Accuracy issues

[peppersass]

Accuracy Issue #1:

We've had a Tesla Gateway with two Powerwall 2 batteries and a solar array since October 2020. Ever since installing the system, I've noticed that its readings for total consumption by the house are 3%-6% higher than my GEM/Dashbox readings for Total Consumption (I have Split-200 CTs on the left and right Mains going into two separate GEM channels.) The difference ranges from about .5 kWh to about 1.6 kWh per day. Since the Tesla Gateway readings are very close to the monthly readings of the power company meter that appear on our bill, I assume the GEM/Dashbox readings are incorrect.

I don't believe the Split-200 CTs on the left and right Mains are the problem because I see a similar discrepancy between the Tesla Gateway readings for our solar production and the readings from the GEM's Split-200 CTs on the solar array output lines (the solar array is not connected to the main panel, so the readings are not included in Total Consumption.) Also, the Tesla Gateway readings are within 1% of the readings sent by the solar inverter to the manufacturer's website (that discrepancy is probably caused by voltage drop on the 200 ft cable run from the solar array to the Tesla Gateway.) So, two different sets of Split-200 CTs, both reading 3%-6% too low.

The only thing I can think of that would account for this is a problem with PT accuracy. I've tried calibrating the PT voltage with my Fluke 189 multimeter several times, but that hasn't brought the GEM/Dashbox in line with the Tesla Gateway and power company readings. Note that my GEM and PT are about seven years old.

So, my first question is this: Can the accuracy of the PT deteriorate over time or fluctuate randomly, requiring replacement?

Accuracy Issue #2:

Total Consumption consistently reads about 1 kWh per day higher than the sum of a group I created that includes all circuits except the Mains (I call it All Circuits.) The difference is about 4%-6%, but almost always close to 1 kWh, so I thought it might be a polarity issue with a 240V channel (in the main panel we have five 240v circuits and 35 120V circuits, all monitored by micro 40 CTs.) But the 1 kWh difference occurs every day and only one of our 240V circuits operates every day: our well pump. Until a couple of days ago, when it was moved to a subpanel, I had one CT on the pump breaker with a multiplier of two because it doesn't use the neutral. Daily pump consumption tends to run about .100-.350 kWh, so even if the reading is half what it should be, it's still doesn't account for most of the missing 1 kWh. Note that If I compare the sum of All Circuits with the Tesla Gateway readings, the difference is almost 10%!

It doesn't seem like this discrepancy could be caused by a faulty PT, which would affect the Split-200s and micro 40s equally, right?.

My second question is this: Is it possible I have one or more bad micro 40s? Do they deteriorate over time?

[Side Note: Groups don't work in the Dashbox Live tab. It says "Undefined". Is that a known issue?]

[Teken]

Accuracy Issue #1:

We've had a Tesla Gateway with two Powerwall 2 batteries and a solar array since October 2020. Ever since installing the system, I've noticed that its readings for total consumption by the house are 3%-6% higher than my GEM/Dashbox readings for Total Consumption (I have Split-200 CTs on the left and right Mains going into two separate GEM channels.) The difference ranges from about .5 kWh to about 1.6 kWh per day. Since the Tesla Gateway readings are very close to the monthly readings of the power company meter that appear on our bill, I assume the GEM/Dashbox readings are incorrect.

I don't believe the Split-200 CTs on the left and right Mains are the problem because I see a similar discrepancy between the Tesla Gateway readings for our solar production and the readings from the GEM's Split-200 CTs on the solar array output lines (the solar array is not connected to the main panel, so the readings are not included in Total Consumption.) Also, the Tesla Gateway readings are within 1% of the readings sent by the solar inverter to the manufacturer's website (that discrepancy is probably caused by voltage drop on the 200 ft cable run from the solar array to the Tesla Gateway.) So, two different sets of Split-200 CTs, both reading 3%-6% too low.

The only thing I can think of that would account for this is a problem with PT accuracy. I've tried calibrating the PT voltage with my Fluke 189 multimeter several times, but that hasn't brought the GEM/Dashbox in line with the Tesla Gateway and power company readings. Note that my GEM and PT are about seven years old.

So, my first question is this: Can the accuracy of the PT deteriorate over time or fluctuate randomly, requiring replacement?

