Sense or Meter Low Accuracy

Hello All!

I read through @kevin1’s awesome thread on accuracy and how the CT’s should be positioned for best accuracy. I have a 300 amp service split between two 150 amp panels. The first one is the panel Sense is installed and the second is the middle port of the sense going to the other panel

I’ve set them up where they are perpendicular to the wire and checked they are fully closed to try to ensure accuracy as per @kevin1’s post.

So here’s my deal though…My power company is saying I’m using way more power than my sense is. At the last billing my power usage was up to 30903 kWh

So my new billing period will be from Feb 11 to Mar 10th. At this time sense see my total usage as 516.17 kWh for the billing period.

However my meter is telling me that it’s actually around 726 kWh.

It’s off by around 28% . So here comes my obvious question? How did I install the sense wrong? Mind you I do actually suspect my meter is wrong because power bills are insane with summer of around 3000 kWh supposedly.

Any ideas?

Here is the sense total

I can also get a table of it if anyone wants it.

Do you have realtime access to the Smart Meter readings? Most utilities with smart meters are enabling that these days AFAIK.

With rt access (or even just checking manually) you can compare your daily totals between the meter and Sense.

@ixu sort of. So Georgia Power enabled hour by hour breakdowns last November but it’s not actually an hour by hour breakdown or daily breakdown. Here’s what they do…they take the power usage over the hour and they average it. So say I’m using 600 watts for an hour, they will represent that to be 1 kWh. Say I’m using 1.5 kWh in an hour, that’ll be 2 kWh on the site. They round the first 1-999 watts to 1 kWh and the rest the standard way. They then take the 24 hours of rounding and make it the daily total for each day. So it’s quite inaccurate compared to just reading the meter. Here’s an example of this month so far.

As you can see this morning it stays the monthly power usage so far is around 664 kWh through 28/02/2022 while the picture of the meter states around 726 only halfway through 28/02/2022. Only one month has the rounding coming in higher than the actual meter.

Here’s what the hourly looks like:

and here is the sense of the same day

So the Georgia power rounding says around 38 kWh were used on the 27th while sense states around 28.5 kWh.

Take periodic photos of the meter I suppose and correlate with Sense usage.

Noon-to-noon might be more convenient than midnight-to-midnight, but you get the idea.
Even on a daily basis with actual (according to the meter) usage around 30-40kWh @ 30% discrepancy you should pick up the anomaly in “ground truth” (vs ultlity-transmitted) data every day when comparing to Sense.

OK, so assuming you do have an anomaly and we assume the Sense CTs are accurate to within 1% or so (normal) then the next step is verifying that your CTs are placed in the right location relative to what you are metered and that the meter is reflecting your actual usage. Can you send pics of the meter relative to the panel/s? Is the cable routing obvious?

You are configured like this?

Are there any potential wire taps between the meter and Sense that are using power? EV outlet, that sort of thing (This is perhaps the most likely scenario)

1 Like

As an aside here, can somebody enlighten me regarding the look of those Flex sensors above? Are they a different build?

@WizardofCoz, thanks for all the details. I would agree with the checks that @ixu is suggesting. It does look like you have installed correctly as long as there aren’t any other taps or splits between the revenue meter and those mains in the top of the breaker box.

The other thing I would do in your shoes is try the export function which seems to be available to you. You might discover that it outputs the real accurate hourly (or even 15min) usage values, instead of the rounded values. If it does, I would do a Sense vs. utility scatter plot for hourly usage. If your utility does round the hourly data in export (and that rounding is annoying !), then I would do a daily Sense vs. utility scatter plot with the general methodology here.

1 Like

Screen Shot 2022-03-01 at 11.12.55 AM

“All that power” but limited signals it seems.

Ok there is no EV outlet split since the house was built in the 90s and I don’t have an EV (yet).

As to the build of the flex cables they are actually the solar flex cables because they ran out of the other flex cables according to the note they put in the packaging. So instead of making me wait while they got more flex cables they gave me solar cables instead.

I don’t get real time readings sadly and the data output is the same sadly. I figured the same as you did when I started deep diving this issue. They round… a lot.

The data for Feb 27th looks like this

Hour kWh Temp
11:00 PM 1 47
10:00 PM 2 47
9:00 PM 2 47
8:00 PM 2 48
7:00 PM 2 48
6:00 PM 3 49
5:00 PM 2 51
4:00 PM 2 52
3:00 PM 1 52
2:00 PM 2 53
1:00 PM 2 51
12:00 PM 1 51
11:00 AM 2 50
10:00 AM 2 49
9:00 AM 2 49
8:00 AM 1 49
7:00 AM 1 50
6:00 AM 1 50
5:00 AM 2 50
4:00 AM 1 48
3:00 AM 1 51
2:00 AM 1 50
1:00 AM 1 50
12:00 AM 1 51

So yes I’ll take pictures from 12 pm to 12 pm as that’s simpler. I can’t get you a picture of the breaker boxes in relation to the meter since they are on opposite sides of the same wall. Meter is outside. There is however something in the way between the panels and the meter. Two ATS(s) that do have a split for the battery charger of the generator. That battery charger has been off for the duration of this billing cycle.

