TC Reading Fluctuating Badly

Loose connection to the thermocouple, poorly controlled heater, other failing electronics, lack of troubleshooting skills and tools. Take your pick or add your own.

Possibly a bad thermocouple.

Could be. Possibly, it could be one of those at post #1 also.

Take one more TC Measure same point simultaneously with both and see if both are showing same trends or not. If yes, your temp source is fluctuating and if no,your TC is not OK.

Another possibility is a badly-fluctuating process temperature.

You mean somebody may have not closed the door.

Without details about your temperature control system, this appears to be a process that is out of control rather than a defective thermocouple. If this is the PV of a PID conatroller, It appears to be badly out of tune with the paroces dynamics. Another possibility is the placement of the thermocouple giving excessive lag between the measurement and the process input.

I assume the issue is the ~45° jitter from 740° to 785°, centering around ~760°.

It is not likely a loose connection for a thermocouple measurement. Almost all commercial thermocouple inputs have a burn-out sensing circuit that would drive the indicated temperature value upscale or downscale on detection of even a momentary open circuit. With the exception of a single truncated down spike on Thursday afternoon, the jitter does not max out or drop to some minimum like 0.

"Bad" thermocouples typically exhibit drift that tends to move in one direction, either positive or negative, away from a true (unpolluted) reading to what/whererever. A bad thermocouple might center the up and down jitter around some point other than ~765° with because of the its offset drift, but the oscillations are not typical of a bad thermocouple in a stable process.

There can be dozens of reasons why the temperature control is not straight-line, but I don't think it the thermocouple itself is the problem. I don't think you have to check it with another thermocouple, I think those others who have pointed out process control issues have the correct approach.

I am returning to this post to make an important point. What you have been measuring is the temperature of the hot junction of your thermocouple which is not necessarily at the same temperature as the medium that you want to measure.

First, look at the location of your thermocouple. Consider the effect of thermal lag, that is, the time it takes your TC to respond to a temperature change in your process. Also consider if your TC could be effected by temperature changes external to your process. As an example from my own experience, a TC was mounted on the outside surface of a pipe. Not only was it effected by the temperature of the medium inside the pipe (what I wanted to measure) but also the outside air temperature and flow. When someone opened a window the wind blowing around the pipe would cause the measured temperature to drop.

If you manually switch the power off to the heater then monitor the TC output and if fluctuation are gone then the source may be common mode high AC voltage into thermocouple amplifier. If it still persists then sure thermocouple is not working properly or amplifier is not working properly.

Check the continuity of the thermocouple and you may find just <10 ohms in continuity and if more than 1K appears on the multimeter then Tc is open or loosely connected.

The problem occurs after we install new TC.

1.Possible cause: defect TC

Action: Remove TC and calibrate at workshop.Test TC at 800Degreec for 1 hour

Result: No issue. TC stable at +/- 3 degreeC /1hr

2.Possible cause: differ brand TC

Action: Remove TC and use the previous brand.

Result: No difference. reading still fluctuate

3.Possible cause: Ground loop

Action: remove screen cable at JB

Result: reading become more fluctuate

4.Possible cause: Temp effect at extension nipple

Action: insulate with insulation tape

Result: No difference. reading still fluctuate

5.Possible cause: Lightning Arrester faulty

Action: replace new

Result: No difference. reading still fluctuate.

6.Possible cause: Genuine Process issue

Result: the other TI reading at upstream and downstream of process is stable.

Some possible causes that we not yet verify

1.if temp element is not properly insert into thermowell? say the insertion of element is half or 3/4 of thermowell? Can this lead to reading fluctuation?

2. defected cable (from TC head to JB). How to check the integrity of the cable?

Cause could be poor grounding of screen. The ground itself may be bad. You can check integrity of cable with a multimeter on resistance range. Position of thermo-well insertion can not be a cause.

Apit7,

If you suspect that the problem is the cable,

1.- Put a millivolts generator directly on the JB terminal strip by five minutes and check the trend.

then

2.- Put a millivolts generator directly on the TC head ( I mean on the wires going to the JB) by five minutes and check the trend.

