Is it possible the car is thinking in terms of 240V even though you have 120V? 240V/8A would tell the circuit to draw 16A - but that should (I hope!!!) flip the breaker fairly quickly. But the pic shows 124V.
That is almost entirely unthinkable. Resistive heating in wires is due to current flow (amps) regardless of voltage and watts, so it is fundamentally critical that EVSEs and EVs to implement and respect amp limits, not watt limits. It would be shocking if an EV got released to the public with the charge rate limit implemented in terms of watts, precisely because it would be so likely to lead to circuit breaker trips (at best) or fires (at worst).
That is almost certainly measurement error by the power meter. There is no way the EVSE would have passed UL testing if it advertised 16A to the EV (implying a 20A circuit) while having a NEMA 5-15 plug attached (continuous loads are limited to 80% of circuit size, so 12A draw on a 15A circuit).
On the other hand, if the bug were in the EV, causing it to draw more than the EVSE was advertising (not unheard of but exceedingly rare), the EVSE is supposed to be monitoring the current draw and immediately open the contractor to cut off power if the EV ever draws more current than the EVSE is offering.
Both your EV and EVSE would have to be utterly defective to see them draw 16A continuously on a 5-15 plug.
Sorry, I still owe you all a round of testing that I wasn’t able to get to this last weekend!
But in the meantime, another power meter arrived, and this one showed (the hopefully accurate) expected results:
- When the car is limited to 8A, it draws about 7.9A
- When the car is set to “MAX”, it draws about 11.5A (vs the charger’s rated 12A)
That’s completely in line with expectations, and makes me feel a lot better about the whole situation. (In case it wasn’t clear, I went with the thicker extension cord before the charger, just because it had fewer moving parts/electronics… less that could go wrong, I think, than an J1772 extension cord. It’s still oversized for a 12A load.)
On the other hand, I had no idea that power metering is so difficult — I only had a half a semester of electricity stuff, and we skipped over any complex math. I’ve had many Kill-a-Watts, but wanted something with wi-fi integration (c’mon, it’s 2025 and the Kill-a-Watt hasn’t changed in like 20 years).
I chose the Shelly at first because it was well-reviewed for Home Assistant integration (open-source smart home system), but apparently it can’t do its primary function very well… and their support was totally unhelpful… so that’s going right back. I ended up going with a “Poniie” (I swear these names are worse than randomly-generated) Kill-A-Watt clone that’s good up to 25A and has wi-fi reporting.
But still, out of curiosity, I will try to find some time to do more testing on both. It’s just hard to find a resistive load around the house… we don’t have any space heaters or toasters, and most of the high-draw appliances we do have are either inductive and/or too “smart” in some way. I don’t have any incandescent lights left, either. Hmm. Maybe there’s a blow-dryer somewhere…
Coffee maker? Hot water/teapot? Seems a success with the new power meter. 
If it’s a bit inductive, then - hopefully - it shouldn’t be a problem. (And if it does have a problem, it’s a really lousy power meter.) The latter is the most challenging. Modern electronics use SMPSs, and the current waveform can look crazy. It can sometimes be so wacky looking that it takes a good bit of effort to get a decent trace on the oscilloscope. (By contrast, the voltage waveform will usually look like a pretty sine wave, even with a SMPS. However, super-choppy current waveforms can cause significant distortion in the voltage waveform.) Traditional methods for measuring power are useless for these, and you have to resort to what I described previously: simultaneously sampling current & voltage using a quality/pricy S&H, ADC, shunt, and divider, and then performing some math. If this can be done for less than $30, then… that’s impressive.
If you aren’t too concerned about accuracy, it’s <$1 worth of extra parts. The device already has an ESP8266 or equivalent to drive the WiFi connection. A shunt resistor is easy; you can use an array of standard SMD resistors if required. A divider is also basically free. You might add a capacitor or two to filter the noise a bit.
The ESP ADC has a 10-bit range across 0-1 volts. You wouldn’t use the full range here on the shunt, and would probably want to stick to 0-0.1 V. That means ~1% accuracy on any given current sample. Voltage would be a bit more. You could increase the shunt range with an op-amp.
It can sample at around 1 kHz, though, and if you get a reading every few seconds you can improve the accuracy substantially beyond the base level. The chip can easily do the integration math.
The ESP ADCs aren’t known to be well-calibrated, but they could do this at the factory fairly easily.
If you wanted to make a good power meter instead of just an adequate one, you could still do it for <$5 (given the ESP chip). Reasonable quality 2-channel external ADCs are just a few bucks each.