Needed microcontroller for temperature sensor data logging in high temperatures

Thread Starter

joris484

Joined Mar 28, 2019
7
Hello everybody, I have no experience with microcontrollers. My project is temperature sensor (thermo couple) which generates some mili volts, amplifying them with Op amp into 3-4 Volts and after that I need microcontroller with ADC (10 bits will be enough). Microcontroller should store the data from measuring into external EEPROM or flash memory for example. The working temperature should be something like 150 C and the size of the microcontroller should be as small as possible. Could you suggest me microcontroller for that purpose. Thanks!
 

ericgibbs

Joined Jan 29, 2010
9,136
hi joris,
Welcome to AAC.
Which programming language are you planning to use.?
The Arduino range of MCU modules would be suitable.

E

BTW: is the a college assignment.?
 

Thread Starter

joris484

Joined Mar 28, 2019
7
Hi Eric,

It is a Master Thesis topic and I work on it in a company. The solution should be placed on a rotating plate. That's why I am searching for a microcontroller with thickness of 2-3 mm (is that possible?). I am not sure about the language, I need to start from somewhere and to work on it in the next 2 months.

Regards
 

Yaakov

Joined Jan 27, 2019
1,611
Hi Eric,

It is a Master Thesis topic and I work on it in a company. The solution should be placed on a rotating plate. That's why I am searching for a microcontroller with thickness of 2-3 mm (is that possible?). I am not sure about the language, I need to start from somewhere and to work on it in the next 2 months.

Regards
How will you power it?
 

ericgibbs

Joined Jan 29, 2010
9,136
hi joris,
Do you have a draft sketch of the project that you could post.? showing the critical dimension areas.
It sounds an interesting challenge.
E
 

Thread Starter

joris484

Joined Mar 28, 2019
7
Hi Yaakov,

Have no idea till now. I found ATmega32 by Atmel - it has 10 bits ADC and is 1.2 mm thick. Would it be a solution for my project? Only the temperature range is up to 85C... I would prefer smth up to 150C
 

ericgibbs

Joined Jan 29, 2010
9,136
hi joris,
Depending upon the rotating surface material and the way is driven, one powering option could be conductive slip rings on the under face of the surface.
We need more 'physical type' information regarding the mechanics of the project, that will determine electronics part of the project.
E
 

Yaakov

Joined Jan 27, 2019
1,611
Hi Yaakov,

Have no idea till now. I found ATmega32 by Atmel - it has 10 bits ADC and is 1.2 mm thick. Would it be a solution for my project? Only the temperature range is up to 85C... I would prefer smth up to 150C
I have the impression you've turned designing this on its head. You've specified certain things that may or may not be actual constraints, and you've already got an imaginary solution you are chasing. I suggest you step back and:

List the things that your device must do from a user perspective, not a solution one. (e.g.: "record temperatures for use in spreadsheets")
List the things that are given, and can actually not be changed. (e.g.: "must use built-in thrermocouple")
List the maximums and minimums you will encounter (e.g.: "measurement from 100°C to 140°C", "10SA/s")
List the environmental conditions (e.g.: "ambient operating temperature maximum 150°C")
List the dimensional and mechanical requirements (e.g.: "maximum height 3mm", "maximum size 10x50mm", "rotation 360° no stops")

Then, with those things in place, you can start to iteratively design a solution that meets the user requirements within the constraints. If you don't do it this way, your "solution" will become the problem instead of the original problem you were working on.

The outcome of this exercise may surprise you, or it may turn out you've already chosen the right thing, but if you don't do it, you are no longer solving the original problem rather one you've created with your attempt to do it.

This needs to be applied iteratively. That is, each time you make a choice for part of the solution, run through the list and add the new constraints you've created by that choice, check it against the intended outcome, and make sure you aren't blocking yourself from succeeding.

If you develop such a list, you can get help more easily too. This specification will answer the many questions you will be asked over and over.

I assume you keep a notebook, if not, this is why you need to. You need to be able to follow back the reasons you are doing things, and explain them to people you are collaborating with. It also helps to know why you rejected something, which you very well may not remember.
 

Thread Starter

joris484

Joined Mar 28, 2019
7
hi joris,
Do you have a draft sketch of the project that you could post.? showing the critical dimension areas.
It sounds an interesting challenge.
E
The most critical dimension is the thickness - up to 2 - 2.5 mm and the temperature - should withstand 150C. I have the full freedom for solutions. I chose thermocouple K type that generates 16 mV at 400C, after that Op amp amplifies the voltage to 3.5 - 4 V (I haven't calculated it yet), 10 bits ADC will give around 0,5C between the points, and the measurements should be stored in memory. As i said, i still have no idea how to power the curcuit
 

jpanhalt

Joined Jan 18, 2008
8,359
Are you required to use thet set-up you describe?

