Help with thermocouple confusion

Hi.

I am really confuse with the result of the inspection we made on our client's molding machines. To give a brief history, they called us complaining about irregularity of temperature reading on some of their machines (molding machines). Each machine have 2 thermocouples, 1 going to a paper recorder whilst the other to a PLC. Apparently, the reading on this two recorders are different, eg., Machine 1: paper recorder reading 164 C, PLC reading 171 C. We took actual temperature via a thermo laser, it reads 169 C. Also the set-up is that the machine is nearer to the paper recorder than to the PLC (I'm not sure if this has something to do with it).

Here's what we did:
1.) Since the user does not know the exact thermocouple type they are using, we conducted a resistance test. To check whether wire distance is a contributor to the error or not, we made a two-point check point; one is nearest to the thermocouple , the other is farthest. SEE CONTINUATION>>

>>CONTINUATION HERE:
Machine 1: On the nearest, we read 11.06 & 8.5 Ohms. On the farthest, we read 9.8 & 9.3 ohms respectively.
Machine 2: Temp (PLC) = 166 C, temp (paper recorder) = 166 C, temp (thermolaser) = 164 C. On the nearest, we read 11.3 & 11.6 Ohms. On the farthest, we read 30.6 & 19.7 Ohms.

QUESTIONS:
1.) Based on the data above, it seems that we are testing it on a wrong foot. The consistent temperature reading of [166 C vs 166 C] machine 2, given 11.2 vs 30.6 Ohms (difference = 19.4 Ohms) & faulty reading of [171 C vs 164 C] machine 1, given 11.6 vs. 9.8 Ohms (difference = 1.8 Ohms). What may we have done wrongly or what must be rightfully done with this matter?

Please help.

Sorry I forgot to add that the sensor is being spliced with an ordinary 2-core wire. Does this contribute?

theripley said:

Sorry I forgot to add that the sensor is being spliced with an ordinary 2-core wire. Does this contribute?

Click to expand...

DOH! It definitely doesn't help your cause. Where is this splice an how temperature stable is it?

Also, did anybody ever calibrate the T/C? Use the old melting glass of water trick to adjust the offset of your readout.

Since the user does not know the exact thermocouple type they are using,

Click to expand...

what type plc? some plc's you se the thermocuple card with a dip switch
is your tc wire red and white type j or red and yellow type k both have differnt scales
maybe this link could be helpfull : http://www.omega.com/techref/thermcolorcodes.html

It may help to understand HOW a thermocouple works.

Basically, a junction of any two dissimilar metals will create a voltage that is proportional to the temperature.

Sorry I forgot to add that the sensor is being spliced with an ordinary 2-core wire. Does this contribute?

Click to expand...

So, where you've spliced the thermocouple wire with ordinary (copper?) wire, you've created new thermocouples, which are called "cold junctions". These can and do create unreliable inconsistencies in the temperature readings. If you need to extend the length of the thermocouples, you need to use thermocouple extension wire of the same thermocouple type. AND this also applies to any terminal blocks where they may be landing in any kind of an intermediate junction box...

first let's cut to the chase ...

Sorry I forgot to add that the sensor is being spliced with an ordinary 2-core wire. Does this contribute?

Click to expand...

smart money says that's where the discrepancy in readings is coming from ... that situation needs to be corrected ... here's a recent thread that might help with the basic ideas ...

http://www.plctalk.net/qanda/showthread.php?p=506759&postcount=1

basically, the points where the "unmatched" (non-thermocouple) wire is spliced into the circuit form EXTRA sensing points (extra "artificial" thermocouples) in your circuit ... if the temperature at those extra sensing points is DIFFERENT from the REAL sensing point, then your signal will have an OFFSET ... in some cases, this offset will not be enough to affect the manufacturing process – but in other cases the offset might cause significant problems ...

the best course of action is to replace the ordinary wire with "thermocouple grade" extension wire ... I recommend that you do that before you go much further ... you'll also want to use special "thermocouple" type terminal blocks for the most reliable results ...

now some random thoughts on what you've posted so far:

paper recorder reading 164 C, PLC reading 171 C. We took actual temperature via a thermo laser, it reads 169 C.

