Mitsubishi Q Series: PLC over-reading K-type thermocouples. Why?

[Mas01]

Hi, Today I connected up a previously spare QD68TD-G module, configured for inputs from eight K-type TC's.
However, when I connect a Fluke to simulate e.g. 10.0 Celsius, with the PLC in Monitor mode, it reads 14.6 Celsius. i.e. the PLC over-reads by +4.6 Celsius. At other Fluke settings, the difference is consistent.
Anyone know why? Or how I can make the correction in the code?
Thanks

 

[parky]

When you say you simulate it using a fluke is the fluke a K type temperature simulator ? I didn't know flukes did a simulator. or are you just comparing a sensor connected to the analog card where both the fluke & K type tc are compared i.e. in water. I would not always rely on meters to give accurate measurement unless it specifically designed for such, I purchased a sensor to fit on my meter (Hantek) and it was a couple of degrees out but then again I had a dry well calibrator where I could test sensors with an accuracy of +- 0.05 Degrees C. anyway, if the difference is always say 4 Deg c just take the value in your "D" register and add or subtract that from it and store it in another register.
There is also a way of setting the zero & span in the card by writing to the buffer memories but it is rather complex. I suggest you get a kettle & boil it, put in the probe & see what it reads, then get a bucket of water with Ice then check it's low range as well.

 

[Mas01]

When you say you simulate it using a fluke is the fluke a K type temperature simulator ? I didn't know flukes did a simulator. or are you just comparing a sensor connected to the analog card where both the fluke & K type tc are compared i.e. in water.

The fluke 754 can simulate several types of TC, k-type being just one of them. This is the one I'm using...

 

[parky]

Thermocouples are not very accurate at low temperatures normally you would use Platinum resistance probes a K type TC from 0 Deg C will only have an accuracy of +- 2.2 Deg C so it is possible that with both the fluke & the PLC this could lead to +- 4 odd degrees. What is that internal reference at the bottom on the Fluke of 14.9 Deg C is it actually reading close to the same as the TC connected to the PLC ?

 

[Mas01]

When you say you simulate it using a fluke is the fluke a K type temperature simulator ? I didn't know flukes did a simulator. or are you just comparing a sensor connected to the analog card where both the fluke & K type tc are compared i.e. in water. I would not always rely on meters to give accurate measurement unless it specifically designed for such, I purchased a sensor to fit on my meter (Hantek) and it was a couple of degrees out but then again I had a dry well calibrator where I could test sensors with an accuracy of +- 0.05 Degrees C. anyway, if the difference is always say 4 Deg c just take the value in your "D" register and add or subtract that from it and store it in another register.
There is also a way of setting the zero & span in the card by writing to the buffer memories but it is rather complex. I suggest you get a kettle & boil it, put in the probe & see what it reads, then get a bucket of water with Ice then check it's low range as well.

Thanks, I see what you mean. So for the fudge-factor I could do something like this, which will reduce the PLC value by 4.4 Celsius.:


u13\G10.4
|---||-----------------[+ u13\G15 K-44 D98 ]

 

[Mas01]

Thermocouples are not very accurate at low temperatures normally you would use Platinum resistance probes a K type TC from 0 Deg C will only have an accuracy of +- 2.2 Deg C so it is possible that with both the fluke & the PLC this could lead to +- 4 odd degrees. What is that internal reference at the bottom on the Fluke of 14.9 Deg C is it actually reading close to the same as the TC connected to the PLC ?

Oh I never noticed that 'internal ref' reading before. I'll have to read the manual and find out what it means.

 

[parky]

Yes that should work, I'm still a bit perplexed by the temperature shown on the fluke at the bottom "Int Ref 14.9 Deg C" that seems to be close to what you are reading, what does that refer to is it compensated or not compensated reading ?

 

[lfe]

The simulator is generating a negative voltage since you have set it to simulate a temperature lower than ambient (14.9ºC in the simulator)ç

Perhaps the Mitsubishi module does not take into account negative voltages and reading never drops below the ambient temperature, try to simulate for example 20ºC

Check that the cable connecting the simulator and the module is the correct compensation cable for type K thermocouple and with the correct polarity.

Thermocouples do not measure an absolute temperature but the difference between the cold and the hot junction, and a thermocouple that can measure temperatures up to 1200ºC, like the K type, will not be very accurate when measuring room temperatures.

