Go Down

Topic: High resolution, high accuracy temperature sensor (Read 5495 times) previous topic - next topic

muppy

Hi All,
i need to build a high resolution and high accuracy temperature sensor based on Arduino
I need accuracy and resolution of 0.05 deg or better.

In my understanding, the key point to achieve a so challenging result is a proper choice of the temperature sensor, is this right ?

Any suggestion about an appropriate temperature sensor for this purpose ?

Thanks
Andrea

muppy

Sorry, there is an error in my previous message.
I need 0.05 deg resolution and 0.5 accuracy.

Andrea

PeterH

Over what temperature range, in what environment?

muppy

environment = air
temperature range from 18 to 25 deg

Thanks,
Andrea

dc42

A DS18B20 can give you accuracy of 0.5C and resolution of 0.0625C. Is that good enough?
Formal verification of safety-critical software, software development, and electronic design and prototyping. See http://www.eschertech.com. Please do not ask for unpaid help via PM, use the forum.

muppy

This is very close to the limits of what i can accept.
Do you know if there is a sensor which will provide better performances ?

Thanks,
Andrea

robtillaart


if you do multiple readings you can increase the decimals. (averaging ~16 readings gives you one extra decimal resolution)

Another option is to check if there is a physics process that responds to temperature that you can measure indirectly.
I am thinking of expansion of some metal, and measure the expansion precisely.
Or you could measure a gas in a closed room and measure the pressure differences..
Don't know if that would be fast enough...

How many reads/minute do you need?
Rob Tillaart

Nederlandse sectie - http://arduino.cc/forum/index.php/board,77.0.html -
(Please do not PM for private consultancy)

muppy

I need more or less 10 reading per minute.
Ok, i can accept to average over 16 measurements.
Thanks!

Nick_Pyner

#8
Jun 02, 2013, 03:47 pm Last Edit: Jun 02, 2013, 05:10 pm by Nick_Pyner Reason: 1

I need more or less 10 reading per minute.
Ok, i can accept to average over 16 measurements.



You need to check the return time of the DS18B20. I believe it is normally used for one reading per second which implies about four averaged readings per minute. If you really need to take that path, a PT 100 is probably a better bet.

PaulS

Quote
I need 0.05 deg resolution and 0.5 accuracy.

This makes no sense. A sensor that reports a temperature like 78.95 degrees when it is really 78.5 is useless. The accuracy has to be close to the resolution in order for the data to be useful.
The art of getting good answers lies in asking good questions.

sonnyyu

#10
Jun 02, 2013, 05:06 pm Last Edit: Jun 02, 2013, 05:08 pm by sonnyyu Reason: 1

Sorry, there is an error in my previous message.
I need 0.05 deg resolution and 0.5 accuracy.

Andrea



environment = air
temperature range from 18 to 25 deg

Thanks,
Andrea


How did you make air to stay still between each measure?
Person's body temperature is 36.75 C or 36.8 C or less 37 C make any difference?

afremont


Quote
I need 0.05 deg resolution and 0.5 accuracy.

This makes no sense. A sensor that reports a temperature like 78.95 degrees when it is really 78.5 is useless. The accuracy has to be close to the resolution in order for the data to be useful.


Not if you care about relative measurements.  Even though the overall accuracy is only .5 deg, the relative measurements between steps will be much more accurate and may be entirely useful depending upon the application.
Experience, it's what you get when you were expecting something else.

pito

#12
Jun 02, 2013, 05:20 pm Last Edit: Jun 02, 2013, 05:23 pm by pito Reason: 1
Your temperature reading is, for example:

Code: [Select]
T = Value_under_specific_resolution +/- (1% accuracy + 1digit)
T = 20.25 +/- (0.2025 + 1 * 0.05)
T = 20.25 +/- 0.2525 degC

resolution = 0.05
accuracy = 1%

Go Up