Click to expand...

so the customer is complaining about a "reading" difference of 7 degrees between the two display devices ... is that correct? ...

your reading with another (presumed accurate) instrument shows a temperature about MIDWAY between the two displays ...

to be honest, that's probably not enough difference to worry about – but – we're going to accept the fact that since the CU$TOMER is worried, then we need to go further ...

I'm reminded of the old adage: a man with ONE watch always knows exactly what time it is ... but a man with TWO watches is never quite sure ...

here's some light reading for you ...

http://www.omega.com/thermocouples.html

and here's a question: how is the manufacturing process working? ...

(a) if it's producing good product then the customer is just concerned about the "displayed" temperature readings ... in that case, maybe a little bit of calibration on the chart recorder could make the two readings match ...

(b) if the manufacturing process is less than optimal then maybe you need to work on the CONTROLLING end of things (presumably the PLC) and worry about the chart recorder's displayed value later ...

basically are we dealing with something that REALLY does need to be "fixed" ... or are we just trying to deal with an operator's discomfort at seeing TWO different readings for ONE actual temperature? ...

continued ...

Also the set-up is that the machine is nearer to the paper recorder than to the PLC (I'm not sure if this has something to do with it).

Click to expand...

the real question there is:

what is the room temperature at the chart recorder – compared to the room temperature at the PLC? ...

the trick about a thermocouple is that it only works based on the DIFFERENCE in temperature between the "hot junction" (where the probe is located) and the "cold junction" (where the meter/display/instrument is located) ...

an illustration:

suppose that you have a thermocouple measuring device – and you put the probe into an oven ... the oven's temperature is 200 degrees F – and the meter accurately displays that temperature for you ... good so far ...

now suppose that you put the ENTIRE measuring device (meter, probe, all of it) into the oven – and watch the meter indication through the oven's window ... here's the trick ... the display will now read ZERO ... that's because the "cold junction" (meter end) of the system is exactly the same as the "hot junction" ... specifically, since there is no DIFFERENCE between the ends of the system, you'll get ZERO reading on the meter ... yes, the PROBE still sees 200 degrees F – but since the meter end of the line also sees 200 degrees F there is no DIFFERENCE – so no signal is produced ...

so ...

suppose that (everything else being equal) your customer's probes are measuring the temperature – but – suppose that the chart recorder is located in a COOL place – and the PLC is located in a WARM place ... in that case, you could certainly expect to see a difference reflected in the two temperature displays ...

but – (do NOT skip this part) ...

many (most?) thermocouple devices have a feature for Cold Junction Compensation (CJC) ... in basic terms, this system "takes the temperature" of the meter's/module's "room temperature" surroundings, and then applies a correction factor to the reading from the probe ...

so ... if both your PLC's thermocouple module AND the chart recorder BOTH have this Cold Junction Compensation – AND – if those features are working correctly, then the "room temperature" effect should be negligible ... (be sure to notice the IF statements included above) ...

Since the user does not know the exact thermocouple type they are using,

Click to expand...

now that has GOT to be remedied before you can go much further ... you've GOT to know what type of thermocouple you're using – because different types give substantially DIFFERENT readings – even when sensing the SAME temperatures ...

in some cases, the colors of the insulation on the thermocouple is enough to tell you what type it is ... search in the Omega links that I gave you above for the "color charts" ... but ... in older systems, the heat has discolored the insulation – so the colors might not be recognizable ...

here's an old-timer's trick ... see if a magnet will "draw" one of the leads on the thermocouple ... if it does then the type is PROBABLY (but not certainly) a type J ... that's because a J type's positive lead is made of iron – and will "draw" a magnet ...

continued ...

we conducted a resistance test.

Click to expand...

that probably did not accomplish much ... the reason is that the thermocouple was probably generating some amount of voltage – and that "extra" voltage would have affected your resistance reading ...

TIP: if you get a chance to do the resistance test again, experiment by taking a reading – and then REVERSING the ohmmeter's leads – and see if the resistance reading changes ... depending on the temperature of the thermocouple probe, you could get a SUBSTANTIAL difference in the ohms reading ...

What may we have done wrongly or what must be rightfully done with this matter?

