Acquisition of temperature values

Hi,

I would like to start a project related to the acquisition of temperature values. The specifications are as follows:

  • acquisition of at least 36 probes (I could use also more arduinos)
  • the probes are thermocouples type T premium (0,01°C)
  • range of temperatures: from -200°C to 400°C

My questions are:

  • which board do you suggest? arduino Mega?
  • which module/ amplifier do you suggest for the acquisition of the temperatures?
  • have you any examples or references?

Thanks in advance

Here is list of temperature sensor types, one example of which, a Nickel Chromium / Constantan type thermocouple which has a range from -200 to + 900 degrees Centigrade. http://www.electronics-tutorials.ws/io/io_3.html

It looks like you'll have to design an amplifier to condition the voltage for an Arduino to get it more into the range 0 to 3.3/5 volts

How widely distributed (geographical area) are the sensors going to be or are they local to a very small area ?

Can you give more details of what you are actually measuring or is this a more theoretical school assignment ?

It will be very expensive to achieve 0.01 degree C accuracy. This is not something a beginner can hope to construct.

http://www.gecinstruments.com/?gclid=CjwKEAiAvs7CBRC24rao6bGCoiASJABaCt5DpcpRmLXE6xTtSrFtnVaPaa1Mj6_MZgLTn_GYrwYUDRoCKt_w_wcB

I'm not sure that 0.01C is acheivable over that range, whatever the cost.... 1C might be.

and what would it mean, anyway ?

regards

Allan.

Hi, How long will the leads be to the thermocouples?

Each thermocouple will need a T-type version of this breakout board. https://www.sparkfun.com/products/13266

This will enable multiple units to be connected to an arduino controller.

If you are going to need to add to the length of the T-type thermocouple leads, you will need to use compatible T-type wire.

Tom.... :)

Or a single signal conditioner with a big analog mux...

regards

Allan

6v6gt: How widely distributed (geographical area) are the sensors going to be or are they local to a very small area ?

Can you give more details of what you are actually measuring or is this a more theoretical school assignment ?

Thanks for your answer, the sensors are very close. For my work I often have to map the temperature of equipment such as autoclaves, dryers, freeze dryers, etc. Because of I am a supporter of Arduino, I would try to see if I can use it for this job. For this reason the probes are very close. I use type "T premium" thermocouples (resolution 0.01 ° C), in order to make the calibration of the thermocouples with a maximum uncertainty of 0.20 ° C.

thanks in advance

jremington: It will be very expensive to achieve 0.01 degree C accuracy. This is not something a beginner can hope to construct.

http://www.gecinstruments.com/?gclid=CjwKEAiAvs7CBRC24rao6bGCoiASJABaCt5DpcpRmLXE6xTtSrFtnVaPaa1Mj6_MZgLTn_GYrwYUDRoCKt_w_wcB

Thanks for your answer, for my work I use thermocouple type T, in order to make the calibration of the thermocouples with a maximum uncertainty of 0.20 ° C and I would try to acquire the data with Arduino.

regards

TomGeorge: Hi, How long will the leads be to the thermocouples?

Each thermocouple will need a T-type version of this breakout board. https://www.sparkfun.com/products/13266

This will enable multiple units to be connected to an arduino controller.

If you are going to need to add to the length of the T-type thermocouple leads, you will need to use compatible T-type wire.

Tom.... :)

thank you Tom, I buy thermocouples ready, I never make extension, so as to avoid mistakes. This board looks interesting, but I see that the resolution is 0.25 ° C, which is too big form me (it uses 2 bits for the decimal part), I would need at least 4 bits (0.07 °C). Do you know other boards?

thanks

Hi, For the resolution you are looking for you may need to go to industrial/commercial thermocouple interfaces.

For what you want to do, I see full industrial loggers ready to go.

But you may find something if you dig deep and long enough.

Tom.. :)

TomGeorge: Hi, For the resolution you are looking for you may need to go to industrial/commercial thermocouple interfaces.

For what you want to do, I see full industrial loggers ready to go.

But you may find something if you dig deep and long enough.

