Go Down

Topic: high accuracy temp sensor!? (Read 5 times) previous topic - next topic

giorgiodidon

hi all, is there an high accuracy temperature sensor? i mean something with 0.1°C accuracy at room temp. most of the usual sensor like the sht15/10 or the lm series or the dallas ds18s20 have an accuracy of 0.5°C.
thanks
giorgio

Ran Talbott

You have to be careful about the distinction between "resolution" and "accuracy":  there are a couple of chips that I know of (the TC77 and DS1631) that provide readings down to a 16th of a degree C,  but the datasheets say they can have an error of +- a whole degree.  So you can get very accurate relative measurements (i.e.,  "The room is an eighth of a degree warmer than it was 10 minutes ago"),  but the absolute value could be off (i.e.,  it might be telling you the temperature is 17.125 degrees C when it's actually 16.875).

The TC77 is available inexpensively from Mouser (and probably Digi-Key and others,  but I happened to be ordering from Mouser when I decided to buy a few).  I have a couple dozen DS1631s that I bought last month with a plan to mount them to DIP adapter boards for people who don't do SMD soldering (much like Sparkfun and NKC do for other sensors).   If you want to help get me off my butt and doing that by being my first customer,  send me a PM or email,  and we may both benefit  ;)

Ran

giorgiodidon

hi, i'm not so interested in resolution (2 decimal after point is enought), what i'm looking for is a temp sensor that can give the absolute temp with an accuracy of more then 0.5°C let's say 0.1-0.3°C
giorgio

retrolefty

0.1-0.3c absolute accuracy is going to be a very expensive proposition. I'm sure there are lab equipment suppliers that might have such measurement systems (would take more then just one sensor) available but the price, calibration requirements, etc are pretty formidable I think.

In the process control industry the gold standard they used when wanting max accuracy of field temp measurements is the 100 ohm platinum RTD sensor but most vendors spec them to be in the 1 degree F accuracy range if I recall correctly.

Lefty

giorgiodidon

you are right, i googled for this kind of high accuracy sensors, the best one seems to be the platinum ones,
but maybe the best way is to apply some kind of compensation to the single sensor (common ones like ds18s20 or sht10 etc.) using 0°C liquefying ice and the given accuracy curve... i'm interested in a short range of temp near 37°C (i'm setting up an eggs incubator!!).

IanH

Why not go for resolution and calibrate to get your accuracy?  If you take water and ice floats in it, it is 0°C, boil water and your at 100°C.  If you can get ahold of a calibrated instrument you could compare something around the temp you are targeting and calibrate to that.

retrolefty

Quote
Why not go for resolution and calibrate to get your accuracy?  If you take water and ice floats in it, it is 0°C, boil water and your at 100°C.  If you can get ahold of a calibrated instrument you could compare something around the temp you are targeting and calibrate to that.


Well absolute accuracy can only be calibrated if the calibration procedure and equipment used is something like 10X more accurate then the basic sensor being calibrated.

Take for instant using boiling water as a 100C reference temperature. Is it really exactly 100C ? Water boils at 100C when the atmospheric pressure is at a certainly exact barometric pressure. What is the barometric pressure at when you perform your calibration? I guess you would have to measure the pressure and calculate the error, with what? and what is it's accuracy and how will you validate the calibration of the pressure sensor?

It's all a chain and circle thing and trying to validate a temperature measurement  to tenths of a degree is a very expensive and time consuming process. Ice bath and boiling water are very easy, inexpensive and useful calibration sources, but they won't validate accuracy to tenths of a degree.

Lefty

divdog

hey how did the incubator project go?

giorgiodidon

hi, the incubator worked well, i calibrated the temperature with a fever thermometer (that has a 0.1 C accuracy), i noticed that the ds18s20 had already a good accuracy.
than i used a servo and an aggs plastic container to move the eggs during incubation period.
for heating i just used a solid state relay for turning on an off some 12v halogen lamps.
next time i will use some IRL mosfet and a pid algoritm to improve temp stability.
do not chose glass container such as old aquarium for incubator set up they are not suitable do to humidity condensation, i had some diffuculty to maintain adeguate humidity level...
giorgio

Go Up