Accuracy Issue #2:

Total Consumption consistently reads about 1 kWh per day higher than the sum of a group I created that includes all circuits except the Mains (I call it All Circuits.) The difference is about 4%-6%, but almost always close to 1 kWh, so I thought it might be a polarity issue with a 240V channel (in the main panel we have five 240v circuits and 35 120V circuits, all monitored by micro 40 CTs.) But the 1 kWh difference occurs every day and only one of our 240V circuits operates every day: our well pump. Until a couple of days ago, when it was moved to a subpanel, I had one CT on the pump breaker with a multiplier of two because it doesn't use the neutral. Daily pump consumption tends to run about .100-.350 kWh, so even if the reading is half what it should be, it's still doesn't account for most of the missing 1 kWh. Note that If I compare the sum of All Circuits with the Tesla Gateway readings, the difference is almost 10%!

It doesn't seem like this discrepancy could be caused by a faulty PT, which would affect the Split-200s and micro 40s equally, right?.

My second question is this: Is it possible I have one or more bad micro 40s? Do they deteriorate over time?

[Side Note: Groups don't work in the Dashbox Live tab. It says "Undefined". Is that a known issue?]

I’ve documented all the steps taken in my projects thread to fine tune and get the last ounce of accuracy from the GEM, PT, and CT’s.

Every component including the GEM has a variance which is stated in the manual and sales page. I’ve never seen a variance from the PT / CT due to age. I have fine tuned all settings and parameters in the PT / CT’s while using accurate & reliable Fluke RMS DMM.

The tolerance between the standard PT VS high tolerance is extremely small but is something to consider if you want that last Nth degree of accuracy.

The line voltage must be measured where the PT is inserted no where else. Fine tune the PT while monitoring over a 5-10 minute period if the line voltage varies a lot. If voltage is steady just insure the same is reflective in the GEM.

All CT’s can be adjusted but must be done using a highly accurate amp / current meter. I highly suggest you use small, medium, and high loads which are resistive and validated to be accurate and consistent as your reference.

It must have taken me a week to find all three loads which I could use that was highly reliable and consistent in its use.

You can’t use a plain incandescent light bulb say 100 watts and see 100 watt draw out of the hole. As every bulb also has a tolerance and variance which took me weeks to find!

Ultimately this was done more for my OCD vs need. As if the bulb was only 97 watts that’s fine too but it’s just easier to insert and see a highly accurate load that spans 5, 60, 75, 100 watts.

I did the same with medium / high loads by using a portable heater, heat gun, and hair dryer.

Much larger loads like my dryer and HWT I used no less than five different meters and energy data loggers to validate the readings before making CT changes.

Any long wire run will add to the variance so try to keep them short / reduced in use. Inserting a jumper in the CT channels also removes some ghost noise or improves low level power acquisition.

It should be noted do not place the PT on any UPS no matter how good or type. Doing so will cause errors and untold grief for you!

The system can be powered by a UPS but the PT Voltage monitored can’t and shouldn’t. Lastly, never change the orientation of the PT unless you intend to spend hours recalibrating.

[ben]

Is the solar connected via 240V or 120V? Do the powerwalls get used often?

So, my first question is this: Can the accuracy of the PT deteriorate over time or fluctuate randomly, requiring replacement?

You can try cleaning the contacts of the stereo jack with rubbing alcohol to see if there's any corrosion/etc. messing up your readings.

My second question is this: Is it possible I have one or more bad micro 40s? Do they deteriorate over time?

Haven't heard of any deterioration over time.

[Side Note: Groups don't work in the Dashbox Live tab. It says "Undefined". Is that a known issue?]

I'll need to look at this again, Live is a bit of a pain as you can group channels over multiple devices (so grouped packets won't be synchronized).

[peppersass]

Thanks, Teken. I’ll give calibrating the PT another try with your recommendations.

You say each CT can be calibrated. How?

Is the solar connected via 240V or 120V? Do the powerwalls get used often?

Solar is connected directly to the Tesla Gateway via 240V. Same for the Powerwalls. Neither is connected to the main panel. The Gateway is effectively a smart traffic cop that feeds power to the house from the grid, solar and/or batteries, depending on conditions. I have a pair of split-200 CTs on the solar feed before it enters the Gateway. The two CTs go into 1,2,3 and 4 on a GEM channel. I have one Split-60 on one leg of each Powerwall inside the Gateway. They go into 1,2,3 and 4 on a channel in the GEM that’s multiplied by 2. The GEM readings for solar and the Powerwalls are close enough to what the Tesla app reports that I believe I have the CTs setup correctly.

