Checking accuracy

I’m trying to compare my IoTaWatt to my utility bill and getting frustrated. There is a big gap on imported power – off by 30-50%, but exported power is fairly accurate at 5-7% off.

I had previously had concerns about power factor accuracy and @overeasy suggested there might be an issue with tape on the CT. I haven’t been able to get into the main panel (to congested to safely do myself, need electrician), but as I look at the data, power factor only seems oddly low on import-- export is at unity power factor.

Is this an issue of just tracking voltage reference on a single side? Is there a way to use two voltage transformers for split-phase service?

One more bit of data-- I am using the 5-second Watt values exported to Influxdb and integrate for energy there. I understand the limitations of doing the integration for a month directly with Graph+.

And a further edit… looking at the data I don’t think the problem is with power factor; at worst it seems consistent or predictable. Average load between the two HECO utility phases seems reasonably balanced at about 4.5A on one and 4.0A on the other average during import periods.

If you are trying to measure import and export you should setup an integrator. See the docs for an explanation why it’s needed and how to do it.

There is probably a way to query it from influx but I haven’t looked into it. Ordinary integral query in influx will only yield net energy.

I currently send to Influx an “Export” and “Import” Watt value calculated as such, rather than using an integrator:


(HECO1 + HECO2 - HECO1 abs - HECO2 abs ) ÷ -2


(HECO1 + HECO2 + HECO1 abs + HECO2 abs ) ÷ 2

I’ll look up how the integrators work, but it seems to be the same operation.

Either way, my net energy still is not reflected accurately when I just integrate HECO1 + HECO2.

Any other ideas?

Some further analysis. “HECO” is the sum of all loads on the panel (6 total), and HECO1 and 2 are the direct phase measurements.

Accuracy of the sum of parts vs direct measurements seems good with up to ~3kW load, but gets worse with higher loads (in export). On import, there seems to be a fairly constant offset of about 250W.

I would need to have the whole context to figure out what s going on here. To start:

A picture of the panel with CTs.
The inputs setup display.
The outputs setup display.
A screenshot of the status display. You imply that there is solar by mentioning export, so a display with solar producing and without.

I don’t have panel pictures, but I did some detailed checking of the PV measurements against Enphase’s system. Taking out the power consumption for the Envoy box, I saw a 7% delta there as well. My guess is that I have voltage imbalance, as this also relies on two CTs for the total (since they carried a neutral to the Envoy box).

Investigating a little further, it turns out one of my three IoTaWatt’s is on a different phase, and I was able to graph the voltage difference, which varies widely between +6V and -4V. The average though is closer to 2V/2% though, so that alone would not explain the 7% delta.

Here are the configuration images, along with current status and a graph of amps at the same time.

It will be a month or so though before I can get panel interior photos.

Your HECO01 is the clamp type CT, and judging by the low reading and PF, I suspect is either not closed or still has the protective paper between the iron cores. It needs to be inspected, checked for paper between the surfaces,nand reinstalled.

You should also look into using an integrator to track import and export.

Below is the status with PV producing. I will check the main CT when I can get the electrician out to the house.

Have both checked and reinstalled. Without knowing anything about your service, I can’t determine anything definitive about the mains except that together they don’t appear to provide enough power to support the loads that are measured.

FWIW, the incoming utility is 200A, 120/240V via a service transformer that is shared with about 20 households. Utility service lateral is #1/0 aluminum IIRC, which would be normal utility procedure. The panel has no main; it is a 6-disconnect rule panel with (2) 100A sub-panels, (3) split air conditioners, and a 40A PV breaker supporting a 7kW microinverter system. House was built in 2008, PV installed in 2014. Annual energy consumption is around 10MWh.

I had called attention to problems here before with odd power factor readings last year, and it turned out that one of the breakers had some hot connections. Those were torqued down and the readings are generally consistent with my expectations within the system, with the exception being that I have more “unmetered” loads on my sub-panel branch breakers than I would have expected-- it is about 30-40% of my low-state loads (late night, lights off, etc). During those periods the sub-panel reports about 500W of loads (from the main panel), but my metered branches only total ~300W. (The unmetered loads should be near-zero during these times.)

I’ll have the CTs checked once I can get someone out to help. When I get home I can also now check the integrator values against the utility meter for import/export/net over a defined period.