Disk hardware emulation, was Re: Grandfather system RTE6/VM?

From: Peter C. Wallace <pcw_at_mesanet.com>
Date: Mon Dec 1 18:12:09 2003

On Mon, 1 Dec 2003, Tony Duell wrote:

> >
> > On Mon, 2003-12-01 at 16:32, Andreas Holz wrote:
> > > Hello all,
> > >
> > > I would like to see a substitution of the now rapidly dying
> > > MFM/SMD-disks. I would appreciate if I could swap out a defective
> > > Maxtor-XT/ drive, and an IDE-drive into my Symbolics or PDP simply by
> > > connecting the "black-box" to the existing cables.
> >
> > Agreed. Or given the capacity of modern (ish) IDE/SCSI drives versus old
> > hardware, being able to host several virtual drives on one physical one
> > would be interesting, not waste drive space, and make it trivial to back
> > systems up.
> >
> > Several machines' drive controllers could be interfaced across a network
> > to one host. The "black-boxes" would talk ST506 on one side and ethernet
> > on the other, say.
>
> I'd rather have one interface/drive per system. It could probably fit in
> place of the existing drive (certainly on machines like the PERQ where
> there's plenty of space round the hard disk), and would mean the machine
> stays self-contained.
>
> >
> > Interesting idea anyway... but given the nature of ST506 is it feasible?
> > Doesn't ST506 have analogue elements to the interface, and a tight
>
> All the signals on the interface connectors are digital, but the data is
> the 'raw' data from the head, not packed into nice sectors and bytes
> (doing that is the job of the controller). So the timing is, to some
> extent, analogue.
>
> However, suppose the data rate is something like 5MHz (I think that's
> right for ST506). If you sampled the data output from the controller at,
> say, 50MHz, wrote the pattern of 0's and 1's to the new hard disk (OK,
> very wasteful of disk space, but then we're proposing replacing a 20Mbyte
> sisk with a 20Gbyte one or something :-)), and then turned the data on
> the new disk back into a 50MHz signal that you fed back to the
> controller, I think it would work. Might be interesting to try, anyway...


This is what I was trying to suggest, with the added feature of slight data
compression by counting the clocks between data transitions, that way a 16X
clock (80 MHz for MFM data recording) doesn't use 16 bits per data bit, only
4...



>
> -tony
>

Peter Wallace
Received on Mon Dec 01 2003 - 18:12:09 GMT