Re: Write Signal on Disk II Question

[ferdimh@gmx.de]

>One very interesting thing I noticed is that on a blank track I could
>very few pulses from the MC3470, something like few hundred per
>revolution. Seems like the AGC is not that picky and doesn't generate
>much noise, at least on my drive.
The Disk ][ has a pot to turn down the sensitivity by kicking the
MC3470 slightly out of correct bias.  One can even turn down the
sensitivity that far that there will be NO noise at all on a blank
track.
This pot is needed to allow for 2 zeroes to be read at all. If it is
adjusted to maximum gain the drive won't read two zeroes reliably.
But this trimming has side effects, since it increases jitter. So
there a compromise needs to be found.
>Just for the record, I notice a good mechanical jitter in successive
>revolutions both due to the spindle motor regulator and the rubber
>belt. Sampling at 2 MHz is okay as long as the writing drive was not
>too slow and the reading too fast, because the two consecutive zeroes
>become a problem then.
I noticed about 5% jitter on the track 0 on a freshly inited disk and
I used clone direct-drive drives.
If there is data on the disk, the different patterns all cause much
more jitter. My program is able to read a freshly inited disk without
errors. The problem only arises when there's data on the disk.

At the moment I'm thinking of a real analog PLL to recover the
original clock and use it to clock a shift register.
BTW, a system that just syncs to the data stream is NOT a PLL. A PLL
would be able to adjust the clock (i.e. when reading a disk which was
written on a 320rpm drive on a 280 rpm drive it would sample at
220kHz). The Disk ][ controller doesn't do that (no floppy disk
controller does!), and it's not really needed. A real PLL just
provides more margin which might be useful in tricky situations (like
copying copy-protected disks).

PS: I was never interested in reading copy-protected disks since I
have none. I was actually just looking for a programming challenge.