Re: Commodore 15 Second disk format routine disassembled

On Fri, Jun 19, 2026 at 2:38 PM groepaz <groepaz_at_gmx.net> wrote:
>
> Am Freitag, 19. Juni 2026, 14:14:33 Mitteleuropäische Sommerzeit schrieb
> Francesco Messineo:
> > On Fri, Jun 19, 2026 at 1:58 PM groepaz <groepaz_at_gmx.net> wrote:
> > > Am Freitag, 19. Juni 2026, 09:34:14 Mitteleuropäische Sommerzeit schrieb
> > >
> > > Francesco Messineo:
> > > > On Fri, Jun 19, 2026 at 12:37 AM Maciej Witkowiak <ytm_at_elysium.pl>
> wrote:
> > > > > In its extended format disk the disk will have tracks 35-41 formatted
> > > > > and
> > > > > it reports 840(!) free blocks.
> > > > >
> > > > > It not only formats the inner tracks, but, if I got the disassembly
> > > > > right,
> > > > > the whole disk is formatted with the same density as tracks 1-17, with
> > > > > 21
> > > > > sectors on each track. I'm not able to reproduce this in VICE or YaPe.
> > > > > I
> > > > > don't have a real 1551 to check how reliable this is either, but
> > > > > indeed
> > > > > it works on emulated Pi1551.
> > > >
> > > > and how does it configure the clock divider to allow 21 sectors on the
> > > > smaller tracks?
> > >
> > > The density config is independent from the track. The problem here is that
> > > the faster/more dense setting becomes less reliable on the inner tracks
> > > (which is why it exists in the first place)
> >
> > let me change slightly the question: how does the firmware pack 21
> > sectors on the smaller tracks without running out of space?
> > With standard shugart/IBM controllers (so fixed clock rate independent
> > from the track number), they are forced to use the lowest amount of
> > sectors that fit in the smallest track in less than one rotation
> > period.
>
> Yes. And it is the same for every track.
>
> > So I'd expect on a 1541 format that 21 sectors do overrun the track
> > length well before reaching track 35.
> > If you have a flux image of such a formatted floppy, I'd like to see it.
>
> No idea what you are asking. Writing a track works exactly the same, no matter
> if Track 1 or Track 35. If you can write 21 sectors to one track, using a
> certain density, you can do the same on every other track.

yes you're right, I was not reasoning correctly, the problem is "only"
the linear (not angular) velocity on smaller tracks that makes bitcell
flux reversals too close to each other.
I apologize for the wrong question.

IZ8DWF