Re: envstat wrong: who's at fault?

[der Mouse <mouse%Rodents-Montreal.ORG@localhost>]

Last night, I wrote, of the issue I had with envstat/envsys values,

> One thing I have not yet done is to plot the figures to see what they
> look like - I'm letting the battery drain to the point of the machine
> going dead while recording this information; once that's done I'll
> have some figures worth plotting.

I've now done this.  The pictures are interesting, and clearly indicate
that the discharge rate fairly consistently under-reports the actual
discharge rate (in terms of the reported charge values, that is; I have
no independent measurements of either figure, so I can't tell which
one, if either, correponds to the electrical reality).

ftp.rodents-montreal.org:/mouse/misc/envsys-vaio/ contains two
pictures for this (and a third picture; see below).

drain.both.gif is a plot of the discharge rate reported by envstat (red
line) versus the actual discharge rate obtained by taking the discrete
derivative of the reported charge values and converting to the same
units (green line).

drain.report.gif is a plot of just the red line from drain.both.gif,
with the vertical scale greatly enlarged.  I did this because the scale
compelled by the green line in the other picture makes the red line's
structure pretty much invisible.

The X axis in each case is minutes since start of the run.

The physical reality leads me to believe the actual discharge rate was
fairly constant, so I think (a) there is *substantial* noise in the
reported charge values and (b) the reported discharge rate
significantly under-reports the actual discharge rate.  The integral of
the reported discharge rate (area under the red curve) is less than the
actual difference from start to end (area under the green curve) by a
factor pretty close to 1.3.  The reported rate's integral is 2397.83
watt-minutes while the actual rate's integral is 3133.2 watt-minutes,
the latter agreeing reassuringly well with the difference between the
first and last charge figures in the raw log - 52.46 to 0.24, for a
difference of 52.22 watt-hours, or exactly 3133.2 watt-minutes.

The third picture, charge.gif, is basically the same thing as
drain.both.gif, only the data were collected while recharging the
battery rather than draining it.  Again, red is the reported rate
(charge rate, this time) whereas green is the discrete derivative of
the reported charge values.  The last half-hour of this graph makes me
go "w t *f*!?", but I've looked at the raw logfile manually, and that
jump really is there in the envstat output.

/~\ The ASCII                             Mouse
\ / Ribbon Campaign
 X  Against HTML                mouse%rodents-montreal.org@localhost
/ \ Email!           7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B