7 weeks in, first accurate device -- how's that compare?

I know they say it takes “a while”, but I’m curious how my experience compares to others. At seven weeks, I have what I think would be considered its first accurately identified single device – an incandescent lamp (found as “Unnamed device”)

By ‘accurate’ here, I mean the Sense alert for this device is only triggered by this device, and no other actual devices – it’s not a composite. Sense has detected a dozen other “devices” but they are composites of actual devices (eg, my electric kettle, toaster, microwave, and dishwasher are seen by Sense as one device, when not unseen altogether) But I don’t mean Sense always detects the light. It only detects it about half the times it’s on. But at least when it says the lamp is on, it is true.

I’ve seen a few Sense owners that seem to report better/faster results, but the forum is pretty new and there’s no way to tell how representative it is. I have to admit that so far it’s not only not useful, it’s hard to say it’s even interesting.

Is my Sense performance an outlier, or is this typical?

I would have said you are slow… but since you are adding in the definition of “accurate” you might be similar to many.

As an example this weekend I tried to properly label my refrigerators in Sense. I have two regular fridges and one smaller wine refrigerator.

I decided to unplug them to isolate them and then label them. I did this on a Sunday morning with virtually no other loads on in the house other than “always on”. Long story short none of the 2 devices Sense had identified as a refrigerator correlated to my actual refrigerators and I could not figure out what they were. When I unplugged my refrigerators the Sense devices stayed on.

Furthermore Senses see both my washing machine and my garbage disposal as the same device and there are a few other similar examples.

Net I have about 15 devices found in Sense… I can trust only about 5 of them… the other 10 are either unknown or are confused for multiple actual loads.

How long has yours been running?

Did yours actually say it had found ‘refrigerator’? All mine are “Unnamed device #n” and “Unnamed motor #n

6 Weeks

Yes it found Fridge and Fridge 2

I think you’re within the bell curve. I’m about 8 weeks in (after having my original device replaced), and I have 18 devices found. Of those, there are 9 actual appliances - more or less correctly identified, but many with separate legs of power reporting as separate parts (I’ll be posting another thread shortly). For example, my stove is identified as three different devices so far - 2 legs of the actual stove elements and one big burner. My heat pump is recognized as two units based on 2 legs of power. My water heater, on the other hand, is hit or miss identified as one whole unit or half a unit, or sometimes a whole unit but only part of the time, or sometimes not at all. I also have some device that’s identified as drawing relatively low wattage (70-90w), but I think it’s not picking it up all the time. The one time it was on long enough to d/c appliances, I ruled out the fridge, the television, and everything else…except for my router bundle. Obviously that one is a little difficult to rule out, so that’s what I’m left with.

Not complaining at all about device discovery. Again, I think you’re within the bell curve. Sometimes it’s certain about devices and you don’t have to rename them (like my microwave and garbage disposal). Sometimes it’s not certain and you have to rename them, or it only finds one particular leg or part of an appliance. Sometimes it thinks it’s one thing and you have to rename it. But in the end I think you’re within the norm.

2 Likes

Well, I have to retract the first accurate device milestone: today the dishwasher also triggered this device.

I tried this with my fridges also. I unplugged each one and my Sense ‘Fridge’ stayed on even though I could see the total house wattage drop and come back up. I finally figured out that Sense had identified a fridge correctly but the algorithm is not good at detecting when the fridge shuts down. I was able to isolate the fridge by the wattage it was drawing. The ‘Fridge’ routine needs some work. It’s good at catching the compressor startups but not that great at the shutdowns. One of my units has a separate compressor for the fridge and freezer sections, so Sense should have a ball with that one.

1 Like

@dan Thanks Dan. Great insight and experience. I will try this again to see if I can identify the three fridges I have.

I’m now at 5 weeks post installation and my Sense claims 28 devices.

Confirmed devices:

  1. Garage door opener (I think this is the only device Sense named)
  2. Fridge compressor (Subzero)
  3. Fridge light (Subzero left door)
  4. Fridge light (Subzero right door)
  5. Fridge compressor (Kenmore)
  6. Aquarium heater
  7. Halogen lighting in two different bedrooms (one device for both)
  8. Gas clothes dryer motor
  9. Trane HVAC fan motor
  10. Same Trane HVAC fan (2nd stage)
  11. Instapot
  12. Electric oven (baking element)
  13. Same electric oven (broiler element)
  14. Some device associated with (but not nearly big enough for) the pool pump

The remainder of devices are unnamed motors, lights and other devices I haven’t been able to confirm yet.

My biggest surprise has been that it has accurately identified some very small loads (10 watts) but apparently has trouble figuring out my biggest loads (two EV chargers and a pool pump). These three are easy to spot with hourly smart meter data.

Who cares what the 4MHz load signature looks like when a 2.5kW load comes on for several hours a day at the exact same time? These “identifications” are all guesses, and it seems this would be an easy one. Apparently not, at least for a machine.

Well that’s a lot better than mine. At least yours seems to be finding distinct individual loads (eg, oven element), even if they are just components of an actual device. All my “devices” are composites of multiple real devices. Knowing that the dishwasher|frig|toaster|microwave turned on is pretty useless.

I also have two EV chargers, which Sense doesn’t register at all.

This week it does seem to have learned the frig light. So I get a notice when the frig door is opened. I think this now qualifies as the first unique device it’s learned.

Nice progress Steve. What trane unit do you have?

