Announcement

Collapse
No announcement yet.

Occasional invalid or wild kWh Values

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

    Occasional invalid or wild kWh Values

    I have just started to put together my first Z-wave system with the primary goal of monitoring energy usage in my home. My setup is only a few weeks old at this point.

    I bought most of the parts many months ago but only recently started installing them when my refrigerator went over to the dark side!

    My setup is as follows:

    Micro Smart Switch DSC18103 (one on Fridge and one on Freezer)
    Fibaro FGK-101 with DS18B20 (Mounted on Fridge)
    Smart Switch DSC24 (On soda machine)
    Home Energy Monitor DSB09104 (on house mains)
    HomeTroller Zee S2
    CentoOS 7 running Cacti
    The poll time for each of the energy monitors is set to 2 minutes.

    Cacti queries the Hometroller every 5 minutes with a JSON "getstatus" request for the "value" from each of the devices, which is then saved in a RRD and can be graphed.

    My problem is that there are occasional wild readings.
    For example, last night at 2050 the HEM (on only one phase) jumped from 11.927kWh to 44.603kWh and then back to 11.988kWh.

    At first I thought my Cacti script was somehow messing up, so yesterday I implemented some diagnostic logging in the script. This confirmed that at least some of the glitches were in the data from the Hometroller.

    Then I looked in the log file on the Hometroller and found a matching entry:
    "Mar-17 8:49:36 PM
    Z-Wave Device: Node 7 House Mains kW Hours 1 Set to 44.603 (44.603 kW Hours)

    These "glitches" are on both phases of the HEM and are frequent enough so far as to make graphing almost worthless. They have occurred on the other energy monitors, though not nearly as often.

    I did a log of digging around and was able to find a few references to "wild" readings. Some discussion of firmware versions but no real guidance.

    I've been working with electronics and computers all my life but am a complete noob with z-wave, so I don't know how robust it truly is. I remember reading trade-publication articles about it when it was being developed and being very impressed.

    The possibilities I see at this point are
    1) the energy monitor is supplying a bogus reading.
    2) The data is being corrupted in transit on the mesh network
    3) The hometroller is corrupting the data when storing it.

    Personally, my money is on #1.

    I guess my question is, does anyone have any experience with this sort of thing or can anyone provide some guidance as to where and how to figure this out?

    Thanks, Ken
    Attached Files

    #2
    Is there a chance the data is accurate - such as a refrigerator/freezer/heat pump turning on with the initial current surge? (Do the specific fridge/freezer sensors show the spike...)

    Comment


      #3
      I wish it were that simple.

      The data that I am reading from the device is the kilowatt-hours (kWh) which should always increase until it hits the upper limit and then it will wrap to zero.

      I am using a program (Cacti) that normally is used to record data through network switches which is also an increasing counter, though they can wrap more frequently.

      The value is read every 5 minutes. In my "spike" example, the reading started at 11.927kWh, 5 minutes later it reported 44.603kWh and 5 minutes after that it reported 11.988kWh. For the kWh to increase 33kWh in 5 minutes, the house would have to draw over 3,000 amps! And the kWh would never go back down.

      Once I filter out the bad reads the data looks more like you would expect and I can see the effect of various loads in the house.

      I have implemented some filtering in Cacti to discard values that are out of range and my graphs are now usable, though I am sometimes seeing large gaps of missing data. I would really like to get to the bottom of this, but it's going to take some time and effort.

      Comment

      Working...
      X