Thermocouple reading

[engr_joni_ee]

Hi,
I am looking for electronics circuit for reading K type thermocouples. I have looked at some references but not able to understand where we need to apply voltage to power them up. It just says that one terminal has to be on the common reference temperature point which at some places is referred as cold junction and then we just measure the voltage across the thermocouple but the question is where the biasing voltage come from ?

 

[betwixt]

Thermocouples are voltage generators, they do not have an external power source!

You could think of them as batteries that convert heat into voltage. The problem is the voltage is very small so it needs amplification to be useful and some voltage is produced even at temperatures below zero so you can't directly convert the voltage into a temperature. There is a predictable voltage produced per degree of temperature but no reference to compare it to. The cold junction compensation can be done by using two identical thermocouples, one for the measurement and one at 0C so the difference can be observed -or- more commonly an electronic circuit is used to simulate the voltage a cold junction would produce.

Brian.

 

[carpenter]

Thermocouple is voltge cource and any IC for TC use external BIAS voltage/current ,it helps, for example, to detect an open thermocouple, but in principle it is not necessary.
You do not specify the purpose of the circuit.
If for MCU, I recommend to use MAX31855 as lowcost solution for K or MAX31856 more precise 19bit solution for all type of thermocouple.

 

[danadakk]

An ap notes on the subject (single chip solutions) -

https://www.cypress.com/file/123806/download

https://www.cypress.com/file/137891/download

https://www.cypress.com/documentati...oc-4-and-psoc-5lp-temperature-measurement-rtd RTD

https://www.cypress.com/documentation/code-examples/ce210383-psoc-temperature-sensing-rtd

Right hand window shows you chip resources used/left. As you can see most resources available for
other design needs. ARM core in part as well. A $15 board would suffice to proto this design. Note
A/D in this part is good for 20 bits, and if you need to filter readings there is onboard gen purpose
DSP good for 120 taps or cascaded biquads, wizard is used to config it. Uses DMA, can run in background.
All onchip.


Bluetooth thermocouple -

https://www.cypress.com/documentation/code-examples/ce217643-ble-temperature-measurement-psoc-6-mcu-ble-connectivity

 

[engr_joni_ee]

Hi, I have a question on MAX31855. I am wondering how the VCC is controlling the resolution ? My actual temperature range is -150 C to +150 C and I would like to have resolution of about 0.25 C. What is the effect of long wire thermocouples with 3 meter on the accuracy of readings ?

In MAX31855, is it possible to improve the accuracy of thermocouple by narrowing the dynamic temperature range through biasing ?

 

[carpenter]

Hi, I have a question on MAX31855. I am wondering how the VCC is controlling the resolution ? My actual temperature range is -150 C to +150 C and I would like to have resolution of about 0.25 C. What is the effect of long wire thermocouples with 3 meter on the accuracy of readings ?

Power-Supply Rejection MAX31855 is -0,30 °C/V
Power-Supply Rejection MAX31856 is +0,15 °C/V
Resolution MAX31855 14bit 0,25°C
Resolution MAX31856 19bit 0.0078125°C

The length of the thermocouple no longer plays a role from the thermocouple principle.
The thermocouple does not measure the temperature at any point, but measures the temperature difference between the hot and cold ends of the thermocouple.
The thermocouple itself and measuring the exact temperature of the cold end is also usually the biggest source of errors when measuring the temperature with a thermocouple.
If you want to measure accurately, the MAX31856 is a better choice, but with a standard K-type thermocouple, the thermocouple error (aging, etc.) is always greater than the MAX31855 error. If you use a high-precision thermocouple and perform a calibration, the second largest source of error will be accurate cold-end measurement .

 

[FvM]

MAX31855 has fixed resolution of about 0.25 degree, not controlled by VCC or software.

Accuracy doesn't depend on thermocouple length, except for possibly increased interferences in a noisy environment.

 

[engr_joni_ee]

Hi,

I guess there should always be a cold end reference temperature which has to be at the same temperature all the time which is a challenge and that effect the accuracy. Then I guess RTD is better option to get 0.25 C precision for temperature range -150 C to +150 C.

Is there any 8 channel RTD current source ? or a IC which can read 8 RTDs ?

