USB meter.
And I've just re-took data from a Forma, and, yeah, power off is useless. A whole 1mA less than suspend, yaaaaay!
c.f.,
https://github.com/koreader/koreader...ent-1214257679
For reference, a power on sequence will pull more or less that load figure for its whole duration, yielding on the Forma a whopping 9mWh/1mAh (those are *very* optimistic numbers, because I'm using KFMon's boot spinner, which is *extremely* more efficient than Nickel's (Nickel's basically full loads the CPU, this one... doesn't. At all. Meaning my boot sequence is both shorter and less power hungry than a stock one.)).
(There's a tiny bit of noise in those values because that meter has no auxiliary power, so the meter's own drain accounts for some of this. It's extremely minimal though (orders of magnitude below those figures) and I couldn't be arsed to find the right plugs for my other meter. Also, it's the same meter used for those Elipsa numbers (and PW5 & Sage numbers), which makes for perfectly valid comparisons. TL;DR: The point of this experiment isn't in the absolute values, but in the deltas between them).