Seems like you are running at a pretty high detected device ratio - that’s a place where the double counting of Always On between the main Always On and smartplugs slice could help displace Other.
The lag time might be a result of progressive detection. As I understand it, there is a tradeoff between broader detection capabilities and lag in displaying on/off. In other words, one can detect more if Sense can looks at a little wider time window around transitions.
There’s really not a way to adjust the Always On bubble value to remove the Always On components picked up by the smart plugs. The accounting in the Always On device is mainly focused on figuring out how much of the Always On bubble is unknown.
If a smart plug has a big dynamic component and only a small Always On component, then you should just keep using it as is.
If a smart plug has mostly an Always On component (i.e. my switch that uses 38-39W continuously), then it might be worthwhile removing it from the smart plug and simply entering the Always On component in the Always On list.
The simplest way I can think of to pick up leakage current with Sense would be to shut off all your breakers except for known dual-phase (240V) perfectly balanced loads running. Anything with a big 240V motor, for example, should have balanced usage of the phases.
You could put a Flex CT or a clamp meter around BOTH L1 and N and again around L2 and N and you should get 0amps.
Real electricians, here or in the real world, will probably have better ideas.
That said, it’s unlikely I think that you are experiencing such a dramatic “leakage”.
We got a little distracted from the scale of the discrepancy.
I lean toward what is usually the case and @kevin1 suggests above:
Another way to sort that out, beyond tracing wires, is to shut all the breakers off that you have access to (except the one that Sense is on) and see if you get to 5 watts or so (what Sense uses). You can of course do some math and leave on certain devices that you need to run. The key of course is to do that when you have the suspiciously high usage.
You can also do math on the TOTAL POTENTIAL usage of your house. Add up all the wattages. Look through the peak usage in Sense. Did it ever go above that?
Ah, and if you have patience you can watch your meter in realtime along with Sense. Only practical when you have a HIGH load running.
Currently the support team has me adjusting my wireless router to achieve a better signal strength. It looks to be helping but I think there’s still a summary reporting error in the app. I need a bit more time of comparison to be sure.
I read my smart meter every night at 00:05 and put the info in a spreadsheet.
I missed a couple of days in october 2021 so when I compare complete months after that:
From nov 2021 - feb 2022 I got 4157 kWh from the grid and delivered 3250 kWh to the grid.
Same period according to sense: 4160 kWh from the grid and 3231.5 kWh to the grid.
From the grid: 99.93% same
To the grid: 99.43% same.
Good enough for me
@croerig, hmmm. So far, every case of poor accuracy vs utility meter, where the user has posted the details on this forum and actually investigated, has turned up as a measurement problem, from a missed sub panel, to a crosswired condo (meter for one unit is assigned to the opposite one). Can you share more specifics of power usage you are seeing in Sense vs your utility meter ?
Here are some good pointers on finding the source of the differences:
Hi, @croerig, I’m sorry you’re having accuracy issues. There could be a few different reasons for this, as @kevin1 pointed out. Can you reach out to our support team with details on what’s happening? email to email@example.com. And keep us posted on what you learn.
I am a new two week user on Sense. I am very satisfied on the usage rate reported by Sense compared to the local SC utility. For seven days I read the utility meter at midnight and compared with Sense. They agreed spot on as others have said. The Sense meter is in the main panel. Most of the breakers are in the sub panel 40’ away.
Now Sense has found the first stage of my two stage AC unit. However the wattage shown on the AC power meter is the same day and night and is about 1785 watts. In reality the usage at night when the outside temperature is about 1600 to 1650 watts and during the heat of the day it can reach 1850 watts. This is in agreement with the AC design. Even though I like the ability of Sense to find the AC unit I wish Sense could display the actual wattage on the AC power meter. I am going to use the dedicated flex line in the main panel to monitor the stove. I hope in the future that Sense will provide some local current transformers for large 240 volt units. I use 12 KASA TP125 plugs. They work great with Sense. Overall Sense has made a great product.
@joeengr, welcome to the world of Sense. Glad you are doing the experiments to double check your results and build confidence.
On the AC, you’re running into one of the challenges associated with the main Sense detection mechanism which is based on “immediate detection” - that is detecting based on an on and off transition that are only 1/2 to 1 second long, then doing an immediate display of the on or off in the bubbles. This immediate detection model can over/underestiamte devices like your AC that have some degree of variability to their final operating power consumption.