May I suggest the Maxim MAX31856 that includes noise circuitry, amplifier, cold junction compensation (on board or external), linearization, and digitization with 19-bit resolution (if needed).

That connects directly to the thermocouple(s) and to the MCU via SPI. In fact, there is an arduino breakout board available cheaply. As for memory, I am not familiar with the Atmel, I am using a simple PIC16F chip and linear memory (RAM) for data. Your temperature range at 10-bits should be no problem.

I don't think putting you IC's directly in the solution is a good idea.

EDIT: Here's a link to that device: https://datasheets.maximintegrated.com/en/ds/MAX31856.pdf
 
Last edited:

ericgibbs

Joined Jan 29, 2010
9,136
hi.
Ok, lets assume we can mount a MCU that gives the required clearance.
This suggests that the OPA between the TC and the MCU is also going to mounted alongside the MCU.
How do you plan to display the MCU ADC values.?
You could supply power by using conductive slip rings on the underside of the rotating plate.

E
 

Thread Starter

joris484

Joined Mar 28, 2019
7
hi joris,
Depending upon the rotating surface material and the way is driven, one powering option could be conductive slip rings on the under face of the surface.
We need more 'physical type' information regarding the mechanics of the project, that will determine electronics part of the project.
E
Everything should be placed on the steel plate - it is part of wet clutch. The power should be enough for 3 hours measuring the temperature. The plate is rotating in oil. After that the clutch is dissasembled and the data from the memory is transferred to PC for analyses.
 

jpanhalt

Joined Jan 18, 2008
8,359
Everything should be placed on the steel plate - it is part of wet clutch. The power should be enough for 3 hours measuring the temperature. The plate is rotating in oil. After that the clutch is dissasembled and the data from the memory is transferred to PC for analyses.
If everything is going to be immersed in an isothermic oil bath, how will you get a "cold junction?" Perhaps you need to change the design to a platinum resistor (for accuracy) or another type of resistance thermometer.
 

Thread Starter

joris484

Joined Mar 28, 2019
7
If everything is going to be immersed in an isothermic oil bath, how will you get a "cold junction?" Perhaps you need to change the design to a platinum resistor (for accuracy) or another type of resistance thermometer.
That's a very good point!
My other option is RTD Platinum sensor like this:
https://shop.bb-sensors.com/en/Temperature-measurement/Temperature-sensors/Platinum-temperature-sensor-Pt200.html
In this case I need to pass current through it, after that I need voltage devider and again microcontroller with ADC. Am I right?
 

Yaakov

Joined Jan 27, 2019
1,611
That's a very good point!
My other option is RTD Platinum sensor like this:
In this case I need to pass current through it, after that I need voltage devider and again microcontroller with ADC. Am I right?
Can you see that by choosing a part
Of the implementation before design you set yourself up to fail? You really need to step back and make some drawings and write a spec. Good luck.
 

jpanhalt

Joined Jan 18, 2008
8,359
That's a very good point!
My other option is RTD Platinum sensor like this:
https://shop.bb-sensors.com/en/Temperature-measurement/Temperature-sensors/Platinum-temperature-sensor-Pt200.html
In this case I need to pass current through it, after that I need voltage devider and again microcontroller with ADC. Am I right?
The times I've used a platinum thermometer in an experiment (more than 50 years ago), we used an Wheatstone bridge. I suspect there have been some improvements.

What level of precision will you need? 10-bit precision will give you about 0.2°C precision (maybe a little less). How frequently and for low long will temperatures be recorded -- just to estimate memory requirements. You may want to use non-volatile memory so data aren't lost if there is a power interruption.
 

Thread Starter

joris484

Joined Mar 28, 2019
7
The times I've used a platinum thermometer in an experiment (more than 50 years ago), we used an Wheatstone bridge. I suspect there have been some improvements.

What level of precision will you need? 10-bit precision will give you about 0.2°C precision (maybe a little less). How frequently and for low long will temperatures be recorded -- just to estimate memory requirements. You may want to use non-volatile memory so data aren't lost if there is a power interruption.
The expected temperature at the sensor is up to 300°C for a small amount of time. 10 bit will give approx. 0,3°C of precision. If I use 8 bit, this will give over 1°C precision (which is still OK for me). The temperatures will be registered for 30 min (in an interval of 0,5 sec) which will give 3600 records.
 
Top