RTD or Thermocouple for Bearing Temperature


Thread Starter


Is there any recommendation or standard determine which temp. element to be used for measuring bearing temperature? Which is preferred RTD or thermocouple and why?
There are advantages to both types, and disadvantages as well. I've seen both types, though I more commonly see T/Cs used in the last 20 years.

RTDs are, I've read and been told, better for low temperature applications, while T/Cs are better for higher temperature applications.

RTDs require a power source and a rather complicated measuring system.

T/Cs are self-powered, require cold junction compensation at some point in the circuit and since the T/C output is not usually linear with temperature there is also some "complicated" measurement system required.

In today's industrial environments a LOT of manufacturers make the decision about which type of measurement/monitoring system to use based on one of three criteria:

1) previous history of usage; if they've been using RTDs for some time, they're predisposed to use RTDs, and if they've been using T/Cs for some time, they're predisposed to use T/Cs

2) cost--pure and simple; in the absence of any other criteria the cost of one versus the other will likely be the deciding factor, and if they don't have to supply the measurement/monitoring system (cost, again) they will likely choose the least expensive type

3) customer specification--they will provide whatever the buyer wants and factor the cost into the proposal/price for the equipment.

In my opinion, I prefer T/Cs because they don't require a power source, have only two wires (better RTDs require a three-conductor, twisted, shielded cable per RTD, and the best RTDs require four-conductor, twisted, shielded cables per RTD (or two twisted, shielded pair cables per RTD)), and, because the magnitude of a bearing temperature is (to my way of thinking) not as critical as detecting a sudden change in temperature (rate of change).

Yes; a "high" bearing temperature reading is not good (in most cases), but I want to know how long the temperature has been increasing and how fast it has been increasing to whatever "high" temperature it is currently at. Good equipment operators don't just look at magnitude (like many control systems do)--they monitor rate of change and use that as an indicator to "sound an alarm" before the magnitude reaches the alarm point. And, T/Cs can do that just as well as RTDs. Every bearing failure I've ever been asked to analyze was preceded by a change in temperature over time--sometimes it was weeks, other times it was hours or minutes. but had the rate of change been monitored and not just the magnitude it might have been possible to reduce load and try to take some other corrective actions before the bearing temperature reached the alarm/trip point and prevent more severe damage and reduce outage time.

Hope this helps!

I was told 30 years ago that API had a standard recommendation, but that it is only "enforced" by companies that choose to enforce it, and was really only used when someone is writing a specification for a piece of equipment. I don't know if that's true, but I do know that some companies have their own standard and, right or wrong, that's what they'll use. And the standard can change as soon as the person-in-charge changes.
RTD's provide better coverage at lower temperatures since you are measuring resistance and not millivolts, but can be sensitive to mechanical damage.

T/C's tend to tolerate rough services.
I worked in mining for many years it was common to use 100 Ohm Pt 3 wire RTDs to protect the mill bearings and lube system, I don't think they are any more complex than thermocouples. Large motors often have 10 Ohm copper RTDs in the windings or sometimes thermistors.

Consider a multi-input RTD or Thermocouple analog module for the PLC.
You can use multiplexed modules because nothing is going to change between scans. This is much cheaper than providing 4-20 transmitters for each temperature input. I don't recall ever seeing transmitters used on bearings only process.

If you were short on wire the RTDs can share a common or could be wired 2 wire compensating for the various cable lengths in the PLC, I don't think they would be any less accurate than thermocouples.

Other points to consider.
Thermocouple break protection? Fail High Trip on under temperature as well as over temperature; this can provide additional protection in case of a power supply outage.
You don't need 4 wire accuracy for bearings.
Availability of multi-conductor thermocouple cable
In my experience in a large Utility Plant with large multi-stage high pressure BFW pumps, medium and low pressure pumps; motor winding temperature measurement used RTD's and bearing temperatures used thermocouples.
Roy Matson makes a great point--T/Cs require T/C-specific extension wire to connect them to the control system/monitor.

And, I agree, four-wire accuracy isn't needed for motor bearings.

Again, I don't think there's a real "standard" for this--it just depends on personal experience, personal preference (of the specifier) and cost (overall cost of installation and monitoring).

Both sensors are fine for most applications. I also think I've been told that smaller bearings more commonly use T/Cs, but that's a recollection from long ago which may or may not be true today.
Do you see any reason for this selection??

> In my experience in a large Utility Plant with large multi-stage high
> pressure BFW pumps, medium and low pressure pumps; motor winding
> temperature measurement used RTD's and bearing temperatures used
> thermocouples.
The classic very thin Platinum wire wound around a ceramic mandrel did not hold up well in vibration.

But I've MgO thermocouples break in vibration, too.

But I honestly do not know whether the thin film RTD element RTD's are better in vibration or not.
One consideration that hasn't been mentioned in this chain is measurement time lag. This would not impact bearing temperature or motor winding temperature monitoring, but if the measured temperature is used for control purposes, a small lag is frequently required. In this service, thermocouples are usually superior to RTD's.
I think it is:

-manufacturers preference and experience

-with magnetic fields and induced currents in a motor a RTD is probably better

-the right T/C may be more vibration resistant for bearings and also cheaper

that's all I got
Wire-wound sensors don't tolerate vibration stresses well at all, 5-10 G's at best.

T/C's with heavier wire gauges tolerate the G-force's better, but if the vibration levels cause it to collide with adjacent surfaces it too ill fail.

I've only seen bearing and winding RTD's used (copper, nickel, or platinum elements). Two wire RTD's were common as the distance for the connection wires was quite short. three and four wire rtd's are common where the resistance of the lead wires is significant.
There is a great article on motor bearing temperatures at


The typical area of concern in the article seems to be about the boiling point of water. The article also discussed how case temperature can be used with an adder as an alternative.

If the number of wires and vibration are a concern, at 100C another alternative to an RTD that can be looked at is a thermistor. The units shown at


can be adjusted to cover that range without a problem.

https://en.wikipedia.org/wiki/Thermistor has a great discussion of the thermistor as a reference.

Generally RTD Pt 100 (duplex/double element) was use for bearing temperature measurement due to RTD was better for lower range temperature than T/C.

cons: T/C element installation is required special cable depend on T/C type if the T/C design is separate from transmitter.