yngndrw

Hey, been a while since I was last on here.

Anyway, I'd like to make a Thermometer which has K Type thermocouple inputs.

I've bolded the main questions which I would like answering.

I've recently been looking at the process of reading temperatures with K Type thermocouples and I found it to be the following:
1) Calculate the temperature of the "Ice Point Compensation" thermistor.
2) Use a K Type lookup table to calculate the equivalent voltage for this Ice Point temperature.
3) Get the voltage of the K Type and subtract (Or add ?) the compensation voltage.
4) Use the K Type lookup table to convert this new voltage to the temperature of the probe.
Firstly, is this process correct ?

Secondly I don't quite understand the tables.
Here's a table that I found: http://www.instrumentation-central.c...ypeKTableC.pdf

From what I understand the bold column down the left in the temperature in degrees C and the values in the middle of the table are the corresponding voltages of the thermocouple in mv, assuming an Ice Point of 0C. However I don't understand what the "0 -1 -2 -3 -4 -5 -6 -7 -8 -9 -10" headings are across the tol of the table, what are these and how do they affect me ?

Thirdly are there any formulas for conversion of K Type themperatures and voltages which can easily be done on a PIC18f (In C) ? There must be formulars for conversions both ways, I.e. Temperature to Voltage and voltage to Temperature.

Finally I've been looking at an opamp to use to build the differential amplifier for the thermocouple it's self. I've just got a sample of TC913A from Microchip. (Datasheet link) Would these be suitable ? What other alternatives are there which would be better, or would these ones be more than suitable ?

Thank you very much for any help,
-Andrew.

dknguyen

MY SUGGESTION:
THe best alternative? An IC that does ALL the work for you!

Just use an AD595 or AD597 from Analog Devices that does the amplification, cold-junction compensation and linearization for you and outputs an analog voltage of about 10mV/C.

Or use a MAX6675 from Maxim that does all that, except uses an 12-bit sampling and an SPI interface.

ANSWERS TO YOUR QUESTIONS:
You CANNOT convert the thermistor's temperature back into an equivelant Type-K voltage and then add it to the thermouple voltage and then use the combined voltage to calculate temperature as you have done in your steps. Why? THe thermocouple temp-volt curve is NONLINEAR. Luckily, there is a more intuitive straightforward way of doing it with less work (ie. less using the look-up tables back and forth):

1) Calculate the temperature of the "Ice Point Compensation" thermistor either with a approximating equation or lookup table
2) Use a K Type lookup table to calculate the RELATIVE TEMPERATURE MEASURED BETWEEN THE TC ENDS
3) ADD the TC temperature to the Ice Point Compensation Temperature (think about it, you are taking a relative temperature measurement and referencing it to an absolute temperature measurement)
4)You can, if you choose, to convert everything from C to K before adding it. It's not needed since +/- signs are given for both thermocouples temperature difference measurements as well as absolute temperature measurements so it will work itself out, but can make it easier to spot out mistakes.

You have to understand a thermocouple's output voltage represents the temperature difference between it's ends (it only measures relative temperature). So you stick an absolute temperature measurement at one end and calculate the absoltue and relative temperature INDEPENDENTLY (you use the tables/values/equations for the thermistor, RTD, or silicon sensor to get an absolute temperature and you use the Type-K table to get the TC temperature difference). THen you add it up, since one is a relative measurement and one is absolute. What's the point of this? The far end of the thermocouple can handle far higher temperatures than the absolute temperature sensors.

THe conversion formulas are all approximations so they will probably be complicated multi-order polynomials with a lot of coefficients and higher powers. Not so easy to rearrange reverse those formulas to reverse their input/output so you might not find any that let you change temperature back to voltage (obviously a look-up table lets you do this). Few people would go to the working out the equation that approximates the temperature-to-voltage conversion relationship because TCs aren't used in this way, and the only reason you need it for your original process was because it's incorrect. So you don't need it after all.

