On 12/02/2010 02:28 AM, Teo Zenios wrote:
Anyone who has come up with ideas, i'd encourage you to move forward
with at least a theory or virtual model, if not a physical device.
How hard would it be to make an adapter from SCSI (or laptop IDE) to
SD cards (the ones all these digital cameras use)? I figure there will
be millions of those things in the 256MB to several GB range (perfect
for old systems), they should be fast enough, small enough, and about
worthless plus easy to ship cheaply. People are using IDE to CF
adapters but it seems to be SD is more common and much smaller.
A little bit of history and how we got to SCSI to anything...
In the early days pre-SCSI there was the WD1001HDO , What that was is a
bus level interface not unlike IDE
(8bit bus, a few address lines, read, wrie and select) to a board that
had CPU, RAM,EPROM, and the full
interface from the cpu to ST506 disk drive(MFM). The idea was to make
using the ST506 drives easier
as it was largely interfaces like and IO chip. These were used in many
systems and also existed in forms
like the S100 version (aternate vendors did this like KONAN, Teltek).
Many systems had a functionally
similar board that plugged into do this and for PCs the most well known
was the WD1003 (hint remember
that one).
SCSI to MFM.. A starting point.
the typical bridge board sucj as the WD, Adaptec, or Xybec were very
similar. they contained a SCSI to
microprocessor interface (raw ttl, ASIC, or 5380), A CPU typically Z80
or 8085 and 4-16K of rom and
2-8k of ram followed by (for the ones I have) a WD1010 MFM chip set and
TTL ST506 interface or maybe
the WD11C00 ST506 D interface chipset. The rest is software that does
three major things, SCSI protocol
and interface management, SCSI command structure(read write and
housekeeping), and the needed
routines to read, write and manage the disk hardware and the drive
itself. All on a board that fits in the
common (then) 5.25 footprint. These were emerging soon after the
WD1001HDO as the naucent SASI
bus that later would be come SCSI in all its flavors.
SCSI to IDE..
First why IDE, IDE drives are still available and CF or solid state IDE
is the same interface and parts in the 32MB
to many GB range are available.
A bridge board for SCSI to anything will look the same as most of the
boards already known save for
the one area gets way simpler. the scsi interface can still be done
with TTL, Z80 or 8085s are easily found
in small volumes, Eprom/flash for fixed storage is easy and same for Ram
though a few K is all thats needed
such as a 2116 or maybe 2164. The departure is that the whole disk
interface just goes away and can be
replaces with a trivial bit of TTL to interface to IDE/CF as the "DISK
and interface".
Fundamentally the IDE interface was the WD1003(remember that) and a MFM
drive amalgamated into one
so that the PC bus could literally be connected to the IDE (integrated
electronics disk) directly. Well, it was
never direct as engineers decided a little buffering and decoding of the
PC bus was needed but that was
very trivial. It became a standard. The IDE command set is the WD1003
command set though later
versions allowed for LBA instead of CHS and a few extensions. So what
does that ahve to do with SCSI...
Well, if you can put the whole WD1001HDO or 1003equivilent on the drive
then the interface is ready for
CPU/ram/rom to do intelligent interfaces like SASI and SCSI. Due to
miniaturization in the same time frame
the SCSI to IDE was skipped as the SCSI interface with all the
electronics could be put on the drive.
So what we have been talking about is revisiting the SCSI to IDE bridge
that never emerged as it was a
evolutionary step never needed. Until now.
So to do this all was need to do is build the SCSI interface with TTL or
a FPGA to CPU, add the cpu plus some
ram and rom and then employ the common and electronically simple IDE
interface to add the drive . The drive
could easily be a IDE drive or CF, MMD, FLASH but the idea is the host
interface to whatever. this whole thing
can be done with common parts and need not include ASICs or FPGA, gal or
other programmable logic outside
of Eproms. Parts like an ARM with integrated peripherals are appealing
for this as they offer speed and
may limit the needed external hardware but mounting a BGA or TQFP is at
best unappealing to out of
reach. Additionally many older systems that use SCSI do not stress the
performance needed and a z80
or 8085 is fast enough. Plenty of people have built 8085 or z80 to IDE
so this is known and in the public
domain. there are examples of Z80 to SCSI using nothing but TTL and
software the Visual 1050 used
that and there are others.
So SCSI to IDE is electrically fairly simple. The software is harder to
call but if a z80 with 4k Eprom
could do SCSI to MFM and IDE is simpler or minimally similar to the MFM
chip sets the task should
be about the same magnitude. I don't presume to know all of SCSI but
the AmproLB+ had a CPM
Bios that spoke SCSI to a SCSI hard disk so there is code there and
likely other places to use. I also
had to rewrite it to go from a SCSI bridge with 10mb drive to a 45mb
SCSI drive so there is no
real mystery there. The tasks that need to be performed are likely not
many or complex and it
would only take a spec for a SCSI drive to find out what those commands
and actions would be.
So if SCSI drive supplies went away that's a solution. Myself I'm not
planning on doing it as I have dozens
of SCSI drives in the 45mb to 1.3GB range (DEC RZ2x, RZ55,RZ56 and
Barcudas) to insure my needs. I also
made a point to collect SCSI drives with failed media and electronics as
they can be used to reeanimate
a similar dead drive. On the other front I have a boxes (imagine two 10
ream paper boxes) of IDE drives
in the 10mb to 500mb range and a smaller collection of 1gb to 80gb
drives pulled from those PCs people
just toss. Since I also maintain DEC systems a significant pile of 5mb
to 31mb drives of the ST506, 512 and
Q540(AKA rd52), RD53 and a few RD54 drives. Sure some might fail from
sitting, I've not seen that. Also
drives with known poor histories (as a family) have been eliminated by
use to extinction already, for example
ST251s and JST and some flavors of WD and Segates. Oddly enough the
oldest is a ST506 that refuses to quit
and it's been used hard for a long time. I also have a lot of the
RD53s with the head stick problem
(micropolos 1325s) I've fixed all and they all have long use that says
for that class of drive opening them
without doing the extremes of a clean room has no ill effect.
The only bridge ibe interested in is Q-bus MSCP to IDE/CF/MMD/FLASH to
fill the need in the DEC space as
Q-bus to anything is not well filled with cheap anything especially the
RQDXn series Q-bus to MFM.
In the end If I had to and I didn't have the SCSI spares I see no great
difficulty in wire wrapping a
SCSI to IDE bridge. Some research would be be needed to unsure the
board implemented the SCSI drive
it replaces but if the drive existed the spec should be on hand anyway.
I'm rather pragmatic and practical
on all this as much of the hardware I have has 20+ years of my
experience as well and is maintainable
for a long time to come. When it comes to the system people are
worrying drives when Power supplies,
and a whole host of other subsystems are also likely to fail. Systems
become unsupportable when the
sum total of the parts has an item that has become unobtainium and there
for unrepairable if failed.
Allison