How accurate is the temperature sensor?
  • I am using twine in my wine cellar to measure temperature..however, compared to my existing equipment which measures both ambient and in bottle temps, the ambient temp on twine seems a few degrees low which may make sense because it is inside the case. What benchmarks / lab tests have you done confirm temp. accuracy? In most applications 1 or 2 degrees is acceptable but not in a wine cellar...would love to connect an in bottle temp probe to my twine to improve accuracy within .5 degree. My current cooling unit keeps the cellar within .75 degrees at all times and want to use twine as a backup in case that sensor fails....
  • 18 Comments sorted by
  • I just noticed the same thing. I have 4 different things sitting on my desk right now measuring temperature:

    Oregon Scientific THN132N - 78.8°
    Oregon Scientific THGR122NX - 78.4°
    Smarthome INSTEON Wireless Thermostat (2441ZTH) - 80°
    Twine - 75°

    I even took off the cover to see if that would make a difference - doesn't appear to. All 4 are sitting literally inches apart. Not sure what to make of that.
  • A man with two thermometers never knows what temperature it is! The cover will slow down how Twine's temperature reading adjusts to changes, but it should get to the right temperature. There's little variation between Twines. Note that if temperature will go up a few degrees if you're in setup or fast mode - communication makes Twine work harder.
  • My twine's temp is showing it to be 6 degrees higher than what it is in the room. If setting up the twine or updating it increases the temperature as suggested above, it would have been helpful to put that information on the website so that a user will know where there is a disparity.
  • In between bug chasing, we're writing real documentation for each sensor. I'll be sure to include that setup and fast mode drive the temperature up. But in normal operation Twine uses a minuscule amount of power, and won't create an amount of heat outside the margin of error.
  • Just a follow up: I intentionally raised the temp in the room by a couple of degrees. The twine is staying about 2°-3° cooler than the other sensors. I wonder if perhaps there's a calibration issue?
  • My Twine is located outside my house. It measures within a couple degrees (+/-) of the Airport temperature which is very near my home. Close enough.
  • The green cover may be causing this. I'm sure the sensor is accurate, but it may take longer to get precise readings...
  • Amprobe TH-1 = 70.9%

    Twine = 71%

    pretty close. wonder how thisvaries from Twine to Twine
  • More thoughts - the case may increase response time and may also affect internal ambient temp by acting as a thermal insulator, especially if external temp is close to internal temp. I will run calibration tests against an Amprobe temp meter.
  • I'd be interested in results when running with the cover off, as well as on. With power being consumed in the unit, I'd expect a temp higher than the ambient air temperature.
  • My twine is also showing a temp way too high for the room.
  • I'm getting a higher temperature here as well, at a rough guess 4-5 degrees celsius.
  • I have removed my cover and it is still reading 2-3 deg F low to the monitoring equipment that is inches away.
  • I've put twine on top of my room thermostat which is at 18C. Twine reports 61F now which is 16.1C. It's easy to compensate the temp difference in the rule set so for me not a big deal. But for some reason It seems I only get an e-mail when I use the temperature rule in Fahrenheit. anyone else experienced this?

    Edit: reading is now 64F which is only a 0.3 difference with the room thermostat. That's pretty good!
  • Thanks for reminding us that the temperature differences in the sensor can be compensated for with the rules (so long as the difference stays constant). I had been thinking that it would be nice to have a user supplied offset in the hardware/firmware, but that's not necessary.

    In another thread the developers have acknowledged the problem with the Celsius rule, so presumably it will get fixed.
  • Just a note, that if you are comparing the Twine temperature to some other sensor's temperature, you have an instrument pair where each one has it's own error. Unless one of the pairs is a lab calibrated thermometer, you can't be sure what amount of error to attribute to each.
    There is also a granularity issue. If you are working in Fahrenheit, your accuracy is almost twice what it is in Celsius, because the Twine reports only in whole number degrees. Thus, if your "trigger" temperature needs to be fairly accurate, working in Fahrenheit is preferable.
    Right now, my Twine is reporting 74 F, and a mercury in glass chemical lab thermometer is reading 23.9 C (=75 F). This kind of thermometer comes with and accuracy of +/-0.2 C. For the picky, the thermometer bulb is right beside the middle of the Twine's bottom, and the Twine is resting on its "right". (I have no idea where the temperature sensor is actually located on the Twine circuit board.)
    So roughly speaking, my Twine is reporting about 1 F low, at these particular conditions. (Note that the Twine could actually be reading 74.49 F, which would round down to the displayed 74 F, and the actual temperature could actually be 23.7 C (=74.7 F.)
    Since most users will probably use the temperature for much cruder decisions than would require fractional degree accuracy (has my furnace died and my house temperature gone below 45 F; or, has my freezer stopped working and is about to thaw food), in engineering terms, my Twine (at least my Twine) is close enough for practical purposes.
  • mine is sitting within 5cm of three uncalibrated digital thermometers. all three say 19c. the twine varies between 19c and 22c, usually 21c or 22c. imiho, this is not good enough for reporting ambient to a human.
  • I just checked my twine against the thermocouple on my fluke 87. It is within +- 1F. The twine has considerable thermal lag. Good enough to help monitor my garage to keep it from freezing

Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!