Thermocouple calibration

I have three different instruments, thermocouple shield, 2 x DMM all using type k.
Over 0 to 200 c I get readings that vary by +- 5 degrees.
Does not appear to be systematic.

I assume this is due to differences in calibration.

I had hoped to use the DMM to better calibrate the arduino, but do not know which reference to use.
Using a ir thermometer gave errors up to 15 degrees compared to thermocouple.

Which reference is likely to be the most accurate ?

Boil some water and put in the thermocouple and see what readings the various meters give you, should be 212F or 100C. You might have to adjust a little for elevation is you are not at sea level but that should be a simple calculation if you google for it. And a 'ice bath' can give you a decent 32F or 0C reference to use.

Lefty

At 0 and 100 they all pretty much agree,except the ir thermometer.
Over the 200 degree range they vary.
Unfortunately the critical temp is 140 deg and at that point they disagree badly.

What are you using for your 140 degree calibration source? What sort of circuit is your shield?

Did you connect the wires of your thermocouple directly or did you extend them ?
And if so, how did you do that ?

Shield is based on max 31855 uses standard ,un extended themocouple with plug into proper socket on the board.

High temp uses custom amp , ADC type circuit with a lookup table, it's this I need to calibrate.

140 reference , do not have one, only things that boil at that temp appear to be toxic and unobtanium.

Edit ,please ignore th high temp reference it's not relevant yet, just need to calibrate the 31855 for now.

If that's the case, I'd use it as my calibration standard.

Good luck.

Then how do you know it's inaccurate at this temperature?

Boardburner2:
High temp uses custom amp , ADC type circuit with a lookup table, it's this I need to calibrate.

So are you okay with calibrating this now? You haven't posted a schematic diagram of this circuit.

You might try loading your water with some common table salt. There are many references about the water solute boiling points with common ingredients available as always on Wiki.. A really great resource.

Bob

Docedison:
You might try loading your water with some common table salt. There are many references about the water solute boiling points with common ingredients available as always on Wiki.. A really great resource.

Bob

Unfortunately, that will only raise the boiling point about 3 degrees Celsius before the solution is saturated and will dissolve no more salt.

Quote from: Boardburner2
140 reference , do not have one, only things that boil at that temp appear to be toxic and unobtanium.

Then how do you know it's inaccurate at this temperature?

He doesn't know, but rather he knows that his various meters, arduino, IR don't agree at that around that temperature value.

A typical quandary, man with one watch always knows the time, man with two (or more) watches never quite sure. That is why primary calibration standards are so important and usually expensive if guaranteed accuracy is required or desired.

Lefty

retrolefty:

Quote from: Boardburner2
140 reference , do not have one, only things that boil at that temp appear to be toxic and unobtanium.

Then how do you know it's inaccurate at this temperature?

He doesn't know, but rather he knows that his various meters, arduino, IR don't agree at that around that temperature value.

A typical quandary, man with one watch always knows the time, man with two (or more) watches never quite sure. That is way primary calibration standards are so important and usually expensive if guaranteed accuracy is required or desired.

Lefty

Yep, it's a dilemma.

If those temperatures are in celcius, that sounds awfully close (and past) manufacturers design maximums.

Electronics derate above design temperatures.

The data sheet says.

Note that the MAX31855 assumes a linear relationship between temperature and voltage. Because all thermocouples exhibit some level of nonlinearity, apply appropriate correction to the device’s output data.

No kidding, for the cost of the chip why did they not do this ?
There are published tables for all the thermocouple types.
I can't use them with this chip as its already digitised the output.

I can probably knock up the same linear circuit cheaper with amps but with much more hassle.

I've been having look at other bits, borrowed aquarium thermometers , Weather monitors,etc.

Amazing, 0 and 100 they all agree but I suppose that is not too surprising.
Elsewhere though , wow some big disagreements.

This paper discusses correction for the non-linear response of a K type thermocouple. You could just try the correction factor they came up with.

https://docs.google.com/viewer?a=v&q=cache:diwsbQCjYi4J:gr.xjtu.edu.cn/LiferayFCKeditor/UserFiles/File/Nonlinearity%2520correction%2520of%2520the%2520thermocouple%2520based%2520on%2520neural%2520network.pdf+non+linearity+correction+k+thermocouple&hl=en&gl=us&pid=bl&srcid=ADGEESihE0fNTAXSA_GvwYsLC-H08jXXpFSDbRYqeBggrzdOO9sHsNIXpUKocRuipSv2ilM8vcgavJSmtaGOVcClXfptKK3LNYsSbzNiB4J3EuJbn6tiGBTEhj0-WMQ1oMGXwPFxN37F&sig=AHIEtbQdxuCh0TXlb0saORpodwMbUpUt1Q

Boardburner2:
The data sheet says.

Note that the MAX31855 assumes a linear relationship between temperature and voltage. Because all thermocouples exhibit some level of nonlinearity, apply appropriate correction to the device’s output data.

No kidding, for the cost of the chip why did they not do this ?
There are published tables for all the thermocouple types.
I can't use them with this chip as its already digitised the output.

I can probably knock up the same linear circuit cheaper with amps but with much more hassle.

I've been having look at other bits, borrowed aquarium thermometers , Weather monitors,etc.

Amazing, 0 and 100 they all agree but I suppose that is not too surprising.
Elsewhere though , wow some big disagreements.

Now that I read the data sheet, I see it's not very accurate. +- 2degrees C isn't that great.

I don't know what your budget is but check the price of lab grade thermometers. You can use hot cooking oil to make a bath to check the calibration of your sensor and create a lookup table to linearize it.

It's complicated !
I have wombled an old toaster oven to hack into a black body radiator.
These devices have no known calibration curves and have to be calibrated a degree at a time with tables.
I was hoping to use contact measurement of the radiator to calibrate.
If I can't get accurate reading for a 200 deg oven , can I trust it at higher temperatures where I have no other temperature references to check with.

After 30 years as a control electrician i would tend to believe the thermo-couple and not the IR
unless the unit has emmisivity settings and retails over 500 dollars

how to calibrate MAX31855 ?

internal temp is ok but measured values are ...??
and why so different values?

Internal Temp = 23.62
C = -15.75
Internal Temp = 23.62
C = -16.75
Internal Temp = 23.69
C = -17.00
Internal Temp = 23.69
C = -15.75
Internal Temp = 23.62
C = -16.75
Internal Temp = 23.62
C = -16.00
Internal Temp = 23.62
C = -15.75
Internal Temp = 23.62
C = -15.75
Internal Temp = 23.62
C = -16.00

if i heat thermocouple with a lighter temp goes up...like normal... but when it cools down , values are.. dont know what ..

Internal Temp = 24.44
C = 48.25
Internal Temp = 24.50
C = 487.00
Internal Temp = 23.50
C = 721.25
Internal Temp = 24.25
C = 645.25
Internal Temp = 24.06
C = 758.25
Internal Temp = 24.37
C = 816.00
Internal Temp = 24.37
C = 281.50
Internal Temp = 23.37
C = 141.00
Internal Temp = 24.19
C = 87.75
Internal Temp = 24.50
C = 61.75
Internal Temp = 24.31
C = 47.25