ben wrote: ↑Mon Aug 16, 2021 12:37 pm
Yeah, I would assume whatever the hot water heater reads should be accurate if it's a built-in display. Can you post the model? Maybe they have a minimum accuracy value.
It's a Rheem Professional Prestige ProTerra model # PROPH80 T2 RH375-30. They don't provide a spec for power measurement accuracy.
ben wrote: ↑Mon Aug 16, 2021 12:37 pm
The GEM should pick up those spikes, you can try looking at Peak Watts to see. Peak Watts is the highest value between packets (based on your send interval) logged on the minute.
No spikes in the Peak Watts display. In fact, we were on vacation from 7/30-8/5, and I set the HPHWH to Vacation Mode, which I believe holds the water at 65F. The GEM reported no power whatsoever going to the HPHWH during that period, while the HPHWH showed regular spikes of about .2kW.
ben wrote: ↑Mon Aug 16, 2021 12:37 pm
The combination of 2 hour bars on the DashBox seem to reflect the one spiked bar on the water heater display pretty closely.
Yes, I had noticed that. That's what makes me think the .2kW spikes are where the difference in daily consumption comes from. It's not clear whether each bar represents an hour or what hour it's supposed to be.
ben wrote: ↑Mon Aug 16, 2021 12:37 pm
Are you 100% sure on the wiring? It's possible the CT polarities are cancelling from each other. You can remove one set of CT leads and see if the value jumps up.
The short answer is: removing one set of wires halves the power to about 225W when the HPHWH is running at max.
(When I say max, I mean compressor and fan on. The real max occurs when the compressor and fan are running and the resistive element is on. That's more like 4kW.)
I don't see how incorrect wiring could result in the GEM showing 0W for the .2kW spikes shown by the HPHWH, yet show about the same consumption when the HPHWH is running at max power (albeit with the hours not quite lining up.)
But I'm never 100% sure about anything, so I'll review what I did. Initially, I had a pair of Split 60s on L1 and L2 in the subpanel until the micro 40s I ordered arrived. That included the HPHWH and the well pump, but the well pump doesn't go on very often so most of the time the readings were just from the HPHWH when it was running. I always orient all CTs in the same direction: K facing source and L facing load for Type A split CTs and the wire side facing the load for Type B micro 40s. When using the Split 60s, I put one pair of wires on ports 1 and 2, and the other pair on ports 3 and 4. I switched the leads of one pair back and forth to determine which wiring gave the max power. Typically, it's about 450W (450 kWH per hour) when the HPHWH is running, even though the HPHWH might show a peak of 900 kWh at about the same time period. I couldn't find any combination of wiring that resulted in a higher reading. When I got the micro 40s, I put one one each lead to/from the well pump and hot water heater. I combined the black wire from one CT and the white wire from the other CT in port 2 and combined the remaining black and white wires in port 3. (And yes, I changed the CT type in the GEM as appropriate for the CT type being used.)
To make sure that there aren't significant losses in the ~30-foot twisted pair cable between the main and subpanel, I used a clamp ammeter on each lead between the subpanel breaker and the HPHWH while it was running, and got a reading of about 1.8A on each leg (it's an analog clamp meter on the 6A scale, so I had to interpolate between marks.) If I multiply 1.8A times 240V, I get 432W, which is about what the GEM says.
I should note that both the well pump and HPHWH are balanced loads, so I could have used just one CT and doubled the reading. I used to do it that way on the well pump when it was in the main panel because I ran out of micro40s, but now I have a bunch of micro 40s and decided to use two on each. In case you're wondering, I checked the wiring diagram for the HPHWH and it only has connections for L1, L2 and GND. No Neutral.
ben wrote: ↑Mon Aug 16, 2021 12:37 pm
Under the Advanced section if you choose Line Graph you should be able to do Watts and VA on the same graph.
Ah. That's cool. Still no indication of the spikes. In the part of the graph where the HPHWH is shown drawing power, Watts looks to be about 95%+ if the VA value. That corresponds to the calculated Power Factor and the product of amps and volts in the GEM Live Data screen, which came out to about 97% of the watts value. So, I don't think power factor is the issue here.
I think the fact that the spiked bars in the Dashbox sum to about the value of the spiked bar in the HPHWH display, and the difference between the daily consumption readings is pretty close to the sum of the .2kW mystery spikes, suggests that the HPHWH thinks those spikes are real and the GEM doesn't. Which one is right?
I've tried to figure out whether the HPHWH is really consuming as much as it says by comparing my All Circuits group with Total Consumption and the figure for House consumption shown by the Tesla Gateway. However, before we got the HPHWH, and as I reported in another thread, there has been a significant discrepancy between these three values. Total Consumption generally ran about 1 kWh or so less than House consumption shown by the Tesla Gateway, and All Circuits generally ran about 1kWh less than Total Consumption. I still haven't gotten to the bottom of those discrepancies. So far, I've shortened all CT leads to the minimum required, eliminated noise on the subpanel CT leads, cleaned the PT plug and verified that the PT is putting out the correct voltage within .1-.2V of what my true RMS DMM says. Next step is to put some known loads on various circuits. I'm still unclear on whether CTs can be "tweaked". Changing Range doubles or halves, so that's way too much for tweaking. Does changing the CT type by small increments do that?
[EDIT: Possible explanation for the discrepancy: See my last post in
this thread.]