Shop OBEX P1 Docs P2 Docs Learn Events
Digital thermometer with better than 0.5C resolution? — Parallax Forums

Digital thermometer with better than 0.5C resolution?

CuriousOneCuriousOne Posts: 931
edited 2013-10-08 22:06 in Accessories
Hello.

I'm looking for digital, factory calibrated thermometer, which will have better than 0.5C accuracy in range of 15-95C. Does such ones exist? so far, I've found only MAX31723 to be close to required specs.

All interfaces will be OK, but SPI is preferred.

Comments

  • WBA ConsultingWBA Consulting Posts: 2,933
    edited 2013-10-05 17:15
    Check into Sensirion SHT11 or SHT15 series sensors as they have 0.01C resolution. They measure temperature and humidity with an interface very close to I2C. Sensirion also makes newer sensors that communicate in true I2C, SHT21, but I have not seen any breakout boards for them. They are very easy to use. Breakout boards are sold by Parallax, Sparkfun, me (both SHT11 and SHT15 versions), and a few others I am sure.

    PS, right now Parallax's is the deal of the day too!
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2013-10-05 21:05
    CuriousOne,

    Your thread title mentions "precision." The body talks about "accuracy." Those are two very different things. Which do you mean?

    -Phil
  • CuriousOneCuriousOne Posts: 931
    edited 2013-10-05 22:00
    I need to have temperature reading accuracy in range of 0.1C-<0.5C for temperatures from 15C to 95C. And also I need good speed, at least 2 readings per second.

    Regarding the accuracy vs precision, according to picture in wiki, I need them both :)
  • Tracy AllenTracy Allen Posts: 6,658
    edited 2013-10-05 22:42
    You have to read the data sheets carefully, I mean, read between the lines. The MAX31723 looks pretty good. I like the way they present the accuracy data:
    Screen shot 2013-10-05 at 10.25.51 PM.png

    It shows a typical (average?) accuracy of +/- 0.1 °C over a wide temperature range, bracketed by a bowtie curve at 3 standard deviations on each side. That would make it seem that if you were to draw a random specimen of the part from a parts bin, you would stand a pretty good chance of hitting your required accuracy. At least, at around room temperature. The spread is much higher at your high temperatures.

    Another one to consider is the LM92, i2c interface, claiming +/- 0.33 °C limit of error at 30 °C. If you dig into the data sheet though, it is not so good as it moves away from room temperature.

    Andrew mentioned the Sensirion products. You didn't mention needing RH, but Sensirion also offers the temperature-only STS21 (i2c interface) which claims a typical 0.2 °C "typical" tolerance over a fairly wide range.
    Screen shot 2013-10-05 at 10.36.15 PM.png


    What is the application or location? Sometimes you can improve the results with your own calibration, in which case you can get along with a generic (cheaper) sensor.
  • CuriousOneCuriousOne Posts: 931
    edited 2013-10-05 22:48
    My friend's son wants to build Thermal cycler: http://en.wikipedia.org/wiki/Thermal_cycler

    Sure, there are a lot of DIY ideas available online, but most of them are either way too simple and less precise, or precise, but way too expensive ($599 for thermal cycler DIY kit, and case is made from plywood :D) So we want to build something that will have good accuracy and precision, but won't have aircraft prices.

    I want to build it around BS2 oem and parallax usb datalogger.
  • Tracy AllenTracy Allen Posts: 6,658
    edited 2013-10-07 08:40
    I'd just forge ahead and build it, because there are bound to be a host of issues more problematic than the accuracy of the temperature sensor itself. The temperature sensor(s) will be located in thermowells in the block, so when the time comes for the final fraction of a percent accuracy, you can at that point decide on a really accurate sensor. Either you calibrate the individual digital sensor against a standard, or move to an intrinsically more accurate sensor in the form of a thermistor or RTD and hi-res ADC.

    The friend's son is not named Frankenstein is he? :-)
  • CuriousOneCuriousOne Posts: 931
    edited 2013-10-08 22:06
    They want to merge potato with tomato, so plant will have potato roots and tomato berries. So far, they're getting totally opposite result :-)
Sign In or Register to comment.