Are the dozen vehicles you got CAN logs from brand new vehicles? How much variance is there in age? I find it unlikely there be only a +/-0.5% variance in capacity simply based on aging. At least 1-2% loss should easily happen within one year (15k miles) at least from the PIA survey results.
If the CAN report is anything like the CAC in the roadster, it is not a true measure of battery capacity, but rather a rough estimate (and that one does change with conditions and software updates).
As a matter of fact two of them were Tesla-owned demo cars with < 50 miles on them. One didn't even have the shipping protections removed yet. (No, this wasn't local, so no one go and try to get my local service center folks hanged or anything. And no, I'm not going to mention who/how/when/why/where/etc on this to protect the mostly-innocent.)
And I'm quite certain the capacity values reported by the BMS are very accurate (except in the case of a calibration issue from short cycling over an extended period). Additionally, the "60" BMS also correctly reports the capacity that matches external testing.
lol, and then again, some will never agree Tesla that isn't telling the truth!
Yeah, this seems to be a recurring theme for some reason. There really is no getting out of this one. I wouldn't have posted the info if I wasn't certain it was correct. Reasoning behind why they marketed it as an 85 isn't particularly my concern. It's an ~81 kWh pack, it's marketed as an 85. This is false, end of story. There's no real way around that fact. Get over it, do something about it if you must, but don't deny it.
We have it from Jerome Guillien that the A packs do indeed use different cells. I think you just meant they each hold the same capacity?
Interesting. I'd be curious to the source of that info from Jerome, personally. I have four A-pack cells in two of my custom cell torture devices still, and their characteristics match exactly (literally something like a 0.2% margin) of newer cells. I could even pretty likely accurately determine their cycle count by fitting their data with the newer cell data.
Calendar life degradation is virtually non-existent with these cells when stored properly (40-60% SoC, room temp or lower), which is impressive. Cycle count has a measurable effect on capacity, and seems pretty predictable for various condition sets.