 

[betwixt]

But do be careful if you extend the wires to the thermocouple. They work on the principle of dissimilar metals in contact with each other causing a potential difference across the joint, the same applies to any other junction in the loop so if you change material at other junctions they too will produce some voltage. For example, the thermocouple wires are mostly chromium and mostly aluminum, if you solder them to a copper extension you get a transition to tin (solder) then to copper and both of those will introduce an error proportional to the temperature at the respective joints.

Brian.

 

[danadakk]

Hi,

I guess there should always be a cold end reference temperature which has to be at the same temperature all the time which is a challenge and that effect the accuracy. Then I guess RTD is better option to get 0.25 C precision for temperature range -150 C to +150 C.

Is there any 8 channel RTD current source ? or a IC which can read 8 RTDs ?

I tried the key issue for a PSOC, could I route all 8 channels (tool does that for you), and it did.

That would be for this basic design, to mux that to 8 channels -

So I think all this can be done in one PSOC. Note Vref for ADC is +/- .1%. If you need full 20
bit resolution than use external Vref for PSOC. Note window on right shows resources left
for other tasks. This has an ARM core in it for general purpose applications, a large digital
and COM and Display fabric. But part I did this in has limited remaining pins. There are parts
with more I/O. You would need to do an error budget and a quick check on RTD effect on
the IDAC compliance range. Which is in datasheet.

Tool (PSOC Creator) and compiler free. Board would be https://www.sparkfun.com/products/13714 or

https://www.cypress.com/documentation/development-kitsboards/cy8ckit-050-psoc-5lp-development-kit

Both boards have enough GPIO but you would have to check as there are sometimes committed to LEDs
and the like. So schematic check would be prudent. Note part I used had 20 GPIO left for use.

You could do a trial test of a couple of channels with this $ 10 board. Ignore the out of stock comment,
I think they are trying to push sales thru distributors. Its a very popular board, in fact has two PSOC 5LP
processors on it, one the target, the other a programmer/debugger for the target. There is already a working
project in the tool, all you would have to do is add muxes and a few lines of code to mange them and their
"channel".

https://www.cypress.com/documentation/development-kitsboards/cy8ckit-059-psoc-5lp-prototyping-kit-onboard-programmer-and

I attached the component list for the part. In PSOC language a component is an onchip resource.


Regards, Dana.

 

[engr_joni_ee]

Hi,

I still have a question regarding MAX31855. Does it also contain the reference temperature sensor inside to read the T- ? I have looked at the output data format of MAX31855, it actually output both T- and T+. If I use MAX31855, do I need to provide any temperature controlled surface and read that through another reader for T- ?

I guess the price of MAX31855 and MAX31856 is almost the same. The accuracy and precision of MAX31856 is better than MAX31855.

How about if I use T type thermocouple instead of K type thermocouple. Does it improve the accuracy within 1 C ?

 

[danadakk]

Does it also contain the reference temperature sensor inside to read the T- ?

There is an internal reference but datasheet does not spec it directly, rather its error tied
up in the overall Temp accuracy specs. There are graphs vs accuracy in C, seems to indicate
over its range the accuracy maybe 8 bit......if I read DS correctly. Shows just cold junction error
+/- 3%.

the 31856 seems like overall accuracy better than 31855. Cold junction error still significant though.

A complete end-to-end error budget calculation would include noise, mux errors, as well as INL, DNL, offsets....


Regards, Dana.

 

[engr_joni_ee]

Hi,

I don't understand the term cold junction. I am wondering do I need to provide a maintained temperature reference surface for T- outside the chip MAX31856 ? or I just need to connect thermocouple's terminal T+ and T- to the chip and trust the reading that comes out even if the cold junction reference temperature changes ?

How about if I use T type thermocouple instead of K type thermocouple. Does it improve the accuracy within 1 C ?

 

[KlausST]

Hi,

I am wondering do I need to provide a maintained temperature reference surface for T- outside the chip MAX31856 ? or I just need to connect thermocouple's terminal T+ and T- to the chip and trust the reading that comes out even if the cold junction reference temperature changes ?

The datasheet is rather clear with this. You need to read the datasheet and follow the schematics and design recommendations.

I don't understand the term cold junction.

Then you did not read a documentation about thermocouple temperature measurement.
"Cold junction" is important and thus most basic for thermocouple measurement.
There are many documents, tutorials, application notes.
Reliable and good ones usually are from: semiconductor manufacturers, thermocouple manufactureres, universities...

most straight forward:
Go to the internet page of your IC:

There is a section "Technical Docs". (Wasn´t that simple?)

