Sense Correctly identified Dishwasher but is tagging the Toaster too against that category

So, a week into the install and Sense correctly identified the Dishwasher. it also identified “heat 1” which correlated with the Heating cycle on the dishwasher but to my surprise, this morning, the Toaster which Sense has still not idenfied was tagged as Dishwasher used for 4 minutes.

How does one go about correcting these things? Is it a good practice to just delete the device and let Sense re-identify the devices again?

1 Like

Two things:

  1. Having real devices showing up as a single Sense device is a common challenge know as device conflation. It’s especially common for heating devices because pure resistive heating doesn’t give up many clues that distinguish between two devices that use roughly the same wattage. Sense tries to make sure a a device is unique (that there aren’t any similar, but not quite the same devices) before making an identification, but sometimes two devices slip through. And sometimes one device (perhaps the toaster) is only used after Sense has found the first device, so Sense never had the chance to fine tune the detection.
  2. You can try deleting the detected device that is doing double duty. You might also want to try to use the toaster a few more times in are row if you haven’t used it frequently to give Sense additional samples before it tries to re-detect the dishwasher. That might give it more information to distinguish the two.

ps: I live with conflation on a regular basis - I have a floor heating loop that is within 20W of my dryer. Both show up as the Dryer in Sense. Haven’t been able to fix, but it doesn’t bother me that much.

Thank you for that information @kevin1
Btw, how long before a device is identified again? I had the refrigerator show up once and then I think the freezer u to in refrigerator showed up as another device but something wasn’t adding up - I deleted both and they haven’t been re-identified. It’s been almost a week.

How long is not entirely predictable. Basically, Sense keeps looking at all on and off transitions until it sees on and off patterns that are unique and can be paired up (on with off). Transitions from new devices or existing devices that start showing changes in their patterns can slow redirection based on what I understand.

For whatever reason your explanation of conflation doesn’t seem relevant in my situation. Case in point
After I installed Sense my coffee pot and toaster went undetected for months. Then winter came and the very first time I used a space heater bang it was detected and I added it. It measured correctly all thru the winter. I unplugged it and a couple days later I see a space heater bubble. That can’t be right! A little research and I found the conflated item was my Coffee pot or the toaster.
During the entire time the space heater was plugged in I never saw any usage for a coffee pot or toaster so guessing they appeared as “other”. correct?? No requests to merge anything either. The coffee pot is on a timer and used same time everyday.

I deleted the space heater once before trying to detect a 2nd space heater (different brand, different wattage) with no luck.

Soooo my dilemma is should I delete the space heater and see IF the coffee pot is detected now.? It went months before without being detected.

@ron111157, not sure I have any good answers on what’s best going forward. First off, based on what I know, I can see how Sense could behave the way you experienced, when confronted with a number of similar resistive loads at different time points. Take a read of my thoughts on Sense operation here if you are interested in more discussion, especially the parts on the “problem with clusters” plus the section on how models improve / change over time.

There are a bunch of behind-the-scenes adjustments in clustering and in model deployment at the Sense mother ship, that can change the connection between device characteristics and detections. But how and when each kind of action is triggered, is a mystery to me. Pretty clear that deleting a device is a trigger for some set of actions, but not sure it sets in motion what you would want. You might want to contact support@sense.com, though you’ll probably need some persistence to work this one through them.

Interesting read and most of it sounds reasonable. Problem is from reading a good many posts from here I see many of the same issues going back 2-3 years and they are still not resolved. Just how much time should we be patient for Sense to “crack the code” for common household items that exist in most every home today?

Example: After having Sense for about 10 months now if I remove all the Kasa and Hue bubbles I have 6 or 7 detected items. Many of them are not FULLY sensing the product yet either. That’s is probably 25-30% of the home items
I’ve have not had much success with Smart plugs either. I have 11 and NOT ONE device plugged into them has been detected by Sense. Same goes for Hue. I read elsewhere on here where one guy had 30+ smart plugs!! That is like 200-400 dollars extra to make Sense do what it says it already does. Crazy… not to mention the added cost of powering a smart plug. Talk about phantom power!
Speaking of smart plugs there is what 2 that sense can see? There are dozens of plugs, switches and other items and Sense works with 2? As far as I know Sense works with none of the Smart home systems either.

