Reliable temperature measurements

Hello guys,
In recent threads, I've read about so called grounded and ungrounded thermocouples. I've also read that grounded (with hot end directly connected to a metal insertion body) are mostly used when fast response time is needed. This makes sense. Naturally, direct connection enables hot end (measuring end) to be in direct contact with fluid whose temperature is being measured. I wonder if this is dangerous for industrial application where different disturbances are possible (induction). I think ungrounded thermocouples are fare more reliable because thermocouple itslef is isolated from the metal body which represents some kind of shield to disturbances. I recall from theory that metal shield is very good protection from outside induction disturbances. In such case I think that induction is only possible on extension cables if they are not shielded.
Believe it or not, recently on my work (power plant) we had a lot of problems with reliable temperature measuring. Basically, all temperatures suddenly were going mad and there were severe fluctuations which lasted for about two or threee hours. We failed to find out what was the cause, but by testing we discovered that grounded thermocouples were used. Problems gone when we replaced them with ungrounded ones. Interesting, 98% of the time grounded T/C were just fine, but there were periods when readings just "go mad".
Did you have similar experience?
Please share your experience with us.

I've had dozens of similar experiences to yours, where the standard approach is to start with replacing the grounded T/C's with ungrounded T/C's.

The reason for ungrounded T/C's is mundane and typical - they're cheaper. The 'standard' sheathed, mineral insulated (MgO)thermocouples is almost always grounded, with 'ungrounded' being a $1 - $2 option per thermocouple as an 'adder'. In the world of purchasing, commodity pricing is critical, and if the T/C is not specifically spec'd as ungrounded, then it's quoted and purchased as a grounded T/C assembly, low bid wins.

Unless fast response is absolutely critical, ungrounded thermocouples can save lots of headaches.

Most PLCs have a selection of AI cards, where the least expensive card has unisolated inputs, which in combination with an ungrounded thermocouple is a prescription for a ground loop.

In a flow application, the thermocouple is probably installed in a thermowell. I suspect that the mass of the thermowell contributes a greater thermal lag than the small amount of insulation of the T/C tip from the sheath.

If you are using isolated thermocouple transmitters then you can avoid ground loop problems, but are just as likely to have to deal with normal mode electrical noise.

If the possible source of noise is a variable frequency drive, several posts on this thread
http://www.eng-tips.com/viewthread.cfm?qid=203504&page=3
reveal a lot of experience in attempts to minimize noise created those VFD's.

There's lot of insight in the statement by Marke, "If you earth the motor back to the main earth, then the HF energy must flow via the main earth and that spreads the distribution of the energy through other circuits and increases the level of interference."

Dan

danw said:

The reason for ungrounded T/C's is mundane and typical - they're cheaper. The 'standard' sheathed, mineral insulated (MgO)thermocouples is almost always grounded, with 'ungrounded' being a $1 - $2 option per thermocouple as an 'adder'. In the world of purchasing, commodity pricing is critical, and if the T/C is not specifically spec'd as ungrounded, then it's quoted and purchased as a grounded T/C assembly, low bid wins.
Dan

Click to expand...

English is not may native language, but from what I've read in your post, first sentence states that ungrounded T/C are cheaper, while the rest suggests that grounded T/Cs are cheaper.... Don't really understand this.

Sorry, the fault is my mistake, not your English comprehension. It should have stated, "The reason for grounded (not ungrounded) T/C's is mundane and typical . . ."

It takes extra production effort to position the element so that it is surrounded by insulation and does not touch the sheath, hence the extra cost.

Dan