grounded or ungrounded


Thread Starter

Fred Chwalek

I have a Watlow Anafaze temperature controller. The spec states that I should use all grounded or all ungrounded thermocouples. We have a box full of used and assorted thermocouples. How can I tell if they are "grounded" or "ungrounded" and what exactly does that mean? I think it means the thermocouple is isolated from the sheath. Is there an easy way to test this? Fred Chwalek Applications Engineer Beacon E-Commerce, eFilter Division Voice: (704) 398-7916 Fax: (704) 398-7822 Email: [email protected]


Hello Fred! The sheath of a thermocouple cable is usually connected to ground at one end. You will have to check if they are grounded at the thermocouple end in the field or at the control panel end. If the sheath is hanging un-connected at the both ends, the termocouple cable would be ungrounded. The sheath does not come in contact with the thermocouple (grounded or ungrounded). It only serves as a 'shield' against EMI for the signal passing through the cable. Clear skies! Omer
Fred, Your assumption is correct, grounded T/C's have a connection between the sensing element and the sheath. You can tell which is which using a resistance measurement between either sensor lead and the sheath. Of equal or greater importance is the type of T/C you have. Are they type J, K, E, R, S, t, or one of the several Tungsten/Tungsten devices. The wrong one can give false readings by as much as 200 deg F or more, depending on the actual temperature. These devices are relatively inexpensive, and by the time you have gone through your junk box to find one that MIGHT work, spend the $100 or so to get a new T/C of know value. Just my thoughts on the matter.

Vitor Finkel

Sure, Use an ohmmeter. A well insulated thermocouple (ungrounded) should read something in the region of Meghoms between sheath and any TC terminal. Grounded TC should probably read less then 1 ohm. What should you do with TCs showing an insulation resistance of a few Kohms to Hundreds of Kohms, as sometimes it happens, I am not sure. Depending upon your application, and the associated impedances, some of them might be still useful, but beware, those poor insulation resistances may change wildly with time, temperature, etc. Better throw them away. Vitor Finkel [email protected] P.O. Box 16061 Tel (+55) 21 285-5641 22221.971 Rio de Janeiro Brazil Fax (+55) 21 205-3339
Your guess is correct. A grounded TC has the junction welded to the sheath, and an ungrounded one is separated by a mineral filler. To test connect an ohmmeter between sheath and one of the signal leads. The resistance will be extremely high (probably beyond the range of the DVM) if ungrounded, and very nearly zero (a couple of ohms at most due to the wire resistance) for a grounded TC. Note that a physically abused ungrounded TC can become grounded if the sheath has been badly bent, crushed, etc. I'm interested to know why mixing the TC types could cause a problem. I know there can be ground loop problems when using grounded TCs on controllers with single-ended front ends, and would guess that if the controller can handle grounded TCs at all means the front end for each channel is a differential amp. If that's the case, then why would mixing grounded and ungrounded matter? Bob
Bob is correct in the first two paragraphs and then asks a question that I'll take a stab at answering "Why do ungrounded TCs matter?" This is one of those ground loop type problems. In theory they don't happen because all "grounds" have the same potential. As we all know, most every ground in every facility has its own voltage level with problems arising only when the "ground" potentials are different enough to interfere with the signal or the controllers interpretation of the signal. To understand them, you have to leave the ideal world and enter the real world...... Even if the controller input is differential, a grounded TC can cause problems if the common mode voltage of the TC (the voltage between the "ground" at the controller input and the "ground" at the TC) exceeds the rating of the controller. Since many ground loop problems are transient, the symptom will often be a TC with occasionally bad readings which occur when the common mode value has (temporarily) been exceeded and the controller input gets confused. Of course, many times, the problem doesn't cause bad readings, only inaccurate ones. These can be tough to detect/correct. Of course, if the common mode voltage is high enough, the controller input can be damaged or completely destroyed. These usually attract enough attention to get fixed. If several ungrounded TC are connected to the same controller, then the common mode voltage that must be considered is that contributed by each TC individually. Thus, the differential voltage that can cause problems is maximum difference between any one of all the several TCs as well as the controller. If there are many TCs, it takes and exceptional facility to be electrically "quiet" enough to not have at least some problems. Grounded TCs are cheaper, less fragile, have faster response times, can be used with single ended controller inputs, and always work in a perfect situations where all "ground" potentials are the same. So they often get used. Due to the insulation material and related mechanical and manufacturing issues, ungrounded TCs are more expensive, more fragile, have slower response times, need differential controller inputs, and are needed only in real world situations. As for mixing them, I'll bet there is something about how the front end is put together that causes the problems. For instance, if one side of the differential inputs are all tied together when grounded TCs are used (since they are all "ground" anyway), then any differential TC will have one side tied to ground and the other to the now single ended input. In a perfect world, that is OK, but the difference between grounding at the end of one lead vs at the junction will cause its own set of problems. My $.02 worth. Or less. Steve Cliff My opinions. Not my employers or anyone else's. I'll share 'em with you, but you can't have them. :)