You are not the first one who needs to understand thermocouple measurement. (Some years ago I was in the same situation) Thus the manufacturer make huge efforts to provide usful informations. All is free for you. You just have to use it.

You may find similar informations from: AnalogDevices, TI, ST and many others.....

Klaus

 

[danadakk]

https://www.ti.com/lit/an/sbaa274/sbaa274.pdf?ts=1619176388567&ref_url=https%253A%252F%252Fwww.google.com%252F

https://s3.amazonaws.com/files.technologyreview.com/p/pub/legacy/temperature_measurements_with_thermocouples.pdf

https://www.analog.com/media/en/technical-documentation/application-notes/an28f.pdf

Above might help.


Regards, Dana.

 

[engr_joni_ee]

Hi, now I have looked at the TI document, which says that the measurement of the cold junction is done through the thermistor, RTD or a diode.

Coming back to MAX31855, according to datasheet there is a 12 bit D[15:4] internal reference junction and comes out together with the 14 bit D[31:18] thermocouple data. I am wondering how to include 12 bit internal reference measurement towards the calculation of actual temperature ?

 

[c_mitra]

Hi,
I am looking for electronics circuit for reading K type thermocouples. I have looked at some references but not able to understand where we need to apply voltage to power them up. It just says that one terminal has to be on the common reference temperature point which at some places is referred as cold junction and then we just measure the voltage across the thermocouple but the question is where the biasing voltage come from ?

Thermocouples produce a voltage when two dissimilar metals are connected with the two junctions at two different temps.

The effect is called thermoelectric effect. It includes Seebeck effect, Peltier effect and Thomson effect.

Because you are interested in accuracy, remember that the voltage produced is not linearly related with the temp.

Also remember that you have two junctions and you need to keep one junction at a const temp (preferably 0C).

Cold junction temp compensations are always approximate. It is of course easy to get an accuracy of +/-1C with a type K thermocouple in the -100 to +100 range.

The bias voltage produced comes from the electronic energy levels mismatch when two different metals are connected (like a p-n junction). You do not need any external voltage source (but you will need power to run associated electronics).

You need to measure the voltage quite accurately because the potential developed is small.

 

[FvM]

I am wondering how to include 12 bit internal reference measurement towards the calculation of actual temperature ?

According to the datasheet, the reference measurement is automatically added to the thermocouple output. You don't need to do anything about it.

The device senses and corrects for the changes in the reference junction temperature with cold-junction compensation. It does this by first measuring its internal die temperature, which should be held at the same temperature as the reference junction. It then measures the voltage from the thermocouple’s output at the reference junction and converts this to the noncompensated thermocouple temperature value. This value is then added to the device’s die temperature to calculate the thermocouple’s “hot junction” temperature. Note that the “hot junction” temperature can be lower than the cold junction (or reference junction) temperature.

 

[KlausST]

Hi,

I wonder why making things complicated, why referring to third party documents, why bothering with cold junction temperature measurement.

If the goal is to measure the temperature at the thermocouple, then
* configure the MAX31856 (which thermocouple type you use, how you like the cold junction compensation to work) one time during power up
* read the value given in registers 0x0C, 0x0D, 0x0E
... and you are done.

All the measurement, compensation, calculation, linearisation is done by teh MAX31856. No need to bother about it all.
No need to keep any temperature at 0°C.

Please read the MAX31856 datasheet and the related documents.

****
and similarily for the MAX31855. The datasheet clearly says:
The device senses and corrects for the changes in the
reference junction temperature with cold-junction com-
pensation. It does this by first measuring its internal die
temperature, which should be held at the same tem-
perature as the reference junction. It then measures the
voltage from the thermocouple’s output at the reference
junction and converts this to the noncompensated ther-
mocouple temperature value. This value is then added
to the device’s die temperature to calculate the thermo-
couple’s “hot junction” temperature.

and:
A complete serial-interface read of the cold-junction compen-
sated thermocouple temperature requires 14 clock cycles.

--> no need for reading cold junction temperature, no need for compensation, it´s all done inside.

Please read the MAX31855 datasheet and the related documents.

Klaus

Added. I was delayed and didn´t see FvM´s post