What's new in v23 and Web App v5.5: Smarts Plugs and Themes

Does the Sense adjust the “Always On” data to take the new smart plugs into account after I have enabled them? Or does it take time for that to happen?

Meaning: If I had something that was categorized as “Always on” in the past, but now it is on a smart-plug, will I see that power usage be accounted for twice via both the smart plug and the always on category, or should that not happen?

It takes a bit of time, but yes, Sense will integrate those smart plugs into Always On. See this thread for the latest updates on that: What's new in v26 and Web App v7: Always On Stats - #36 by jlj.pers

That link helped, but it didn’t quite answer my question: I’ve attached an example screenshot of my Sense monitor. In the screenshot, the “Servers” bubble is a smart plug (tplink).


So, if you add up the watts of all the bubbles (minus the Solar, of course), the total is higher than the the amount in the left-hand side of the screen. Is there a way to disaggregate the “Servers” smart plug watts and remove it from the “Always on” watts, or do I just have to wait a few days and let Sense do that for me?

I’m guessing you just turned the Tp-Link integration on, based on the News item? If so, yeah, you’ll need to give Sense a bit of time to pull in that data.

But your total can be higher/lower for other reasons, mainly due to Always On being based on a 24-48h lookback window. Thus, even if you managed to turn off every single Always On device in your home at this very second, your Always On would still continue to show 353W. The ‘Now’ wattage is a live reading, and does not include the Always On value in its calculation.

Very helpful! Thank you =).

Hi RyanAtSense. When can we expect Sense to use the ground truth of TP-Link data to result in native Sense detection. I have 10 so I’m less interested in moving them around to save $ and more interested to know when my additional $150 investment is going to result in improvement in the Sense device detection algorithm.

I should add that the HS110s have been functioning great for ~6 months. No issues.

That’s a good question. I’m guessing that the beneficial results of smart-plug “ground truth” should show up in the Sense Data Science updates that we see periodically in the forum. Given the indirect relationship between “ground truth” data and detection, we might not see causal specific resulting features:

  • ground truth improves training
  • improved training gives more accurate detection under some circumstances, but those circumstances might be fuzzy or unknown.

The other possibility is that smart-plug “ground truth” enables detection of some new class of devices, possibly things like some types of power supplies.

But given the wacky variety of things I have plugged into my HS110s and HS300s, I’m not counting on breakthroughs for most of those devices.

I originally thought it was going to be a way to move the HS110 around to all your devices and it would figure out the power that the device uses and identify it, then you move the HS110 to another device. Now I just put it on any device that I cannot pick up otherwise, like my Plasma and my TiVo. But when I first did detection I didn’t know what it couldn’t pick up. Now that I am on round two I haven’t even turned on the HS110s yet.