Can’t find device

I am having difficulty finding what device takes 108w every couple of hours for maybe 20 minutes or so. I’ve turned off every breaker I could while the device is drawing power but can’t find what circuit it

is drawing from. The only circuits I can turn off are the 220v one sense is connected to and the 110 circuit the router and network is connected to. The device seems automatic and turns on at all times of the day. I have isolated any fridges or other things I can think of but can’t find what this mystery device is.

The ~700watt spikes at the beginning of each cycle, and the nature of the 20min on cycle, looks like an AC constant-speed motor … got a pump somewhere in the mix?

You said you isolated the fridges, but that picture looks exactly like my kitchen refrigerator.

1 Like

I kinda agree but it’s fairly sparsely spaced. Maybe for an older freezer. Well stocked? (@foxislandsolutions) … the more full an infrequently accessed freezer is the slower the cycling should be. Some zoom-ins and zoom-outs over time on the waveform might help @foxislandsolutions.

Having a Smart Plug on the fridge of course helps … but it also most likely will show a different waveform than Sense-alone extracts.

There’s lots of analysis in the forum, including my own recent take on fridge cycles here.

I (and others) have posted detailed fridge waveforms like this.

My fridge compressor, for example, runs on a DC inverter and you can see that ramping in the waveform. You can also see the daily auto-defrost heat cycle.

See also the community waveform library. I’ve posted my refrigerator there as have others:

I definitely isolated my freezer and refrigerator circuits. When my mystery device was running I turned off both fridges and it still was pulling 108w.

Just because the Sense continues to show a 108w device-related waveform does not guarantee that that waveform is not associated with an unplugged device. It’s not exactly intuitive.

If that was the case, the unit doesn’t monitor in real time. I unplugged and disconnected the device while sense still showed it pulling power.

It depends what you’re looking at … hence the unintuitive aspect.

If you are viewing the overall Power Meter view for your whole panel then yes, indeed, you would (should) see power drop by 108w if the device (fridge) was on at the time. As the fridge was on, which you can recognize I assume because it’s compressor would be making noise then in pulling the power and seeing the 108w drop you would guess it’s using 108w. Good.

If however, as seems to be the case from your screengrab, you were viewing the Device-only power meter there is no guarantee that that interpolated (by Sense ML/AI) waveform will “switch off” (go to zero) when you unplug the device. The easiest way to think about this is to imagine you had, say, 3 fridges but Sense had only ID’d one of them. What would you expect to happen to the Device waveform in that case as you go around unplugging fridges? The device waveform is rendered from “realtime” power use, sure, but the decision that that part of the waveform (the device) is a device is based on a machine-learned decision – that could be wrong.

There are probably more elegant descriptions of this elsewhere.

Just like Sense learns what a device looks like when it turns on, it also leans how a device turns off. So a Fridge compressor that turns off because the fridge controls turned it off may look different to Sense than when the breaker was thrown. If it sees just a 108w drop in power, it can’t assume that it is the same 108w that it has identified as the fridge.

I believe the Sense does have some logic in it that if the instantaneous power meter goes below the total of the devices it thinks is on, it may try to clean up, but I don’t think that clean up is right away.
There are parts of Sense that are really real time (instantaneous power meter / overall powermeter ) , and there are parts that are “mostly” realtime with some wizardry going on at the same time (bubble view, specific device view).

To @ixu and @ben’s comments, I am now totally confused. Are you saying that what I see on the device power meter might not actually be the device? I have verified all the devices ID’d by sense by catching it when it is on and then shutting it off. Sense may have given it the wrong name, but it is still a specific device.

Hi @jkish

tl;dr - flipping breakers ON/OFF while watching a device detail screen is not a fool proof way of identifying a device

What I’m saying is that what you see on the Device power meter is Sense’s extrapolation of that signal out of all the other electrical signals in the house.
looking a your post

I was reacting to that turning something off by breaker while looking at a device status page and the status not changing doesn’t actually mean that the device didn’t turn off (according to sense). So in your specific case, just because Sense said the unknown Device was on, and you turned OFF the breaker for the fridge, and the unknown device didn’t turn off, doesn’t actually mean that you can eliminate the fridge from the possible devices that unknown is. I’m saying that Sense tracks HOW a device usually turns off. So you turning off the breaker to the fridge may not have triggered Sense’s systems into recognizing that the unknown device (possible fridge) had been turned off.

To test this - I would recommend that you turn off the breaker for your Fridge or 3-4 hours and see if the Unknown device still shows up. If it still shows up, then you know its NOT the fridge and can keep hunting.

Indeed @ben it starts to feel like stereotypical Internet tech support: “Just shut it down for 15 minutes and then restart, that should fix it!” :grin:

@jkish I wish somebody would come up with a really good quick method of explaining this … I’ll add to @ben’s and others’ attempts and try and keep it generic:

  • Imagine your Sense has detected a 10w lightbulb (we’ll call it Light1) as a Device. Great!
  • You switch the light on and off and see it perfectly in your Bubbles and in Light1’s Power Meter.
  • Also, when you look closely you see that the overall Power Meter goes up and down 10w with the light being on and off. Wonderful!
  • Then you install another 10w light, we’ll call this one Light2. Here we go!
  • You switch Light2 on and off and the Light1 Bubble and Power Meter show it perfectly. Oops!
  • What to do?

This may seem like a trivial example (it is) but it represents the near impossibility of Sense reliably tracking all devices. In this example, any device (e.g. a fridge) that has a 10w fluctuation in usage (not uncommon) could fool Sense into thinking the light just went on … or off.

To stretch things a little further … even though a 10w light has a straightforward electrical signature that doesn’t fluctuate up-and-down, because it has (in the “average” home) low power usage it is actually very hard for Sense to separate it from the other noise in the home like a fridge or a washer dryer or another light. When something like a hot water tank comes on and uses 4kW it’s usually the only device doing that so eventually Sense has a good chance to recognize it.

And one more step into the beyond … Sense’s learning could of course (and does, as far as I understand … please correct me techies) incorporate usage patterns. For example if Light1 above only gets used on Monday-Friday and Light2 is Weekends-only, then in theory Sense can (does?) use that information to differentiate Devices. Not that that helps necessarily if the fridge is cycling on and off all week.