 

[Mas01]

Perhaps the Mitsubishi module does not take into account negative voltages and reading never drops below the ambient temperature, try to simulate for example 20ºC...

I simulated 20, 30, 40, 50, 60, 70, 80, 90 & 100 Celsius.
In each case, the PLC overrread by +4.6 Celsius.

 

[Mas01]

Yes that should work, I'm still a bit perplexed by the temperature shown on the fluke at the bottom "Int Ref 14.9 Deg C" that seems to be close to what you are reading, what does that refer to is it compensated or not compensated reading ?

I'll be honest - I don't know much about CJC, other than it's a way of compensating for changes in metal conduction at junction boxes (or something like that).
Just like in this page from the Q68TD spec, the terminal strip in our PLC also includes an RTD. Does that mean the reading is compensated?

 

[lfe]

If you measure the mV at the terminals on the module you are measuring the difference between the millivolts corresponding to the temperature of the hot joint of the thermocouple and those corresponding to the temperature at the terminals.

If the terminals were at 0ºC then you would directly measure the mV of this table but since this is not true then the module measures the temperature and compensates for the difference.

 

[danw]

Internal reference is the temperature of the connector, which is the cold junction temperature whose mV equivalent gets 'added' to the thermal gradient mV reading (Hot junction mV measured at the connector terminals).



I doubt the CJ or reference temperature is a contributor.


I assume that the green plug and green wire are for an IEC Type K T/C? Green insulation is for a noble metal (Pt), Type R/S in the US.

 

[Mas01]

I assume that the green plug and green wire are for an IEC Type K T/C? Green insulation is for a noble metal (Pt), Type R/S in the US.

Yes the green plug and cable is a K-type TC cable. 2-core, white -ve, green +ve. The sockets on the Fluke are also labelled accordingly so you plug it in the right way round.

 

[parky]

The reason for the RTD is to compensate for the Cold junction at the terminals.
The cables internally in the card (or the copper track of the circuit board) obviously are not the correct metal as the thermocouple cable so this is an attempt at compensating for these connections. Connecting different cable to the K type cable creates what is in effect a second thermocouple and has greater effect at low temperatures.
What are you measuring and how critical is it for example what temperature range and will it matter if it varies by a few degrees.
If the answer is it is just an indication of the temperature and could have a wide range without causing a problem i.e. +- 5 Deg. C then you could ignore it and just offset the reading the same way as in the other channels.
To use the RTD you would need to set values in the buffer memories this is a little complex and is done on the first scan of the PLC just as already in your code for example, you already have some code that on a one shot sets the conversion of the TC channels, it is a matter of setting the card into program mode, transferring the values to the buffer memory and checking no errors and set the complete output bit. As I said this is rather complex operation.
I have never done this I have always used the GX Configurator software and to be honest never used Thermocouples on most of the plant I have been involved in as RTD sensors are fare more accurate in the temperature ranges we use i.e. -20 to 200 Deg. C.

 

[Mas01]

What are you measuring and how critical is it for example what temperature range and will it matter if it varies by a few degrees.
If the answer is it is just an indication of the temperature and could have a wide range without causing a problem i.e. +- 5 Deg. C then you could ignore it and just offset the reading the same way as in the other channels.
To use the RTD you would need to set values in the buffer memories this is a little complex and is done on the first scan of the PLC just as already in your code for example, you already have some code that on a one shot sets the conversion of the TC channels, it is a matter of setting the card into program mode, transferring the values to the buffer memory and checking no errors and set the complete output bit. As I said this is rather complex operation.
I have never done this I have always used the GX Configurator software and to be honest never used Thermocouples on most of the plant I have been involved in as RTD sensors are fare more accurate in the temperature ranges we use i.e. -20 to 200 Deg. C.

I'm measuring water temperature. Cooling water temperature. These temps are monitored and used as setpoints for alarms, if the water's getting too hot. It's used for cooling various pumps/motors at the facility.
For info, the read-out of the 'int. ref' temperature is 17.2 Celsius. It must be related to ambient temperature because the room is warmer today.
Yet more info...there are 2 terminal blocks. When I connect the fluke to the TB next to the PLC, the PLC reads the correct value. When I connect the fluke at the Junction box terminal where the TC's will eventually be connected, the PLC over-reads (today it read 17.6 Celsius). The wire from the PLC's TB to the next TB, is probably copper.


Edit...I think I need to move the RTD from one TB to the other TB...possibly. (?)