain constant then you're probably barking up the wrong tree.

Another quick fix option would be to install an external braid to the thermocouple wiring and ground it. You could try this on your faulty machine, if the fault persists then you've pretty much ruled out induced interference as being your issue.