It's not clear how you came to the conclusion that the phantom drain serves no useful purpose.
I said largely..not *no* purpose. Certainly a few (2-3) watts seems reasonable given the various LTE and Bluetooth functions that are operating. It’s true that we don’t know what, if any, conditioning Tesla does at 60 degrees. I suspect that nothing is done though.
And it is absolutely true that I don’t KNOW that the energy is largely wasted. It is just my strong suspicion; I could definitely be wrong.
To my knowledge, self drain of Li-ion is pretty negligible. But, we actually do know (or could know) that the numbers are not HV pack self-drain driven. When in sleep mode, the contactors are disconnected. During this time all the losses must be drawn from the 12V battery, AFAIK. I believe this drain has been measured on a Model S, but similar experiment should be done on a Model 3. So, if we could quantify that, then we would know how much the true sleep mode drain is.
We don’t seem to know what the sleep mode losses are, exactly, without that information. Because as far as I can tell none of the apps actually report this, even if they purport to (based on what
@rawmean has said about the availability of this info in the API). Some apps seem to use battery state info at the beginning and ends of idle periods to infer the sleep mode losses (and Stats specifically only provides the idle mode losses (which are equivalent to total phantom drain) due to the the API limitation as I understand it).
Anyway, the reason I mention sleep mode is that I think we can agree to the extent there are losses there, they have nothing to do with conditioning - because the HV battery is disconnected.
I guess we will find out as time goes by the extent to which I am right; I expect things may get better with time!
Re: your tracking. I understand your separate service would give you exact total usage, but without careful tabulation I can’t come up with how I would use this to infer vampire/phantom losses, even if the charging were entirely at home. Maybe I am being dumb. It seems like it would be hard to separate it out? You could certainly compare to the lifetime kWh usage trip meter in the car to determine how much of the input energy made it to road & other functions, while car was not in park. And get a measure of efficiency that way (but the loss component would be the sum of charger loss, service connection loss, phantom, idling losses when sitting in park, etc.).