Unusual usage reading

I have a weird usage reading as of today.
Nobody has been even remotely to the service panel, nothing was opened or even a breaker flipped.
While looking at today’s production I see weird usage numbers.
The usage suddenly seems to go to 3600+ watts. I know what my users are and certainly nothing is drawing this much and it ends up in the “other” classification.
First I switched off all solar ( i have both micro inverters & string inverter) and looked at the meter and compared with sense: They were in sync and both indicated about 200 watt was used.

Switched Solar back on in 2 steps, first the micro inverters, 1 minute later the string inverter.
After the 5 minute mandatory delay micro inverters came back online and usage still the same.
One minute later the string inverter came online and right away the virtual usage started appearing again.

These were made seconds after each other:
Screenshot Sense:


Subtracting one from the other: 11342-3278=8064 watt into the grid.
photo of meter reading:

10200 watt into the grid.
2000 watt (virtual) difference

I even powercycled the Sense to rule out some weird state, but came back up with the virtual load reading:


Reading from my HAN device from my meter direct for the last hour

Anyone any idea where to start debugging this?
Only thing I can think of is the main CT’s are opened, but they are at the top of service panel and the solar CT’s are at the bottom. doesn’t like they could interfere with each other.
Afaik we did not have a quake that might have moved the CT('s) :wink:

Just as abrupt as it appeared, the usage no longer shows up


The usage is now constantly under 600 watt (apart from EV charging) no more ramp up to 3+kW.
I will keep an eye open for this.

I came here precisely because I too had a very similar incident today - usage readings suddenly went negative while solar had a surprisingly low production day despite high sun and cool temperatures.

It’s almost as though the power calculations were off by 2 kW or so for much of the day.

Things appear to be back to normal basically as of the time I opened my mobile app.

2 Likes

And it is back!

The script that was posted here to read the Sense unit:

Date/Time: 2022/05/29 14:38:15 Power Being Used: 2924 watts, Power Being Generated: 11298 watts, net production 8374 watts
Date/Time: 2022/05/29 14:39:20 Power Being Used: 2823 watts, Power Being Generated: 11291 watts, net production 8468 watts
Date/Time: 2022/05/29 14:40:25 Power Being Used: 2863 watts, Power Being Generated: 11237 watts, net production 8374 watts
Date/Time: 2022/05/29 14:41:30 Power Being Used: 3039 watts, Power Being Generated: 11256 watts, net production 8217 watts

Reading what my meter shows through the zigbee connected HAN device:

{“demand”: -10.011, “raw_demand”: -10011, “multiplier”: 1, “divisor”: 1000, “timestamp”: “2022-05-29T03:43:09Z”}
{“demand”: -10.032, “raw_demand”: -10032, “multiplier”: 1, “divisor”: 1000, “timestamp”: “2022-05-29T03:43:24Z”}

It is seeing -10.000 watt
-10.000 watt meter
sense reporting -8400 watt
Sense is making up a 1500-1800 watt load.

Graph from Sense:

Graph made by wattvision with data taken from HAN device (aka meter reading) for the last hour:

So my conclusion is that Sense is making up some phantom load that is not there IRL.

I will open support ticket and update once sense technicians respond after the holiday.

Yesterday it was clear to see that the (phantom) load was somehow connected to the solar production.

When the solar production got lower, the load also got lower.
When we had some (shadow?) event, you could see that in usage as well (blue circles)

Today however, everything is back to “normal”.


All usage accounted for, no weird stuff (yet)

I did not powercyle the sense unit, or even touched the breaker box.
Imo this is a software issue, not a hardware.

Update
I compared sense data with my utility company data.
Saturday (when I did not have an issue)
Sense:
to grid: 46.1 kWh
from grid: 10.2 kWh
Utility company:
to grid: 46.29
from grid: 10.13
to grid: 3% difference
from grid: 1% difference
That is close enough imo

Now Sunday when I had the phantom load:
Sense:
to grid: 50.3 kWh
from grid: 8.1 kWh
Utility company:
to grid: 59.4 kWh
from grid: 8.97 kWh
to grid: 18% difference
from grid: 10% difference

1 Like

I just got a response from Sense:

After reviewing your thread in the community and the data from your monitor, it appears that the issue you’re reporting may be stemming from micro-corrosion on the contacts of your main sensor cables. It looks like you caught this in an early stage.

To address this problem, we have created a short extension cable that includes slightly larger pins and a bit of protective grease. We’ve found that in most cases, simply inserting this cable between the monitor and sensor corrects the problem and prevents future connection issues.

Please Click Here and we will send you the fix kit ASAP.

Will install when it arrives and give an update.
Good to read it was caught in early stage and that they have a solution already.

1 Like

Out of curiosity, do you know why in the every few mins we have that line every few mins in the solar line… mostly noticeable in the initial climb in the mornings. On the meter it looks like its dropping 250w for a second every 3 mins .

1 Like

Screenshot from 2022-06-01 09-39-50
Screenshot from 2022-06-01 09-40-05
579 watt difference
Screenshot from 2022-06-01 09-40-32
5 minutes apart

I have a real weird setup:
Enphase micro inverters (DL380,M190,M215,M250)
Sunnyboy 2500HF string inverter
Sunnyboy SB7700-21-US string inverter
Fronius 3800 string inverter.
I have monitoring on the fronius & enphase but not in the SB’s.
I have to assume one of them is re-calibrating the MPPT at that point.
But it is weird

I got a sunnyboy 5.0-1sp-us
Just 1 inverter. I recently added 10 modules / 2 strings.
Even when I was just running 1 string , I still got these marks.

1 Like

Parts arrived
4" extension cables

2 Likes

Did this fix your phantom load?

Hope to install it over the weekend.
So far the phenomena hasn’t shown itself again