Click to expand...

the resistance test is probably not going to get you where you need to go ... it COULD – but only under carefully controlled temperature conditions ... on the other hand, did you happen to check the MILLIVOLTS coming from the thermocouple? ... I don't think that's going to help you much either – but it would have made for a more relevant "multi-meter" type test – since the thermocouple works by producing a MILLIVOLT signal (not resistance) ...

now some more questions:

you told us that one of the devices is a PLC – but you didn't tell us how the thermocouple signal is being connected to the PLC ... are you using a special THERMOCOUPLE module? ... or is the thermocouple connected to a transmitter – and then that transmitter's output is being connected to the PLC? ... for example: a 4 to 20 mA signal is often used ...

either way, the "device" that the thermocouple is connected to MUST (I repeat MUST) be properly matched to the type of thermocouple (J, K, etc.) in order to get good reliable results ...

so ...

when you say that the customer doesn't know what type of thermocouple is being used – then clearly THAT question needs to be answered (totally nailed down) before you can assume that the PLC, the chart recorder, or any other device in the circuit, (example: a 4 to 20 mA transmitter) can be set up correctly ...

keep in mind that:

(1) if a thermocouple transmitter is used, it will almost certainly have "calibration" screws to be adjusted ...

(2) if the thermocouple is connected directly to the PLC input module, then that module will almost certainly have some type of "setup" to configure it for the exact type of thermocouple being used ...

(3) your chart recorder almost certainly will have some means of configuring it for the exact type of thermocouple being used ...

from where I sit, it boils down to whether this is a manufacturing problem that actually needs to be "fixed" – or just a situation where an operator/customer is uncomfortable seeing two different "readings" for just one temperature ...

I hope this is helpful ... good luck with your project ...

My money's on the junction box connection - it's probably as a case where copper wire is run instead of thermocouple extension wire, or the wrong type extension wire, creating an intermediate junction.

The configured thermocouple types are probably correct for each device, otherwise the error would be greater.

Cold junction compensation could be really out-of-whack for either, but that's a rare circumstance.

I doubt the circuits include a temperature transmitter. The plastics industry rarely, if ever, uses them. They always use thermocouple-direct connections.

Ron's right about the resistance measurements - useless.

There is one factor that hasn't been mentioned. The plastics is notorius for using grounded thermocouples. There could be a ground loop that appears as an offset in either of the readings. Substituting an ungrounded thermocouple would show a difference in reading for the same machine temperature.

Checking the resistance from either T/C lead (right at the thermocouple with the lead wire disconnected) to the machine should reveal whether it is a grounded thermocouple or not.

Clarifications:
Plastics Industry. Most likely type "J". Black connectors. Wires are red and white. White lead is magnetic and positive.
Other common type is "K". It also has a magnetic lead. Yellow connectors. Wires are yellow and red. Red is magnetic and negative.
See Omega chart.

Cold Junction. There is 2 cold junctions at the screws of the module that is unavoidable as we transition to copper for our electronics. The compensator is used to measure the temperature of the screws, and generate an offset to compensate for the error. Reference.

Use a portable T/C meter, or adapter for your Fluke. You can't troubleshoot this without it.

NOTE:

The color codes being referenced by a few here are US/Canadian standards. There is no true universal standard for this and as such you can't trust that alone to determine what you have.

DamianInRochester said:

NOTE:

The color codes being referenced by a few here are US/Canadian standards. There is no true universal standard for this and as such you can't trust that alone to determine what you have.

Click to expand...

FYI

The link provided by keithkyll does show some international and various countries color coding.

http://www.omega.com/techref/thermcolorcodes.html

I thought my link showed that too! lol

Was referring to the text in the posts. Not the links.

Sorry I forgot to add that the sensor is being spliced with an ordinary 2-core wire. Does this contribute?

Click to expand...

You bet it does! First find out what type TCs are installed (don't depend on the wire color because if these guys used copper wire, then they may have used the wrong TC extension wire also), then order enough themocouple extension wire of that type (probably J or K) to go from the TC to the PLC without ANY splices in between.

It is unlikely that you will ever find two thermocouples that will read the exact same temperature. There are too many variables and errors to allow that. If you can get two to read the same temperature within +/- 3 degrees, that will be about the best to expect.