Thermocouple measurements go crazy when current is applied. 6

[camerontwilliams]

I have a problem with our thermocouples jumping the temperature readings either extremely high or low as soon as we apply a current to the test loop. Once the current is shut off the TC's go back to normal measurements. I was at a lab and they had grounded a part of their transformer to avoid this problem. Does anyone have any idea what to ground or how to fix this problem?

Thank you,

 

[IRstuff]

Does anyone have a schematic of this setup?

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[waross]

I am confused. Current is not normally applied to thermo-couples.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[camerontwilliams]

Yes I have attached the schematic of the test loop.

 

[camerontwilliams]

Bill, we are testing electrical connectors. We have to apply a specified current from the UL 486A - 486B standard depending on the conductor size used with the connectors. We then have to watch temperature rise until we reach a steady state temperature of the connectors. The connectors cannot exceed 50 degrees Celsius over ambient. In order to accurately measure the temperature of the connectors we mechanically and thermally bond them to the point of contact between the connector and the conductor. Which in turn puts the TC in the direct path of the current.

 

[Skogsgurra]

"puts the TC in the direct path of the current"

So, what did you expect? The few millivolts produced by the TCs will drown in the potential differences caused by your rather high test current.

DO NOT BOND TCs TO ANYTHING THAT CARRIES CURRENT

Insulate them or use NTCs or something like that. This is really very elementary. You should not ask questions like this.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[OperaHouse]

I sent a board off to UL for testing. Apparently some co-op student bonded a thermocouple to 480V trace. It came back pretty nasty, but it passed and there was an apology call. Hate to see what happened to their meter.

 

[itsmoked]

Use ones like these:
Insulated T/Cs

Keith Cress
kcress -

http://www.flaminsystems.com

 

[camerontwilliams]

Gunnar England, no need to be rude. I simply was looking for advice, and for your information Underwriters Laboratory would argue against your point. That was the first thought to cross our minds when analyzing the test criteria. We asked their Principal Design Engineer and he responded that they test with the TC in the path of the current very often using NON INSULATED Thermocouples. Now if you have any useful advice that would be greatly appreciated, otherwise keep your negative B.S. to yourself.

 

[Skogsgurra]

I think that I have some right to tell the truth. And it is not BS. It is actually a solution to your problem.

I get upset when someone who says that he is an engineering professional is as ignorant as you seem to be. And then I say so. Like it or not.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[itsmoked]

Then you need to use a properly galvanically isolated TC front end in your instrumentation. This is where the front-end electronics are isolated, meaning the power to run the front-end and the resulting signal from the front-end are all passed magnetically or capacitively to the data logging or display output. You would need each TC to be on it's very own isolated circuit so they do not become part of the circuit between themselves.

A classic way this was done was with flying capacitor front-ends. The logging/display instrument consists of a single front end but all the numerous inputs are hooked up to that front end individually, one at a time, and sequentially. That prevents any two from ever being connected simultaneously.

You can test this by hooking one of your TCs to a DMM with TC capability and seeing if it 'goes crazy' while having no current path thru the hand-held battery powered DMM.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[camerontwilliams]

No were did I claim to be an engineering professional, in fact I am quite the opposite, hence why I am on here asking questions rather than answering them. You do have the right to tell the truth, however you do not have the right to tell a person who is looking for help what kind of questions they should or should not be asking. That my friend is ignorance at its finest.

 

[analogkid2digitalman]

Another vote for isolation be it within the T/C construction or via a signal conditioner, but this time from the voltage perspective:

IR drops between the monitoring points in the test circuit may cause voltage differentials between the T/C's. This can cause inaccurate readings if the probes all go to a common front end.

-AK2DM

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
"It's the questions that drive us"
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

 

[Skogsgurra]

Even with perfect front end insulation and CMR, you will get influences from the current if you do not attach the TCs properly. There will almost always be some current flowing through the TC, which is now a tiny part of the contact piece you are measuring on.

It can be done, yes. But it requires quite a bit of experience. Much better to insulate the TCs. The good thermal coupling can be had by wrapping the stem of the TC around the metal piece.

Still somewhat surprised to see your question. And your reaction.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[Skogsgurra]

11 Jan 17 22:49
"No were did I claim to be an engineering professional"

There are forum rules and certain assumptions. One of the basic ones can be read at the top left corner (scroll to top).

I really do not understand why you call me your friend? I have not got that impression before. What made you change your mind?

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[camerontwilliams]

Thank you Keith Cress, I will try your method using a DMDM and see what happens. Gunnar Englund, thank you for your help.

 

[Skogsgurra]

OK, I'll show you the way out:
1. Keep everything as is.
2. Run current until the set-up is "hot enough"
3. Switch current off and determine the thermal time constant.
4. Continue running test and switch off periodically to read temperature. Use the measured time constant to correct readings, if necessary. I think not.
5. You may compensate for the reduced duty cycle by increasing current correspondingly. It is RMS that counts.

I can very well understand if you want to call me your friend now.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[waross]

http://www.omega.com/Manuals/manualpdf/M10632ML.pdf[/URL]]
502A-J
4-20 MA THERMOCOUPLE
-------
TRANSMITTER
2.0 SPECIFICATIONS
2.1 INPUT Configuration: Isolated input
Thermocouple type: J (“Iron/Constantan”)
Input impedance: 5 MOhm
Thermocouple break-detect current: 50 nA max
Burnout indication: Selectable up or down overscale
Thermocouple lead resistance: Up to 500 ohms for specified performance
Normal mode rejection: 60 dB at 50/60 Hz with 100 mV input
Common mode voltage, input to case or output: 2100 V peak per high pot. test; 354 V peak per IEC spacing
Common mode rejection, input to case or output: 100 dB min from DC to 60 Hz
Overvoltage protection: 120 V ac max/1 min exposure

That said, there is still a possibility that some test current may flow across the thermocouple and distort your readings, depending on the mounting and contact method. (As mentioned by Gunnar.)
The danger here is that the readings may be off but not off enough to be noticeable.
That is rather than obvious errors, the readings may seem good but still be in error.
The insulated probes linked by Keith may be the best way to go rather than isolation.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[moltenmetal]

We use ungrounded thermocouples as a matter of course. That means the junction is insulated electrically from the sheath by means of a thin layer of MgO insulation packed into the construction. Grounded thermocouples, where the junction is connected to the sheath, are notorious for being affected by stray currents and ground loops even in circuits where there is no obvious source of current flow. In the case of monitoring the temperature of an electrical conductor this is obviously even more important- and additional isolation in the monitoring circuitry is probably warranted as suggested by others.

 

[LionelHutz]

I'm curious what the TC actually looks like since there are so many possible variations. We generally just use a raw element, just the 2 wires welded together into a ball at the end, and UL has never had an issue with it being insulated from a live part.

Still, I would agree that having the actual TC element in the current current path is rather silly. I wouldn't be surprised if that was a misinterpretation of what the UL engineer said.