You'd have to look up ITS-90 to find out what the 1-10 are.

Hope this helps.

yngndrw

Thank you very much your your help.

Regarding your suggestion, I am looking to measure the full negative temperature range of a K Type - Therefore only the AD597 from the ICs you said would be suitable. But then there's the cost of the ICs (Last time I checked, thermocouple ICs were quite expensive.) as well as the avalibility. (I can't find any local suppliers who stock any of them.)

Also I wanted this as a project more than anything, so as awesome as lerring an IC do all the work is, it kind of takes away the whole project.
Thanks for the suggestion anyway, in any other situation using an IC like them would probably be the best option.

Regarding step 2:
I'm just unsure of one thing, you say "RELATIVE VOLTAGE MEASURED BETWEEN THE TC ENDS" but I don't understand why you'd calculate the voltage ? Is this ment to mean calculate the relative temperature or am I missing something here ?

Fortunatly with the process that you suggeted, reverse lookup isn't required so I'll try and find a formula which can calculate it.

I'll look at ITS-90 now.

Thanks for all your help, It's clarified a lot.

dknguyen

Doesn't the AD595 measure negative voltages even better than the AD597 does (which is calibrated to work best for warmer temperatures).

If the temperature in my hand is 25C and hold it in my hand and put the other end in a 1500C oven, the thermocouple voltage will represent 1500-25C = 1475C. It measures the temp difference between it's ends.

To actually know the temperature in the oven, I need to know the temperature of my hand. I can either stick my hand in ice water (which is complicated and bulky but very accurate), or use an absolute temperature sensor to measure the temperature of my hand, then add this absolute measurement to the relative TC measurement to find the absolute temperature at the far end of the TC.

Sorry, typo. It has been fixed.:
RELATIVE TEMPERATURE MEASURED BETWEEN THE TC ENDS

yngndrw

Oops I missread the AD595's datasheet, yes it does go to -200C.

Regarding the ITS-90, I still don't understand the -1 to -10. Is this something todo with the pressure or am I on the wrong tracks ?

According to the datasheet for the AD595, the 0 column is the one that I should use.

Can you please explain the tables ?

Thanks.

dknguyen

I do not know. Sorry. But I did find this and this is my take on it (which is 30 seconds worth of looking at it so take it worth a grain of salt).

They seem to be coefficients for the equations and different coefficients apply under different conditions (also includes inverse coefficients!):
http://srdata.nist.gov/its90/menu/menu.html

Rather than actual measurements, those lookup tables seem to be using the equations and the different 1-10 columns might be showing the order of approximation (number of coefficients being used) for each entry in the table. More coefficients = higher order approximation = more accurate.

negative 1-10 seems to be coefficients for negative temperatures and positive 1-10 seems to be coefficients for positive temperatures.

yngndrw

Ah okay then. I'll have another look around later on today.

Thank you very much for all of your help, it's been very helpful.

JimB

If I understand your problem with the numbers 0 to 10, it is as follows:

To find the TC output for 54 degC,
look down the degC column at the left hand side until you get to 50,
then look across until you get to the column headed 4 degC
the reading at the junction of the 50 deg Row and the 4 deg Column is the millivolts for 54 degC.

See the attached piccy.

JimB


PS: Note this message has been deleted and re-posted due to a screw-up with the attachment. Should be OK now.

PPS: At the third attempt, I may have got it right.

Attached Images

TC Table.PNG (49.2 KB, 11 views)

__________________
Experience is directly proportional to the value of the equipment ruined.

yngndrw

Ahh, that makes sense. I feel kind of silly for not realising that now.

Thank you for your help.

dknguyen

How come on the chart provided the voltage for something like 100C doesn't match the voltage for 90C+10C?

EDIT: Oh wait, it does. I just keep looking at the last entry of the upper row and the first entry of the lower row. It only works for positive temperatures and since negative temperatures were on the first page (I was lazy and never bothered to look carefully past the first page), I kept looking at the wrong entries to match voltages between.