I wonder if I could somehow validate the data myself. I think we already know the car uses about 250w of power in overhead, and I know that it's really pulling right at 40A from the breaker panel. I would think I could take the amount of time, the current out of the wall and the beginning/ending energy in the battery to come up with a number... Sounds like that's enough data to figure it out right?
Example from my session today:
Added 106 rated miles (245Wh/rmi * 106 rated miles = 26kWh)
On a Chargepoint session of 4:37 minutes, at 30A, with voltage between 198 and 201V (as indicated in the car).
Chargepoint says the session was 27.324kWh
Hand calculations using approximate car numbers says 30A*199V*4.62hr = 27.6kWh
So efficiency was about 26kWh/27.3kWh = 95.2% - which is slightly better than I've measured in the past for 6kW charging (I expected 93-94%). Obviously there is large (245Wh) potential rounding error on the charging kWh (about 1% error). Also using the car numbers (26kWh/27.6kWh) gives 93.4% efficiency - so it could be the Chargepoint is reporting a low number due to miscalibration. Obviously there is error on that in-car value as well, since I don't know exactly where the voltage was the whole time.
Still, I wonder whether they have optimized the charging overhead with recent updates - would allow them to claim better MPGe without any increase in range. Probably not. Enough error sources here that the efficiency may well be close to 93-94% as expected.
But way worse than the TeslaFi numbers. This is why I don't trust them (plus the huge amount of variation you see). You can obviously do a similar test yourself.
I've screwed this math up in the past (this post below is incorrect - I failed to multiply through the hours correctly...):
Model 3 Range Constant Check
But if the overhead is actually 250W from the wall, then this implies that AC-DC conversion efficiency is:
Efficiency * (27.6kWh - 4.62hr*250W) = 26kWh
=> AC-DC conversion efficiency is 98.3%. (this is not the overall efficiency, which is worse, of course)
That's probably too good, so either it's not really 250W from the wall of overhead (less), or some rounding error has created error. But it could be as high as 97% I would think. Only way to easily check the overhead is to fit it to other charging rate datapoints.
In any case, I guess it's possible that TeslaFi "corrects" for this overhead, and is presenting that result. But that would not produce data that would be what you would want for your original purpose. And I'd expect less variation on this efficiency over the range of currents you presented in your original plot, if they are removing the charging overhead (it dominates the lack of efficiency at low charging powers).