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

From: Tom Jennings <tomj_at_wps.com>
Date: Fri Dec 5 20:45:13 2003

On Fri, 2003-12-05 at 14:28, Jules Richardson wrote:

> > Construct an electrical interface of the simplest possible hardware to
> > get the job done (electrical interface, buss termination as required,
> > etc) and do the whole job in C or even Perl.
>
> I agree, with the only caveat being that the computer required doesn't
> have to be something costing a large amount of money.

As you point out, it can be a case-of-beer's-worth of computer, for many
simulated devices.

> The only coding effort is to talk to the custom homebrew interface that
> handles ST506 or whatever flavour of drive you're needing to emulate.

I take it ST506 has analog data for a controller-side separator; this
makes it harder, but as others point out, it could be 'tricked' or
probably simulated with an 8-bit flash converter, and/or use Duell's
idea for an analog sector recorder!

Disk size being what it is today, you could probably record each
"sector" as AtoD'd analog data in a userland program.

(My LGP-21's rotating memory probably contains interesting data, so I'm
probably going to copy the tracks (32 physical) to disk with an opamp on
the head windings driving a sound card, and post-process the NRZ data
later. I'll likely take multiple copies of each track for safety.)

> The hardware interface is the problem; in the case of ST506 it sounds
> non-trivial and that's even assuming that once built the data stream
> could be decoded and generated by the software (as I get the impression
> that said data stream can change widely depending on what controller was
> used to format the hard drive that you're emulating)


I'm guessing that for slow interfaces, SASI or maybe ST506, you could
make a semi-generic hardware interface that had a dozen latched output
bits, some input bits, an 8-bit DAC and ADC, and with only enough
hardware drivers to talk to it, do the rest of the simulator in
software.

DMA-speed interfaces like SMD would require hardware buffering, and
delays could generate timing issues in the controller. But I bet a lot
of machines could use a slow interface.
Received on Fri Dec 05 2003 - 20:45:13 GMT