Power Quality - 824x (dips) last 30 days

How was your issue detected?: Flickering lights and stove/microwave clocks resetting

Did you noticed any other signs of this issue around your home?: No

Screenshots from Sense Lab

Raw Power Quality CSV

We noticed flickering lights and clocks resetting on June 14th. We have not changed anything in terms of electrical appliances running in about a 2 months (we started using our central AC 2 months ago). Neighbors said they might have noticed flickering lights but my neighbor is 90 years old so I am taking that with a grain of salt. The only thing I could think of that might be causing issues is my Unifi G4 Doorbell Camera. It is hooked up to a 16V 30VA transformer (https://www.supplyhouse.com/NuTone-C907-16V-Transformer-30-VA). I turned off the breaker for this just in case about 2 hours ago and haven’t noticed anything awry since but it has only been 2 hours. Panel looks good, no sounds or smells.

I called the power company (PSEG) and they will send someone out in the next 10 days, but most likely within 2-3 days. Hoping someone can shed some light onto this. Thank you!

1 Like

@JustinAtSense I know in previous posts, you were able to check with the Data Science team to see if there was anything interesting with the data. I am sure everyone is super busy but would definitely appreciate it if you can check with the Data Science team. Thank you in advance!!

1 Like

Hey there - we used to have the Data Science team manually review a user’s Power Quality if we suspected a floating neutral (which have pretty distinctive waveforms and can be a serious fire hazard.) Since then, we’ve released a lab project in Sense Labs that does this on a reoccurring basis, just because it was quite unsustainable to run manual checks with a small team (you can find the Labs project here.)

We’re very limited in what we can “see” as the cause with Power Quality issues outside of Floating Neutrals, which is why the best feedback/suggestions users have received in the past are typically from other users in the community with experience dealing with these sorts of things. I’d definitely recommend waiting to see if you get any feedback from users over the next few days before I check with our Data Science team.

1 Like

Here’s another way of looking at all your power quality problematic datapoints, comparing leg0 to leg1. Red lines are what I think Sense uses as the instantaneous limits for error reporting - 110V and 130V. Looks like all but 4 of your 824 quality spikes were on leg1. The one issue inside the the 110V/130V box is probably an sustained overage over 126V. Both legs look more spiky than my mains, but leg1 seems to get pulled down much farther, so you might have some kind of corrosion or issue with your leg1 feed.

1 Like

Thanks for the very good analysis and explanation! @kevin1, you were spot on! PSEG (power company) came out after I called again because half my house lost power. Turned out to be a faulty connection at the pole. The lineman said the box on the side of the house was perfect but connection at the pole was bad. He said whoever did that before did not strip enough of the wire. While he was replacing it, he said it sparked on both legs. So he replaced the bolts and connection on both legs and said that was the cause of all the issues.

Pretty awesome that Sense detected that there was an issue. However, it would have been great to get notified when the power quality anomalies hit a certain threshold. I could have called PSEG sooner if I had known :slight_smile:.

@JustinAtSense Let me know if the Sense team would like some free, pro bono help to implement this. I’m a full stack software engineer and iOS/Android developer. Would love to help.


Just for fun, I charted 2 min samples of my L1 and L2 voltages over the past month or so, using the same axes, with the tighter 114V-126V criteria. Looks a little different ! Hope yours will as well.

The same rough view with coloring by time of day. There’s a little bit of a pattern.

Agreed re a notification of these spikes /dips…

1 Like