23 Mar
2005
23 Mar
'05
12:14 p.m.
Dwight wrote:
It was mentioned that one could have no more than
two concecutive zeros. It was mentioned that this
was a speed issue of the disk. Actually there is
another reason. The level of the signal from the disk
is constantly changing. Think of driving over a rough
road and you want to count the rocks but not the
larger bumps. If you just set a fixed threshold
on the shocks, every time you went over a bump,
you would miss a bunch of small rocks while the
suspension system was absorbing the larger bump.
That's true, although actually three or even four
consecutive zeros seem to work reliably enough on
a Disk II with the 16-sector controller state machine,
provided that you keep the motor speed reasonably
constant. It just won't work well for interchange,
or on the same machine over a longer period of
time.
The time constant on the threshold detection isn't
fast enough for just a few extra bit times without
a transition to shift it very much.
In principle, with much more complex (and bigger)
prenibblization and postnibblization code the Disk II
could have used 80 nybble values to encode a
256-byte sector into 324 nybbles rather than 342,
but that would not have been enough to get even a
single additional sector per track.
Eric