We had a previous discussion about using one CT per leg on the Powerwalls. I did this because the Gateway uses a single Split-100 CT (Telsa’s, not mine) on only one of the solar legs, leading me to believe the solar and the Powerwalls are fed into the grid via some sort of transformer arrangement so that the power on each leg is the same. I don’t know if the Gateway measures the Powerwalls with one CT or two because they’re hidden inside the guts of the Gateway. At any rate, none of this affects the discrepancy between the GEM’s Total Consumption figure and the Gateway’s House Consumption figure, neither of which is dependent on the solar or battery readings.

Yes, the batteries get used often. The batteries are discharged almost every day during the “critical peak” period, which is 3pm-8pm (highest TOU rates.) The charging algorithm is less predictable. Generally, if the capacity is less than about 70% the batteries charge from excess solar if available. Sometimes they stop at 70%, sometimes go all the way to 100%. Not sure if this has anything to do with weather (sunshine) forecasts. They only charge from the grid under two conditions: 1) A few times a month the power company dumps the entire capacity down to 20% during predicted high demand periods. In this case only, the batteries are recharged to 100% from the grid after midnight (lowest TOU rates.) 2) If a storm with high winds is coming, the batteries are charged to 100% from the grid if needed (“Storm Watch”.) With 26 kWh capacity, we should be able to go 8-12 hours or longer during an outage,

(More than you asked for:) This is not the normal way Tesla Powerwalls are programmed to charge/discharge. We’re participating in a pilot program offered by our power company under which we got the batteries at very low cost in exchange for surrendering almost all control of them to the power company. The idea is that the batteries and TOU rates encourage customers to shift usage outside critical peak hours, and the batteries provide extra power to the company during predicted high demand periods. They’re hoping that these factors will postpone or eliminate a very high capital cost project to beef up a substation that feeds our area.

Note that to determine our actual power usage, I have to add the net Powerwall consumption to Total Consumption. That’s because the Powerwalls consume more energy than they provide. You would think this would be just the initial 26 kWh charge, but it’s a lot more than that — about 200 kWh since last October. Some of that is the standby power consumed by the inverters, which is about 15W, but I believe most of it is due to the power conversion efficiency being less than 100%. However, the Gateway doesn’t include net Powerwall consumption in it’s Home Consumption figure, so it should match the GEM’s Total Consumption figure.

[Teken]

Possible to see a wiring diagram of the entire setup instead? In the GEMS interval web page you can adjust each CT to fine tune the default values. Keep in mind once logged in it will interrupt the data flow to the Dash Box / Other until you leave and close the page.

Lastly, there’s a constant current setting that can be changed. I’ve never changed it and it’s highly recommended that you only do so with guidance from Brultech.

[ben]

Can you post some graphs (or email me access if possible) of daily on a few circuits? Curious if the readings change at all when it switches to battery power versus panel power.

[peppersass]

Can you post some graphs (or email me access if possible) of daily on a few circuits? Curious if the readings change at all when it switches to battery power versus panel power.

I thought I had replied, Ben, but I don't see my reply here.

The discrepancy between Total Consumption and the All Circuits group predates installation of solar and the batteries, so that must have something to do with calibration.

I don't believe I reported that in addition to the discrepancy between the GEM/Dashbox Total Consumption and the Tesla Gateway's Home Usage, there's also a discrepancy between the two on solar and battery production/consumption. For example, yesterday the solar production reported by the GEM/Dashbox was 1 kWh lower than reported by the Gateway. The discrepancy for net battery was off by about .4 kWh. When I get a chance I'll go back through the history and see if the discrepancy varies. My impression is that the Gateway numbers are always larger, but more in line with what the power company bill says.

I suppose it's conceivable that the waveforms of the inverter-produced power from solar and/or the batteries affects the Split-200 CTs on my mains differently than the Gateway's CTs, or there's some conditioning of the CT signals that's different between the GEM and Gateway.

I'll get you some graphs, but right now I'm in the process of shortening and re-dressing all of my CT wires, which I've been meaning to do for years. All the extra wire for each of my 42 CTs is bundled up and stuffed into cable protective tubing (the open slit kind.) I didn't shorten and dress the wires when I installed my GEM because I wasn't sure whether it would stay in its present position or have to be moved further from the panel to make room for a transfer switch, solar, batteries, etc. With all that now installed the original GEM location is final, so I'll re-run the wires through conduit and pull boxes, then shorten them.

Once I do that, I'll recheck for discrepancies and produce some graphs.