Tom.. :)

Thanks!

Start by looking at sensitivity, that is how many bits of resolution the Arduino ADC has, and the span you want to read. -100 to +400 = 500 degree range the arduino has a 10bit ADC so reads 1024 counts. in your case, the count is already 0.5 degrees.

Even if you typed in a ' - ' to confuse us and you want +200 to +400 you have a 200 degree range, so get 0.2 degree resolution. still not good enough.

when attempting to measure a thing, the rule of thumb is that you have to have between 4 and 10 times the resolution. this does not even have 1 times the resolution you seek.

===================

Thermocouple interfaces.

The thermocouple is a device that measures the difference in temperature AT THE JUNCTION of dissimilar metal. you THINK that is at the sensor point, but since you tie your Constantine side to your chrome plated brass screw, you created a second dissimilar metal connection. ditto with the copper wire.

====================

the voltage created at the junction requires an amplifier. your options here are multiple. get a thermocouple chip that does all the work for you in turning your voltage into a digital signal. - or - design and build an amplifier circuit, then feed that into an ADC of sufficient bits so as to have the ability to see a temperature of 0.0025 , not sure how close a 16bit would get you, but the ADS1115 has libraries available. - or - buy a thermocouple interface that has an output you can use. preferably a digial one, but if you cannot find one, then you need to get an ADC that will have the resolution for the analog output.

=======================

get a high quality power supply. a noisy power supply will cause headaches.

=======================

The TEENSY has a 16 bit ADC, but not sure of the chip, I have seen notes of it having a real world 13 bit resolution.

=======================

Although on paper, this will be a monumental task, IMHO you are not looking for accuracy, you are looking stricly and only for sensitivity and response.

IMHO, every one could be 5 deg off from every other one ( 7 red lines, all perpendicular, one being blue and one being transparent ?)

but. if they react to temperature variations of 0.01 or less, then you can just create off-set values in software so you can see the difference between them.

my suggestion is to get a few that you think will work and test them.

as for your question about processors. what you are asking is not hard once you have the data. an UNO or NANO should be able to do the math. you could use a MEGA for the extra inputs and larger memory.

with 36 probes, using a NANO offer 8 ADC inputs. you could use 5 to connect too the probes and then have each relay the data to one MEGA for the math.

if you go with the ADS1115, it outputs digital and can 'see' 4 inputs. so you would need 9 of them. you can put up to 4 on one I2C line with most of the pre-mounted boards you can buy.

in any case, you have some fun ahead of you.

I use type "T premium" thermocouples (resolution 0.01 ° C), in order to make the calibration of the thermocouples with a maximum uncertainty of 0.20 ° C.

How do you "calibrate" the thermocouples to account for the nonlinearity of the thermocouple response, especially over that large temperature range? Note that Arduino does not support double precision floats, as normally required for polynomial approximations.

If you interpolate the thermocouple table, does the Arduino have enough memory to accommodate it? This is, of course, assuming that you are using an external ADC with higher resolution than the Arduino ADC as mentioned above.

Please elaborate!

I'm imagining 36 thermocouples all in a tight bundle with tape around them, trying to get the same reading calibrated on all of them. A tiny draft on one side of the room will show up as a significant error.

I imagine all of the thermocouples in a sealed box inside of a temperature environment that has minimum change of time. And enven then the having some variations.

this is not a project for the faint of heart !

and would be a great white paper on calibrating multiple devices.

Hi,

How do you "calibrate" the thermocouples to account for the nonlinearity of the thermocouple response, especially over that large temperature range?

The best way is to use a HOT POT and Laboratory Temperature calibrator, you heat the HOT POT and use the calibrator sensor next to the tested TC, comparing readings at different temperatures.

The interface circuitry for the T-type will have all the compensation for the T-type, hence different interface IC's for K-type, N-type etc.

Because thermocouple operation is at atomic level, their output temp=volts is well documented, it is calibration of the interface circuit that becomes the important issue.

Tom... :) PS. I wondered about the choice of T rather than K, you aren't measuring a cryogenic project