I realize Sense has a huge task to see everything BUT the big question is WHEN will it do all the things it said it could do?? Guess it’s back to trial and error methods.

ok, done ranting…lol

In my books, the issue is that Sense results are highly variable depending on home dependent on both devices and house noise. And users really can’t know that well ahead of of time how well things will work out. I have seen substantial progress over the past 3 years in Other reduction. I’ll admit that 3 out of my top 10 devices by usage are on smart plugs, but the remaining 7 are all native detections. And my Other is only 7% (the numbers speak for themselves)

I’ll also make a few controversial statements.

  • You shouldn’t expect Sense to detect devices that are on a smart plug. If you are using smart plugs intelligently, you’re only putting them on devices (or groups of devices) that have substantial variable usage and don’t fit the Sense native detection model.
  • The newest Kasa smart plugs use about 1/5 of a Watt, so I wouldn’t worry about them when it comes to vampire power. They will help you find and alleviate a lot more. And I saw the Kasa EP25 4 pack running at 22$ on Amazon a week ago. You could have bought about 60 of them for the 400$ you suggest.
  • The list of integrations is fairly short but it hits some of the most popular and standard smart plugs between TP-Link, WeMo and Wiser. Hue / Philips are the only smart lighting guys that actually build power characterization into their hub APIs. Sense used to have both Ecobee and Nest thermostat integrations, until Google hosed up the Nest-o-verse. The Ecobee integration seems to be helping with HVAC detection, even though the Ecobee feedback isn’t realtime (the data is only available every 5 min).
  • You might actually want to try out the Home Assistant smart home system. It’s open and has the largest ecosystem of all the home hubs. It includes and native integration with Sense, plus just about every other home automation product you might imagine.

Guess you are lucky to have a high percentage but from what I’ve seen on here you are the exception rather than the rule.
I really do appreciate all you explanations and caveats on how Sense works. Always a good thing to be informed. But what I really like is Results and so far they seem to be elusive.

As for smart homes. Don’t get me started. Maybe, maybe Thread and Matter will solve many of the issues in that area and all the stop gap/add on products won’t be needed.

Last your comment about the Kasa plug. Are you sure you were looking at the EP25P4’s and not the HS103P4’s . I was looking at them yesterday on Amazon and the EP25P4 was $49.99 and the HS103P4 was $26.99. I believe the HS103P4 does not work with Sense

1 Like

Good question on Kasa model. Pretty sure it was the EP25P4 - I remember seeing the deal coming in half the price of the usual. But if not, you’re right -$400 for 30.

FYI. Update

So I did delete the space Heater and I put the coffee pot on a smart plug to monitor it. So far the coffee pot has not been natively detected.
I did find out one way it differs from the Space heater is as Coffee is being made it draws roughly the same wattage But once its made it wattage drops to around 50w for however long you’ve set it to maintain hot coffee. The space heater is full On or Off depending on its thermostat setting. So they do differ in a pretty big way.
Another thing I found out is after deleting the space heater I had a couple cold spells and plugged it back in. Of course the device is no longer in my list but it will show as previously named with a bubble. This happens 50/50. Sometimes the old bubble, sometimes in the other bubble? Even when the old bubble appears it does not show as turning on or off in the list on the left side of the main screen. Basically the system is confused I suppose.

Interesting - Sense is likely to see your coffee maker as two separate devices and detection, given your description. Sense only sees the on and off transitions so a 700W coffee making session looks like a different device from the warming 50W cycle.

Most likely, Yes The coffee making cycle is closer to 900-1000W which is very close to the Space heaters low setting. Not 100% sure but before I deleted the space heater device and the coffee pot would show up in the space heater bubble it was just during that coffee making part.
Doubtful since it went months before and no coffee maker detection but maybe this time it will find it.

1 Like

Update. I have kept an eye on the coffee pot dilemma and found Sense does see it as One device however only partially reports what it sees.

So in conclusion the bubble, device timeline and power meter don’t all read the same.