First an apology if I’m posting in the wrong place.
As a new user, on my 3rd day, I’m wondering what the typical learning curve is. So far all it has found is the fridge. Is this typical and am I just being impatient? i.e. When shouldI expect it to have a more complete profile?
Welcome @steven.richardson
Yes, Patience is key. Some homes have very slow detection while others it can happen quickly. I have an all electric 1600ft home in Kentucky. I’m seven months in and have around 43 detections with 14 being smart plugs. Almost all from the first three months or so. The detection rate varies so widely that I don’t think there is a “normal”. There are too many variables involved.
A couple of us have posted detection rate charts that include both normal detections and smartplugs. My advice - it’s probably better to look at detections on a weekly level as a measure of progress than day by day. And as @samwooly1 suggests, each house can be different.
1 Like
Another thing I’d say is after 3 days and already detecting your fridge is a good sign to me. Not everyone has a detection that fast.
You’ll likely have days like I have had where there have been as many as four in a single day. That is really exciting. Then, the detections can seem to stop and it may be a week or longer. Try not to get discouraged when this happens. I think everyone here has experienced ups and downs. It’s best not to have any expectations or goal for detections and to just ride it out.
So it can take awhile and I need to be patient.
Are there any tricks one can do to help the discovery?
General learning/detection tips:
- Repetition is important - machine learning “likes” to see many, many repetitions of an on and off cycle, under many different power usage conditions in the house.
- On/offs induced by humans may or may not be helpful - not helpful if the on/off isn’t the same as a typical cycle. So flicking power on/off at the breaker box doesn’t contribute, nor does turning fridges on/off via a switch.
- “Noisy” devices in your house can mask detection of other devices - Especially if the noisy device runs frequently. The noisy device won’t be detected, plus it might delay or defeat detection of others that are masked by its random signature. Noisy devices are ones with significant power usage (hundreds of watts) and fairly random signatures. A good example is a plasma TV that is frequently on. The power usage is large and varies with the picture on the screen (no repeating patterns).
- Most “electronics” are going to be hard to detect. Any device that has an AC to DC power supply connected to directly to wall outlet is going to mask it’s true identity, including things like computers, networking devices, AV equipment, some kinds of microwave ovens, and even DC/battery backup garage door openers. Consider using Sense-supported smartplugs for any devices in this category that you really want Sense to do the accounting on.
- Many “devices” have several components - HVAC is a good example. My system has a couple of detected components, the furnace blower and the AC compressor, plus several undetected ones that are visible only through a smartplug (the pre-ignition blower and igniter).
2 Likes
I am also on Day 3. It took 36 hours to do the basic test. After 24 or so hours it gave me a 50% complete and the info that there is a problem and Sense will try to figure it out automatically. I guess it did. Any way to find out what that problem was? I have a central air that goes on often given it is summer. According to Kevin1 I guess this might affect my detection since it draws a bunch of power and might mask or create noise. Anyway only one weird device detected that I can’t figure out what it is. Also it did find two Hue lights. With NDI will individual computers and devices on my network be reliably identified?
NDI only works on a select set of newer networked TVs, as far as I have seen on the forum.
AC offer a relatively predictable and stable pattern (Sense seems to identify traditional AC compressors OK, though not mini-splits). If Sense can detect it, it’s not noisy.
I’m curious to know if anybody here has seen a new device that clearly wasn’t theirs. Yesterday Sense found a new device and named it heat 1.
In looking at the usage I saw this:
On June 4th for example my total usage measured by Sense was 5.526 kWh. I x-checked this with my energy provider’s (PECO) measurement which is 6 kWh.
Heat 1 shows usage of 21.991 kWh for that day.
I had this happen before where Sense told me that my Keurig was using 16 kWh… (I don’t have a Keurig…). So I’m wondering if I see another user’s data in my app.
Something doesn’t sound right in the way you have explained your issue.
If Heat 1 shows a daily usage of 21.991 kWh and Sense shows a total usage for the same day that is lower than that then your Sense is misbehaving and you should ask Support to take a look.
Meanwhile, a separate issue is that if day-to-day you have consistently-ish:
100 - (100 x (5.5/6)) = ~8% under-reading by Sense vs the Utility then you have a problem. Normal margin of error should be less than 1%.
Possible reasons:
- UNLIKELY (though maybe LIKELY based on your other issue): Bad Sense/calibration (contact Support)
- UNLIKELY: Bad Utility numbers (check consistency)
- POSSIBLE: Your Sense CTs (clamps) are, you believe, directly between your metered power (meter) and your panel but somewhere in-between there is a power tap = something using power that will not be clocked by Sense so Sense will consistently under-read your power usage. The only guaranteed way to check if that is the case would be to switch off your main panel breaker (if you have one) or all the individual breakers (to get Sense to read 0W in the Power Meter) and see if your meter is still turning over. If it is, you have a tap off the main feed. Sub-panel?
Thanks for the reply. My utilities provider usually matches to 99%. However, they “round up”. So if sense gives me an hour @ 0.58 kWh, Peco shows 1 kWh. Usually it’s to their advantage… 
Nothing I can do about it.
However, this is the second time I see a “new device” that totally doesn’t belong to me. Also, I live a rather minimalistic live. I just got my monthly bill for June and the heat hasn’t been turned on all month. My total usage was 534 kWh of which, based on previous months usage 230 -260 were AC. No sub panels, meter is basically brand new, all electric updated and I’ve had the Sense device for almost 2 years now. Mu bill is totally fine, it’s just that I think that someone else’s data is showing.
Yeah, something definitely does not sound right here. I don’t see how you’d be seeing someone else’s data, but those numbers do sound way off. Have you reached out to Support? That’d be my suggestion here.
I know of a person that had a breaker for a future irrigation pump. They never installed the irrigation pump but later found that their neighbor tapped into the pre-installed wiring and used as power source. I have also heard of people using someone else’s external power plug to connect an extension cord.
My problem is that I would like to use Sense to track A/C cooling cost. Problem is that sometimes the 5 KW A/C gets classified as Other. Also, noticed that my refrig that is on about 22 hours a day is sometimes getting classified as Other. The refrig doesn’t bother me because it has a pretty constant on/off rate so I can calculate its usage & usage is relatively low (320 Watts or 7 KWh).
But the A/C is a bummer as it is the biggest user of power in the house and can’t be tracked. I can never know how much my cooling cost is for the month!!!
1 Like
I’ve been spot checking how well Sense spots my AC. It seems to get it somewhere between 75-90% of the time for the past week. At least 2 of the times it missed, another major appliance turned on/off within a few seconds of the AC turning on, so I’m not sure if that had an impact. I’m hoping it gets more accurate over time, as I’ve had it a bit under a month and only 6 devices have been detected so far.
Edit: after posting this, I opened up my Sense. The AC is currently running in my house, but the energy is going under Other. Maybe it will get re-classified when it shuts off and Sense recognizes the shut-off pattern?
Edit 2: It did not re-classify it toward the AC. Sad times
2 Likes
Here’s a few months worth of analysis of AC detection at my house vs. commands coming out of my thermostats.
1 Like
Yes, sad times. The A/C is the biggest power consumer and the one that I cannot track its operating times. I was hoping to do keep A/C at 95 all day then cool to 80 when I get home versus keep A/C at 90 then at 85 to see which uses the least energy. BUT that not possible with the Sense as so much goes into the other category. Seriously, considering sending it back at the end of the 60 days since Sense doesn’t do good job of allowing energy conservation thru experimentation.
This topic was automatically closed 365 days after the last reply. New replies are no longer allowed.