Temperature discrepancy

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts
  • therealfronty
    Automated Home Guru
    • May 2021
    • 140

    Temperature discrepancy

    So why does this happen?

    20211015_203336.jpg

    20211015_203348.jpg
  • kevinsmart
    Automated Home Ninja
    • Sep 2018
    • 257

    #2
    There is a nice description by DBMandrake here:

    Comment

    • therealfronty
      Automated Home Guru
      • May 2021
      • 140

      #3
      I've got to be honest I didn't really understand all that, it seems like there's some biasing and averaging going on, but I don't really understand why the wall display can't just display what the sensor is reading, would've thought it would be quite simple rather than trying to overcomplicate it with algorithms.

      Something else I don't like is why the sensors only display 0.5 degree increments when they are obviously more sensitive than that. I have no issue with set points incrementing in half-degrees but I would actually quite like to know how many tenths away from a set point the actual temperature is, then I might understand the TPI schedule and random boiler firings a bit better. I just feel that 0.5 degrees accuracy is too coarse.

      Comment

      • DBMandrake
        Automated Home Legend
        • Sep 2014
        • 2361

        #4
        Originally posted by therealfronty View Post
        I've got to be honest I didn't really understand all that, it seems like there's some biasing and averaging going on, but I don't really understand why the wall display can't just display what the sensor is reading, would've thought it would be quite simple rather than trying to overcomplicate it with algorithms.
        Most normal people would agree with you, but Honeywell don't.

        None of the temperature displays on the Evohome devices including wall thermostats show the real temperature. They are all biased towards the set point up to half a degree then rounded to the nearest half degree after that.

        As there are delays in temperature readings being sent between devices it's perfectly normal to see a 0.5 degree discrepancy between what is shown on two different devices for a period of time as well. (Also I get the impression that not all the devices use the exact same algorithm so there could be a few raw temperature readings that result in a different displayed temperature on different devices due to slightly different biasing and rounding algorithms)
        Something else I don't like is why the sensors only display 0.5 degree increments when they are obviously more sensitive than that. I have no issue with set points incrementing in half-degrees but I would actually quite like to know how many tenths away from a set point the actual temperature is, then I might understand the TPI schedule and random boiler firings a bit better. I just feel that 0.5 degrees accuracy is too coarse.
        Agreed, but unfortunately the only way to get the true temperature reading (which is to two decimal places not just one) is to build something to intercept the wireless traffic, or poll the V1 Web API for the data.

        My temperature graphs which poll the V1 API show raw temperature to two decimal places, and that makes for much nicer smoother temperature graphs than one rounded to 0.5C.
        Last edited by DBMandrake; 18 October 2021, 11:52 AM.

        Comment

        • MJNewton
          Automated Home Jr Member
          • Jan 2021
          • 21

          #5
          Originally posted by DBMandrake View Post
          [...] or poll the V1 Web API for the data.
          Do you know if the Lyric T6 range can be polled via the v1 web API? I am currently using the v2 API to read various things from mine but might investigate the v1 API if other information is available through it.

          Comment

          • DBMandrake
            Automated Home Legend
            • Sep 2014
            • 2361

            #6
            Originally posted by MJNewton View Post
            Do you know if the Lyric T6 range can be polled via the v1 web API?
            I only have an Evohome system so I don't know. Give it a try! It will either work or not. I use the watchforstock Python client bindings - they work with both API's.
            I am currently using the v2 API to read various things from mine but might investigate the v1 API if other information is available through it.
            The only benefit in the V1 API, at least for an Evohome is fine grained temperature readings. (Two decimal places with no funky biasing or rounding)

            In all other regards the V2 API is superior. For example the V1 API doesn't support reading or modifying schedules. The V2 API also uses a better authentication mechanism.

            It's possible to poll both API's for data, (temperatures from V1, everything else from V2) however there is some suggestion from my experimenting that the authentication rate limiting that Honeywell applies might apply cross API - so that it would count towards rate limiting as two separate authentication attempts in a short space of time.

            Whatever method you use it's important to try to cache and re-use session tokens as much as possible to minimise fresh authentication attempts.
            Last edited by DBMandrake; 20 October 2021, 07:31 PM.

            Comment

            • MJNewton
              Automated Home Jr Member
              • Jan 2021
              • 21

              #7
              Originally posted by DBMandrake View Post
              I only have an Evohome system so I don't know. Give it a try! It will either work or not. I use the watchforstock Python client bindings - they work with both API's.

              The only benefit in the V1 API, at least for an Evohome is fine grained temperature readings. (Two decimal places with no funky biasing or rounding)
              I'll take a look - thanks - although I my cursory search has only found documentation for v2 but I'll keep looking, and I'll look at pre-written clients rather than roll my own like I did with v2. Speaking of which...

              The V2 API also uses a better authentication mechanism.

              It's possible to poll both API's for data, (temperatures from V1, everything else from V2) however there is some suggestion from my experimenting that the authentication rate limiting that Honeywell applies might apply cross API - so that it would count towards rate limiting as two separate authentication attempts in a short space of time.

              Whatever method you use it's important to try to cache and re-use session tokens as much as possible to minimise fresh authentication attempts.
              Don't remind me. I was completely new to OAuth and found it a very steep learning curve, not least given the non-obvious impact of rate limiting (in terms of ambiguous non-specific error messages that Honeywell return) which when you're experimenting and finding your way can be extremely frustrating and very off-putting!

              Comment

              Working...
              X