21 Feb
2003
21 Feb
'03
6:43 p.m.
Patrick Finnegan wrote:
I agree with
you here - my PCs all have crap cases. Even the top-end
Coolermaster or whatever cases are still (allegedly) cheaply made. Even the
ones that cost $150!
Case in point - my BBC Micro. That thing has been in constant use
at a
college since they bought it, then it got chucked in a skip and one of my
friends rescued it - printer, monitor, drives and all.
Agreed - most of the PC standards (PCI,
Hypertransport, etc) are
closed-distribution, distributed under NDA and cost a lot. Yes, you could
probably get as much info as you needed to rewrite the BIOS from the Linux
source code, but it would still take an age to do it.
Build an IBM PC clone using an 8086 and a few CPLDs and keep tweaking it
until it runs the IBM BIOS. Then grab a copy of "The Undocumented PC" and
the IBM PC Technical Manual and start rewriting the BIOS.
Unfortunately, I'm trying to do a similar thing with a 386-based board with
a Macronix MX83C306/MX83C305 chipset. Even Macronix say they can't help -
"That product has been phased out for 7 years". Anyone got a datasheet for
this thing? I wanted to port TinyBIOS to this board, but without a datasheet
for the chipset, I might as well program an EPROM with random garbage and
try that...
Also, has anyone got a spare copy of the IBM PC Technical Manual?
1) Is it
well-engineered from the ground up?
PCs: no, a lot of the stuff in even the best PCs is of very poor
OK, I agree
with this first point. All three of my examples of 'real
computers' were "well-engineered from the ground up". 2) Can it be
expected to have a long operating life?
PDP-8: I've got one that is 27 years old and still works great,
OK, I
agree with two points, but realistically, I'd argue point 1
implies point 2. If it's well engineered, it _should_ have a long
operating life. 3) Is it
documented?
PDP-8: very well
PC: barely at all. Just try to get information about what your
OK,
here's the first problem I see. 'Newer' IBM systems - well maybe
all three of my examples - are poorly documented down to the
component level. It's actually probably easier to get documentation
about the p690 than the other two examples of mine, but I know you
can't get anything as good as what you'd need to build replacement
firmware or parts from scratch. For example, I just bought, as a part
of a new package, a Pinnacle Systems DC10+ video capture card. I
*know* that it's been available for at least 4 years, probably longer
than that. There are some people that maintain products for very
long lifespans, such as lab-equipment interface cards.
DIY interface cards are even
more fun to repair - "Oh, the LS373 at position
DX2 has died." [sound of engineer swapping the part over] "Oh, it works
again".
One thing I'm slightly suprised about is that
you're not claiming that
it's easy to service 'real computers' at the component level. Of
course, that also would be nearly impossible with my examples ;-).
My example (the
BBC Master 128) used a lot of custom VLSIs - Acorn asked
Texas Instruments to design them a few ICs to run the system - the old PLDs
used in the BBC B were renowned for overheating and then going "pop". A bit
like an AMD Athlon, really. :-)
Later.
--
Phil.
philpem(a)dsl.pipex.com
http://www.philpem.dsl.pipex.com/