Interesting regression. After three or four days of correctly identifying my frig light (fired on door opening), today it identified frig door opening as two other composite “devices”. It is understandable that my 1,000 watt toaster can be confused with my dishwasher during its dry cycle (similar resistance heating elements), but how that load can be confused with a 5 watt LED is mysterious.

This regression puts me back to zero correct device detection.

Obviously we as customers have no real good insight into why some people are having success and others like you are having constant problems - but we can speculate. I’d be curious to know the age and general layout of your house (ranch, large 2-story, etc,). Would you mind taking a quality res pic of your panel so the rest of us can armchair quarterback your install? It wouldn’t be unheard of to have an improper install despite a qualified (or unqualified) electrician doing the job (see here for an example: https://www.reddit.com/r/Sense/comments/5raux7/day_16_still_no_devices_found/)

It’s a 2,500 sqft 1962 ranch, with 100% new electrical. 100% LED lighting, except one incandescent (which Sense correctly identifies most of the time; it has not detected any other lights). Most of the lighting is proprietary LED “light engine” type devices, not LED replacement bulbs in legacy fixtures. Radiant heating via gas boiler, so no HVAC electrical except ~10watt recirc pump and negligible-load control systems. Two EVs always on 11pm to 6am. Miele frig. Miele dishwasher used once per day. Miele microwave/electric combo oven used a couple times every other day or so. Blender, cuisinart, kithenaid used maybe once a week. Toaster and electric kettle used a few times a day. Coffee grinder used for 10 seconds every morning. Can’t-remember-brand-nothing-fancy clothes washer and gas dryer used 2-3 times per week. One small Samsung TV. Two always-on iMacs. Cable box, phone chargers, and that sort of stuff.

Now you know (almost*) my entire electrical life – better than Sense I dare say!

*almost: there are a bunch of motors (tools) in garage, but used only for minutes at a time every other week or so, and I don’t think this is affecting Sense’s ability to detect my frig light…

I’d be happy to post picture of device (but not home now), though I rather doubt there is such a thing as a improper install that affects device detection. The install pretty much works or it doesn’t – CTs are correctly placed or not, wireless connected or not. Not much else there.

My Sense did have some initial internal problems. When installed it worked for a day, then died. Came back a few days later, but couldn’t make wireless connection. Jiggled connectors once, came back, then died, etc. Jiggling only worked once however. I followed tech support troubleshooting and they were going to RMA it for replacement, when it suddenly just started working – nothing had changed, I hadn’t touched it or even been near it. So we canceled the RMA and it’s been running since that day without incident.

But I have wondered if my hardware could be defective. I figured after a about the two month mark if still no correct device detection, I’d pose the question to Tech Support. It’s totally unclear to me how much of the detection occurs in the device, and what occurs on their end. Does anyone have insight into this?

Bubuski: It’s a Trane “High Efficiency Two Stage” XL 80.

You’re not that different than me - 2100sf ranch, 1999, all leds, nothing too complicated except for the evs :slight_smile:

Post a pic of the panel and a couple 12hr screenshots (6am-6pm, 6pm-6am, etc) of your usage graph just to see if there’s anything funky. There may not be anything out of the ordinary, but maybe there will be :wink:

All the crunching is done in the cloud - the device is just sending electrical data and the analysis and model fitting happens above our pay grade. If your data is clean, I don’t think you should be having so much trouble. If you’ve got a wonky ct or a hitch in your install, then the data being fed to the model could be bad making appliance fitting difficult.

Picking this up again, was otherwise occupied last week. Here’s current status 8 weeks in.

  • Always detects electric Kettle turn on.
  • Miele combo oven triggers “Kettle turned on” every ~2 minutes while oven runs.
  • Dishwasher triggers one “Kettle turned on” event about 30 minutes into run.
  • Always detects Miele LED frig light turn on (door open). 100% accurate, no false positives.
  • Always detects my only incandescent lamp (two 40w bulbs)
  • Seems to be detecting toaster, not yet clear how accurate.
  • the rest is a jumble.

Picture of my Sense hardware and average day snapshot.

Hmm - do you have solar? I assume so based on the “solar” breaker, but is your Sense hooked up to solar?

And when you say “EV” I assume you mean electric vehicles? And they’re plugged in a little before 6pm yesterday?

Also, it looks like you have solar leads (at least I can see there are a pair of leads plugged into your device), but I don’t see solar on your graph?

And there’s some other panel somewhere? I can’t tell where your Sense is getting power - I don’t see the power wires running up into that panel box with the main CTs. (never mind, I think I see the power leads coming up to that 20a breaker, neutral going to the neutral bus).

I have Sense solar, but by coincidence my inverter went belly up just after I got by Sense, and I have not got around to dealing with it, partly because it’s been so rainy and not seemed urgent. The Sense Solar CTs are in the Solar disconnect, which is just out of view of the photo.

The power to the Sense is the uppermost breaker (unlabeled) in the picture.

Yes, I have two electric cars, however the load from 6-9pm is mostly household I think (alas, Sense can’t enlighten me) The EV chargers are normally on timers 11pm-6am, which is the cheap power window. But sometimes we override the charge schedule and start earlier. We only have 120v (~1500watt) chargers, and 7 hrs is only ~1/3 charge on the car, which is fine most of the time because we don’t use more than that on an average day, but sometimes not and I start charging when I get home ~6ish. I don’t recall what was happening that day (even though just yesterday!)