[Sphyrna-devel] Integration time probably found

[Elrond]

Hi girls and guys, ;)

I think, I found the integration time and understood
somewhat, how the integration works / gets "stopped".

- Sending CMD00, reg 2, ls word (byte $8,$9)
  is the max integration time in units of
    (1 / w) seconds
    w = 3928.77 +/- 5.803

  Default: 3999, so about 1sec.

- How did I measure w?
  Simple: I started about 250 different measurements with
  the above word in the range 15--3999 and timed how long
  it took for the measurement to finish. Afterwards a
  simple fit in gnuplot.

- Received CMD10, reg 0x12, hi word (byte $10,$11)
  is the actutal integration time, that was used in this
  measurement. It can be less than the sent "max
  integration time".

- Assuming, that CMD23 (and its max, also found in CMD10),
  are some sort of correlation integrators:

  It seems, that the integration is stopped prematurely, if
  one of the correlators hits 0xfff (will then of course be
  the CorMax). That's the reason, why the
  RealIntegrationTime can be less than the
  MaxIntegrationTime.

I'm going to commit an update to sphyrna (decode.py and
hammerhead.py) soon.

Dmitry is right: We need some nomenclature. I've tried to
introduce some above, but am not exactly happy with all of
them. (I'll probably add some to decode.py's output, let's
see, who likes them).
And we have a possible problem: if it's wrong, we end up
with speaking like "What we formerly knew as MaxCorPos
seems to be the washing machine status" or "The washing
machine status (formerly known as CorMaxPos) is doing
strange things when I switch of my machine".


   Elrond