By far the most common catastrophic failure in all of them is an open-circuit. A robust system will detect that.

Failing calibration is a whole can of worms.

T/C would be drifting as the metals crystallize. Or a circuit problem causing leakage currents AC or DC to get to the analog front-end.

RTD would likely be poisoning from exposure to some process chemical. Vibration putting stress on the element which causes it to act as strain-gauge as well as a temp sensor.

Thermistor just drifting because they like to and they're cheap. They also tend to have high resistances because that makes for easier analog front-ends, unfortunately it also makes for easier drifting.

Keith Cress
kcress - http://www.flaminsystems.com

1. I can only add to Keith's list - Thermocouple Drift

The higher the temperature, the faster the pollution of the hot junction with metallic ions other than the thermocouple alloy so the thermocouple still produces an EMF, but the EMF no longer tracks the established, recognized EMF vs temperature tables, that is, the EMF has drifted from what it should be at any given temperature to something else. There is no correction factor or even a known direction of drift. Sometimes it's up sometimes it's down.

The industries involved in critical heat treatment of aircraft and military components have to comply with NADCAP requirements that limit exposure times and number of exposure cycles of work or control thermocouples at higher temps.

There is a mineral insulation as an alternative to MgO which is designed to be super dry (typical MgO absorbs water moisture which enhances ion transfer at high temps) which claims to delay the onset of drift.

2. 1988 was the last time I saw a field calibration done that actually measured the temperature of the thermocouple input terminal blocks and compares that to the 'ambient temperature reading' of a short across the input terminals. In my opinion, there's a whole lot of controllers out there with lick 'em & stick 'em' cal stickers that say 'passed' but the accuracy of the cold junction measurement is complete unknown.

Hi itsmoked,

By poisoning of process chemical for RTDs, do you mean when liquid reaches the lead or sensor wires? I thought these were protected by some sort of seal and are only exposed when the seal cracks.

Thanks!

Still an intern

Hi ipocoyo; Nothing will protect the elements long term with the exception of, say, a stainless well or something similar. Any 'coatings' will eventually let something thru. It can be reagents or whatever chemicals the sensor is exposed to. Very often it's gases that can penetrate the coatings.

It's not very likely in normal applications for the leads to get involved in the problem other than the frequent mechanical failure due to vibration or flexing. You could have a probe in a vacuum environment and literally suck surrounding gases down the inside of the wire jackets. RTDs are normally plated onto ceramic substrates then coated with something for protection. If the coating and the ceramic are good at protecting from the sensed environment you shouldn't have much problem.

What specific environment are you working with?

Keith Cress
kcress - http://www.flaminsystems.com

"Both RTDs and thermocouples are simple devices, but problems such as calibration drift and
degradation of response time are still encountered in their application in industrial processes. The quality
of the seal is an important factor in the reliability of RTDs. Moisture can intrude into the assembly if the
seal proves faulty, cracks, or otherwise allows moisture to enter the RTD. Moisture is detrimental to an
RTD in several ways. First, it can reduce the effective resistance of the RTD and cause the sensor to
indicate a lower-than-true temperature. Secondly, moisture can cause noise at the RTD’s output and/or
result in erratic RTD output. Furthermore, moisture can cause the RTD internals to interact chemically.
This could damage the sensing element or cause the platinum wire to thin, thus increasing its resistance
at a given temperature. Temperature drift will be the result."

Hashemian, H. M. "Sensor Performance and Reliability", Research Triangle Park, NC: ISA—
Instrumentation, Systems, and Automation Society, 2005.