Announcement

Collapse
No announcement yet.

Composit Error Rate

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

    Composit Error Rate

    Michael,

    How is this error rate calculated? I'm currently showing a composit rate of 624% with the sensor monitor showing 1 error 3 days ago and the 9x9x monitor showing 12 errors, last one being two hours ago.

    Thanks

    Dennis
    Not too old to learn

    #2
    It is intended to be a moving average of error counts heavily weighted to any errors that occurred in the last iteration. I can see that an algorithmic error.

    I think I will change it to show the relationship between number of attempts vs number of errors. It will still be a moving average, but have a more intuitive interpretation.

    Comment


      #3
      Michael, I really appreciate your product especially since it now takes advantage of WeatherPlug data. My home is in the woods and a local weather station is useless for anything but temp and humidity. Hope you don't find me a pill when I prob your calculations. I'm not a programmer and I wouldn't enjoy the functionality I do without your work. Thanks

      Dennis
      Not too old to learn

      Comment


        #4
        Error rate algorithm has been changed. It is now 80% weighted previous vs. current iteration and reflects the number of errors vs. number of opportunitites for errors.

        In the case of the Temp05 each sensor reported is an opportunity and an error is ??? returned for any sensor or 85C returned for Temperature.
        In the case of the DS9x9x opportunities exist while engaging the adapter to query the bus and an opportunity exists for each sensor on the bus. These are reported as adapter and sensor errors respectively.

        In 4.6 the DS9x9x algorithm was made to perform a retry after first sensor error and 5 retries when attempting to output a discrete and the discrete's state does not report the desired state. Initial errors are reported in the error count so it is possible to have an error rate, yet still have totally valid data due to the retry. I elected to count every error so degraded bus physics would not be masked by software retries.

        Comment


          #5
          Super! Getting the true error rate of the buss structure is what I think most of us really are interested in so if we have a problem on the physical network we can focus on tracking it down.

          A request for the future if you don't mind. Averaging the "location" and "City/State" data from multiple WeatherPlug stations are meaningless. Is it too much trouble to just not create those "averages" and drop the virtual devices?

          Thanks

          Dennis
          Not too old to learn

          Comment

          Working...
          X