In this form you can see (maybe) where is the problem,

It could be that you have problem in the wire, JB terminal strip, TC Input Card Slot dedicated to that Thermocouple, ground loop, TC head connection problem ( I mean a wire touching the inside metal of the head).

How much mV on the generator? It would be depending of your TC type (K, J, B)

Regards

JP

After action #3 did you replace the screen cable. If you removed it and it fluctuated more then investigate further.

As we do not know the process conditions point #6 is misleading at best.

Have you tried putting the control loop in manual, is the source of heat an electric heater,try shutting it off and see if temp. degrades at a normal rate.

When looking at someone else's trend we cannot see the trend setup such as sample rate filtering etc. As the temp. is not going either very hi or low it does not appear to be a TC going open.

The insertion length will cause a lag or a slow reaction to a change in temperature.

Again if you can turn off heater and see reaction.

Like all upsets the process operators blame the instrumentation and the Inst Techs. blame the process or operator error.

It you look at the graph properly, the fluctuations are about 40C and there are both low frequency and high frequency components in it. These fluctuations are occurring in days and not in minutes and seconds. They look like to be load injection related and load injection can be fine tuned to have minimal fluctuations.

If you have a chance to insert another TC then it will also see the similar pattern and this ensures that TC is OK.

PID controller tunning may help if you can increase energy step per oC temperature difference or use proportionate energy to load and load changes. I think the volume inside the reactor or heated chamber is fluctuating too much to about 5% of the existing load and is injected uncontrolled very rapidly in short durations of few minuted. It can be more averaged proper control and perhaps 1% of the existing load for the response time of the temperature control performance of the system. Problem will be more serious when load is very little and injected load is comparable in volume. Fully loaded system with small changes will not show such temperature deviations.

I make two conclusions here for this problem.

1) It is not a problem of TC at all.

2) It is purely related to the load fluctuations. It may also be due to improperly PID settings and un-tuned PID for the system.

>1.if temp element is not properly insert into thermowell? say the insertion of element is half or 3/4 of thermowell? Can this lead to reading fluctuation?

When a thermocouple is not inserted fully into a thermowell, the result is a thermal lag, a longer time period for the transfer of heat to or from the sensing element to the process. Thermal lag does not appear as 'noise' jitter in the temperature readings.

>2. defected cable (from TC head to JB). How to check the integrity of the cable?

a) There are cable checkers that use TDR technology (time domain reflectometery).

b) Wire resistance
Thermocouple extension wire has a fairly high resistance. Calculate a value for what the resistance should be - distance from receiver to head * wire resistance * 2. Then disconnect the extension wire from the thermocouple head and twist the ends together, and measure the resistance from the receiver end, and compare what you calculated wire resistance should be.

There's a caveat: the twisted end is now a junction generating an EMF along the temperature gradient from the head end to the receiver end. That generated EMF adds or subtracts to the resistance reading, which uses a voltage source in the ohmmmeter to make the reading. So the closer you can make the 'hot' end temperature to the 'cold' end temperature, the closer you get to a true resistance reading, because an isothermal thermocouple (no temperature gradient from hot end to cold end) generates 0.0 EMF, so it doesn't distort the resistance reading.

What you're looking for is a resistance value much less than calculated, indicating a false junction somewhere along the extension wiring.

If the T/C extension cable's insulation was damaged and one lead is now touching a ground (like conduit), this could be a ground loop.

3. If appears as if the measurement point worked well at one time, then the thermocouple was replaced.

I'm guessing that the thermocouple was replaced because of this jitter noise issue. Is that true?

But the jitter noise issue did not go away when the thermocouple was replaced?

What about a flakey analog input? Have you moved the thermocouple circuit wiring to a different analog input to see if the noise follows the wire, to make sure that the analog input isn't flaking out?

Electrical noise - Joshi mentioned screen grounding. Does the extension cable run in a wire tray that has motor leads or power cabling?

I'll bet you that somebody keeps opening a window...