🚀 go-pugleaf

In article <7697@cbmvax.UUCP> jesup@cbmvax.UUCP (Randell Jesup) writes:
>       The WORDSYNC compares at each bit.  If it sees a match, it will
>re-sync (throwing away any partial word that has net been written).  The
>effect of this is if there are two sync words in the input stream, say
>$4489 $4489, and they are on a word boundary already, the hardware will
>write $4489 $4489.  If they were not on a word boundary (shifted some
>number of bits) you will get only one $4489 in the output.

Does this mean that WORDSYNC has effect during the complete DMA
transfer, i.e., that it re-syncs when it encounters the sync word in
the middle of the track? Or is it (as I originally thought) that
WORDSYNC only determines when the DMA starts and has no effect
thereafter?

I looked into the hardware manual and the wording seems to allow
both interpretations.

The real question is, of course, do I need to write a routine to find
misaligned SYNCS in data I have read with (a replacement for)
ETD_RAWREAD (which knows about WORDSYNC)?

>       PRECOMP 0&1 determine the amount of precomp when writing, from 0ns
>for 00 to 560ns for 11.  For floppies, you need 00 for tracks 0-79 (0-39
>for 5.25" drives), 01 for the inner tracks (140ns).

I know just enough about hardware to be dangerous, so I must ask:
What exactly *is* precompensation, and what is its use?

>Randell Jesup, Keeper of AmigaDos, Commodore Engineering.

Freely_Distributable=Greetings(Not_For_Any_Commercial_Purpose)->
        Olaf.Seibert;

Astounding News! By Express Via Norfolk!   OS+2 is a trademark of
The Atlantic Crossed in Three Days!        Olaf 'Rhialto' Seibert
[E.A. Poe, 1844]                           U211344@hnykun11.bitnet