On Mon, Jan 19, 2004 at 03:39:21PM -0800, jim wrote:
> I'd like to see you think also about making a pdp8 to ide or pdp8 to
> other bus too, while you are looking at it. not to design it, but to
> see if you can adapt the back side of your ide design to suite both,
> since i would guess you would feed some sort of programmable
> asic with the unibus / pdp8 bus stuff, and then drive the ide bus and
> maybe a buffer on the other side.
The bus requirements for Qbus and Unibus are not that far apart. OMNIBUS
or posi/negi-bus stuff for PDP-8 are another matter entirely. Quite
different.
> if you partition the parts of the
> design right, your ide bus could run many older mini's, but I would
> take one or more for both unibus (for pdp 11/15) and for my 8's (L and E)
A PDP-8 IDE interface would be a good idea, but it would probably be better
to be an entirely distinct project from a Qbus and/or Unibus interface.
The SBC-6120 _has_ an IDE interface, but its design is entirely unsuitable
for a TTL or transistorized -8. It has an 82C55 chip as a peripheral to
the 6120 CPU chip, but it depends on a feature of the 61xx-series CPUs to
switch to "console-mode ROM" and take those simple IOTs and emulate them
with a wad of ROM code. I would be quite surprised if you could fit
enough code in two pages to build a device driver that hammered on the 82C55
directly.
There was a SCSI interface for the OMNIBUS. I was supposed to get one from
Charlie Lasner *years* ago to help him merge two divergent firmware streams,
but it never happened. It had a 6809 onboard that parsed a control block
that the OS/8 driver built. It talked to the SCSI drive, got the requested
disk blocks and DMAed the results into the -8's memory. It was a non-
trivial microcomputer sitting on the OMNIBUS. If I were to build one of
_those_ from scratch, I'd probably switch to the 68000, but that's because
I have analyzers and debugging tools for it (Northwest Instruments and Fluke),
not because it _needs_ that much horsepower to talk to a disk. A simple 6502
SBC could probably do the work, but it'd be easier, I think, to drive a 12-bit
data-break interface with a 16-bit-bus uP than an 8-bit-bus uP.
The "problem" boils down to a simple fact: IDE drives want to see a 16-bit
datapath, and on a Qbus or Unibus, you _have_ that. It can be as simple as
mapping IDE registers to Qbus/Unibus I/O space. For a 12-bit machine, it's
a lot of work to map 12 bits to 16 bits. The easy way with an old PDP-8
is to use an I/O co-processor. With the 6100/6120, you can emulate that
functionality with a 16-bit VLSI I/O chip and a lot of ROM code.
It wouldn't be particularly quick, but I always thought that if I were to
whip up a 68000-based OMNIBUS card, I'd also allow it to respond to a variety
of IOTs to emulate whatever peripheral I wanted, especially extra SLUs
or an LP8E. The circuit wouldn't be any harder - it would all be firmware.
-ethan
--
Ethan Dicks, A-130-S Current South Pole Weather at 20-Jan-2004 07:00 Z
South Pole Station
PSC 468 Box 400 Temp -15.1 F (-26.2 C) Windchill -44.4 F (-42.5 C)
APO AP 96598 Wind 10.1 kts Grid 003 Barometer 688.4 mb (10309. ft)
Ethan.Dicks_at_amanda.spole.gov http://penguincentral.com/penguincentral.html
Received on Tue Jan 20 2004 - 01:15:58 GMT