Minimum Consumption Threshold for Detection?

Hey All! As my sense unit continues to identify more and more large-load devices in my home, it seems to be picking up some steam in identifying loads (it makes sense - breaking apart the “Other” bubble would give a clearer image of the remaining energy usage).

I have used HS110s & HS300 to clear up some messy, difficult-to-detect waveforms - like my TV/Home theatre, and fridges. Integrated icemaker, ice dispenser, water dispenser, compressor, defrost coil, circulation fans, etc offered a nearly continual changing energy graph - a difficult task for detection.

That said, I have been making efforts to detect some of my smaller loads (bathroom fans, outdoor LED lights, garage lights (all garage lights on one switch). For example, in the latter hours of the evening, while energy consumption is minimized and fluctuating the least, I will run the bathroom fan timer and let it run a cycle uninterrupted (45 mins; in hopes of helping to capture a clean energy trend).

My question is: is there a minimum threshold that Sense uses to filter out trivial “noise” when detecting devices?
If so, is it possible to identify/reduce this threshold? Might a threshold prevent detection of smaller load (ie. 30-50W) devices? Or is a smart plug/strip the only solution?

Thank you all in advance for your help!

Here’s a thought. It’s just my theory, but I think that the tagged transitions in the iOS/Android Power Meter are the only ones that the Sense monitor registers as “transitions of interest”. Others might be too small or too slow (an EV charging ramp, for instance). Even if a transition is flagged as “of interest”, it doesn’t mean that has been identified - that might come later or even never.

So zoom way in on the Power Meter, then try turning on / off devices in question to see if the Sense flags the associated Power Meter transitions. That might give you better insight into which changes Sense is taking note of. Or my theory might be completely wrong.

ps: I can’t see of away to adjust the threshold, especially in the time domain. You might be able to make Sense more sensitive by tempoararily dialing down the transitions from other devices in the house.

1 Like

Yep - it seems that Sense can register changes as low as 10-11W, which made me think of some sort of threshold (or something similar) which might restrict the device detection, in an effort to seek out the “big fish”…but it’s completely a theory.

The flags you mention makes me believe that it sees everytime the devices are energized, and when it completes its cycle or is powered off. :+1:t3:

I’m not sure Sense is trying to find the “big fish” first. My view is that the issue with many of the small transitions is their lack of ubiquity - too many are similar and undistinguished in their “features” that Sense uses to identify unique fingerprints. Just to put a “face” on it, here’s a still from a Sense video back in 2018 that explained a little about their identification / categorization. It shows one slice of multi dimensional clustering of “features” used as input to identify devices. In this case, it’s a slice of Feature1 (power) and Feature17(p0 -phase). That implies two things:

  1. There are at least 17 features of a transition that Sense uses to identify/classify it.
  2. Overall power is one of those “features”. But my guess is that the population of small power transitions is far greater than the population of larger transitions, based on looking at my home’s “transition” stream. That means that the low wattage region of the cluster graph is saturated with devices, presumably with only small differences in some of the other features.

So maybe not Sense trying to pay attention to the “big fish”, but the “little fish” too numerous to separate. Picture trying to identify different sardines in school of thousands, vs. different 2-3 different sharks…

The video is here:

2 Likes

My lowest wattage devices Sense has found are

  1. A component of my ice maker: 34W
  2. Ceiling fan: 28W (only 1 of 3 fans and only if switched from off to on with the speed on medium)
  3. Two different components of my air handler (I think it’s just the initial startup because Sense thinks the runtime is only 1s): 15w and 4w

Everything else is 100+ watts except for a single 60W incandescent bulb. I’m not sure if I believe the 4W Sense reports for the one device, but it detects it pretty reliably.

2 Likes

Good picture.

There’s the temptation, in the 2-D graphic and mind’s eye, to see the Sense Power Meter waveform as an adequate representation of the data stream. “I can see it, why can’t Sense!”. The data however is significantly more multi-dimensional and of higher resolution than is presented in the UI.

To re-picture the picture: You may well be able to identify an individual sardine in a still photo of a school but good luck being a shark trying to eyeball one to catch as they are swimming around.

1 Like

@kevin1 You’re absolutely right - I slipped into the basic thought pattern of energy peaks and valleys on a 2-D graph, whereas Sense is considering a multitude of other features. The video you linked showing voltage/current relationships and scatter graph showing leading/lagging phase angles helped to show just how in-depth the device detection really is!

The tech behind these little orange boxes is really astounding - I’ll let Sense continue to do the learning, and wait for the next surprise when it locks onto another device! :nerd_face:

Thanks everyone for speaking up and helping a newb to the community!

1 Like