The meter outside can be read for instantaneous use it has an emulated dial…I’m not actually sure how to read it but I know it has 6 dashes and it moves from left to right depending on the current usage. I also know that when I flip the breaker panel’s main switches off the line stops moving even when the battery charger for the generator is on so it’s a low power user. Phew that’s a lot of info. Any thing else I should try doing/ ideas?

Just two thoughts.

  • Figure out the instantaneous measurement coding and compare against Sense instantaneous when your usage is somewhat stable (per Sense). That might reveal a gap.
  • Export the daily usage from your meter - that should reduce rounding as a % of the total. Does the export from your utility produce a .csv file ? If so, then it should be possible to merge with Sense daily export .csv to do the scatter plot for one or more months.
1 Like

Sorry to clarify the data they give me is rounded even when exported. While the meter is capable of greater resolution Georgia power doesn’t use it. They send power data to GA power once a day. It’s not really usable outside of the general pattern. I am trying to figure out the instantaneous reading right now. It’s actually 8 dashes.

Your daily data export is rounded? That seems unlikely.

If you can do as @kevin1 suggests, a scatter plot of Sense (daily) vs Utility Meter (daily), and then let Google/Excel/R do some regression analysis (trendline and R value) you’ll get a more accurate picture of the differences.

Is there a chance that despite main breaker switch off and meter inactivity that you may have a non-Sense-metered tap that is only used periodically? Can’t think what that may be other than a water pump or perhaps a water/heater. If you do daily-type plots you may see whether the anomalies are more daily or random.

My utilities daily data is rounded, yes. Here is the output for February 27th exported.

As you can see it has rounded the hourly data for Feb 27th.

They also take that rounded hourly data and add it up to make the day’s usage on the web-tool.

So each of those days’ total is actually an hour by hour rounding as opposed to actual usage. This is the data straight out of the export tool.

Even the total kWh used tool on the website is just those numbers from the estimates added up for the billing period so far as opposed to an exact reading

It states a usage of 664 kWh which is the sum of everyday except for feb 28th. That is why feb 28th is a grey bar because the reading was not completed correctly because I flipped off the power that day to get the pictures of the sense CTs.

But yesterday the meter itself would tell you that 726 Kwh had been used which is greater than the 703 from today if Feb 28 wasn’t a grey bar.

So yes hahaha they literally don’t give me accurate data from the smart meter but wishy washy approximates even when exporting the data.

I can do this sort of if I go and take a picture outside everyday but since I have a 530x I think there are IR readers I can try and get to keep readings of the meter every few seconds depending on cost I might go for one.

If you can be bothered it would be interesting to call them and have them explain what exact information is sent:

So does that 1 second of data include hourly totals?
One would assume it includes the current meter reading or day-start & day-end.
Wonder what the bit rate is.

Yes it gives all the data once at night is what they told me and then for billing it takes the reading on the meter right as the billing cycle ends.

I believe that’s residential 400 amp service, the meter is a CL320 (320 amp rating at 80%) … 400 x0.80= 320, meaning 320 continuous but 400 amp service.

The panel that has the black and red main on the left… are those 3 party CT sensors clamps? My solar ones do not look like those.

It is indeed 400 AMP service however the panels are 150 amp panels each. As to the CT sensors they are not 3rd party but solar CT sensors. Here is the note that was in my sense packaging explaining what’s up.

They have a little sun icon on them and they don’t lock…the other pair locks closed.

Hmmm, They don’t lock shut ?? From my earlier experience with the dual panel setup and second CTs that did lock shut, I discovered that Sense came up short on usage until I rejiggered my CTs so they were fully closed. Some folk have been know to tape them shut to prevent them from opening, even the slightest bit.

As one of the folks participating in the early 400A split service, I wanted to chime in on accuracy - The 400A dual panel results look really, really close to my utility, with just one caveat ! As always, make sure your CTs are completely closed after the install. I installed my additional CTs on 5/26 and let them go for about a week, only to discover that my dual panel Sense hourly reading was consistently below my utility reading by about 2.5%, or almost 4.5% if I took out a couple EV charging datapoints. I popped into my main breaker box, rejiggered the new CTs on 6/3 and waited a while to rechart. After the CT adjustment, my dual Sense data was essentially dead-on with my utility data on an hourly basis with very little scatter.

Here’s the unity plot of Before and After the CT change. The ideal, of course, is a tall vertical bar right at 0.

And here’s the histogram of the error vs. my utility reading, both Before and After.

Moral of the story - when you install your additional CTs, make sure they are completely closed !

1 Like

They are indeed completely closed. It’s just the main ones have like a lock on them while these are more like a spring close without any kind of lock. They’re also thicker than the main sense CTs. They squeeze to open and close by the spring action.

I will have the power company come out and check the meter. We’ll see if there are any weird splits before the breaker panels. I’ll also investigate the energy use of the generator which I know is split before the panels at the ATS(s) to see if the battery charger is malfunctioning. It’s a 12v lead acid used to crank the generator.

Does your generator have a case heater?