Always On Time Threshold

Device Detection
, ,
#1

I have no more “other” bubble, everything is either identified as a device or lumped into the always on group. What is the on duration threshold for being classified as on? Is there a way to adjust it? If not can we get that as a feature? Besides few dozen IoT devices the rest of the “always on” devices are my lights which certainly are not always on.

#2

Here’s how Always On for the top level is computed every half second. There’s not really a threshold, but it sums the 1% bin for power across both legs for the past 48 hours.

Not sure how you thought Always On was calculated, but it doesn’t look at individual device behaviors with time window thresholds.

If you do want to see power usage for lights on that Sense is not detecting today, your options are:

  • Move to Hue bulbs and use current Sense / Hue integration
  • Find some other smart lighting system that has an integration with Home Assistant. Use Home Assistant and SenseLink (all open source) to back-feed power usage from Home Assistant calculation into Sense.
  • Figure out if there is a way to McGuyver an already integrated KP115 or KP125 into a controller for your lights.
#3

I am not quite sure what you mean by “sum the 1% bin”… 1% of what? 48hrs is certainly a threshold of sorts, and an arbitrary one at that. In my bubble view “other” only shows up when my high wattage grow lights are on. When they disappear, the other bubble disappears. This means my “always on” is getting filled with stuff that has multi-hour power transitions room lights and computers… all of which have power draws in the +100W range. So something is clearly not right if dynamic devices are getting classified as always on.

#4

Thanks @Rmh3093 , you originally mentioned just lights, but now you mentioned lights and computers. I understand where you are coming from - hard to have all usage subsumed by Always On if you have a bunch of dynamic loads. It also sounds like most (or all) of your dynamic devices are not getting picked up via Sense native (transition based) detections. Not a surprise if most of your loads are LED lighting and computers.

If I were you, I would do 3 things.

  1. better understand how Always On works. It looks at the power usage on each of the legs in your house over the past 48 hours. That’s 345,600 1/2 second samples, per leg. The 1% bin will be the 3,456th smallest on each leg. Add those values together to get the Always On for that half second. 48 hours was not chosen arbitrarily - it was chosen to look for real Always On - going shorter would only increase your Always On value. Longer wouldn’t change it much.
  2. I would put each computer (and anything else pluggable) on a traveling KP115 / 125 smartplug for 48 hours or more to have Sense calculate the Always On for each computer and to see the usage patterns.
  3. I would also carefully consider my Power Meter patterns to see how much variation you see over 48 hours. If you see more than 30 minutes in 48 hours below your Always On level, you might have something weird going on. If you want to post a waveform for 48 hours, others can take a look.

There are some rare situations where lots of overlapping devices with different on and off times could overlap and layer in just the right way to prevent Sense from seeing any low water marks that are truly representative of the real Always On over 48 hours. But highly unlikely.

1 Like