Fixing a 386SX laptop

Alexandru Lovin thypope at
Sat Oct 10 19:40:13 CDT 2009

Ok, Sir, you want me to do explaining? I will do explaining then. With all
the respect I owe to everyone subscribed to this list.

 The problem that you are going to run into, is that those WITH those
> abilities have their OWN projects that keep them busy.  You have to
> respect that, and the fact that they are MUCH more interested in THEIR
> projects than they are in YOUR projects.

The same Mr. Guzis (yes, he's an exceptionally respectable man, I noticed,
but I didn't know he is usually busy) told me that enough money should
grease my own projects that I have little to no expertise in. So it's a
question of finding the right man and having enough money to do all that.

Why do I even need a 386SX laptop at 20 MHz to have USB via the LPT port and
a custom card to act as a network card and as much RAM as I can give to it ?
Simple. Testing the limits. I want to test what hardware Windows installs
on. A 386 won't get farther than Win95 and apparently '98 (by running Win98
setup using "setup.exe /nm"). I'm not forgetting Linux but that one is much
more modular so I guess even the latest versions would probably allow the
compiling of the kernel for a 386SX or DX.

I asked for someone to sell me a 386DX laptop. Mr. Guzis kindly guided me to
another website. It seems that a gentleman who has his own page on
classiccmp (a Mr. Dunfield) has put a Eurocom Note3300 up for sale in
September 2008 (along with other old laptops). I'm just hoping he still has
it because he asked for $10 for it, which is absolutely cheap.

Why all this old hardware? A while ago, I sent an e-mail to this same list,
asking if people would be interested in having computers with more than one
old processors. Like a system with 16 AMD 386DX-40 CPUs and a lot of RAM. Or
a newly-made Altair 8800 with 32 Z80 CPUs at 20 MHz or so. The ones who did
answer were not very interested in the idea. But I still want to do it. Of
course, it requires digging for old hardware everywhere I can find it, ask
around for specs, ask investors if they want to finance the idea...

Wanna know the results? I still have Hercules 8-bit graphics card I paid
about 100$ for, including shipping, from the US, in the original USPS box
and with the sponge shock absorbers. That one will most probably fit the
Altair 8800's S-100 bus and will be able to do graphics, although this thing
might be able to display colours in some sort of clone of the CGA mode (it's
from 1988, know anything about it?) and I don't want that. Still have to
talk to an expert, prefferably from the old Hercules team itself. I need to
make a new video chipset, able to display the Hercules graphics mode but in
the resolutions of today, far beyond the original.

I have detailed specs of what to put in which system and how to make them
cluster computers. Of course. But investors barely understand the idea.
Except for one or two Americans (who were really out of time to discuss
this) and one Danish man (who was out of money).

> I know somebody who builds race-cars. (Sorry, Jay, but analogies are like
> . . . )  He is currently building an electric car, because he is
> interested in it.  If I wanted him to replace the distributor on my car,
> his time would be worth more than my car is.

How about that. The current idea is to open up a garage that converts old
people's cars to electric or hybrid. I thankfully already found an investor
for this one, but I'm still writing the business plan because there are so
many parts I have to take care of - engines, batteries (know of any
"molten-salt" or Lithium-Cobalt oxide battery makers?) and even the wires to
connect the batteries between themselves. Make enough money, move on to
making new, standalone electric vehicles. Then establish the other company,
start buying rights to remanufacture the old chips I need. Already talked to
an Intel representative in Romania, he said Intel doesn't license x86 chips
to others. I told him we would specify in the agreement that we only want to
re-create the CPUs and not become a competitor, in all the ways a lawyer
team could find. He didn't have an answer for that. Asked if I could call
him later on about this idea (he was an analyst managing/filtering projects
to be funded by them, and I went to him with this proposal). He said sure.
He added that AMD does license x86 products to others. Ah-hah. Great. So the
licensing of the K6-III+ at 450 MHz will be easier (for a system with 8 of
these) or the Athlon 64 x2 Black Edition at 3100 MHz (for a system with four
of these) or the 386DX-40, or the Turion 64 ML-44. Also, this means I'll
have to get all of Intel's CPUs in one shot, because he thinks that will
cannibalize Intel's own market, which it might.

I managed to find an old 16-bit Linux and an 8-bit OS (made for the
Commodore 64 apparently) that resembles Linux. I have the bookmarks on
another computer. A few tens of thousands (or hundreds of thousands or
millions) of Euros later, they will all be made to run on multiple
processors. Same for FreeDOS and ReactOS.

> What you need to do is to get started towards developing your own skills.
> I think that a Weller TCPN is a good beginning soldering iron.  You will
> want a GOOD VOM, but a Chinese Harbor Freight one MIGHT be adequate
> temporarily while you put the funds together.  Then start shopping for a
> used oscilloscope.
I didn't get a good soldering iron yet, but I will someday. Romanians use
soldering "guns," called so because they actually look like hand pistols,
made during the communist times. They're still made somewhere and still
sold, complete with the body and handle made of ebonite (I'm not sure that's
what you call it in English, it's that material looking like plastic but
much more resistant to mechanic shock and high temperatures than plastic).
Cheap, too - about 1.5$ if I recall correctly.

Right. No, I'll tell you what I need. I need one of those big scanners who
are able to give you a detailed print of ALL the connections inside a given
PCB, including the discreet chips on the PCB for one of those 3D views of
all the connections. I don't need to know how to solder, I need to know what
to find and where to find it.

I could then send that 3D view to my Aussie friend and he'd be able to
decipher it and tell me what I need to know. He told me how to bridge
sixteen 286 motherboards - via expanded memory controllers, because expanded
memory works in blocks. Starting with the 386DX system, it would all be
easier because you can build a PCI to PCI bridge.

The Intel analyst also told me that redesigning an existing northbridge or
CPU costs millions, so redesigning chipsets to accomodate more CPUs is not
an option. But since he wasn't interested in the idea, I didn't tell him
about PCI to PCI bridges or hardware EMS controllers for the 16-bit systems.

> Doing it for free IS WORTH IT!

I'm with you on that. But there's absolutely no time to become a world-class
soldering man like the motherboard guy is (he replaced the original tip of
his soldering gun with a bent nail and he can replace SMD chipsets with
that). I have to do something to earn money, and the job diversity around
here is probably exponentially bleaker than it is in the US. Wanna do game
design? Up yours. Wanna do this? Ok, you're hired. You get 330$ monthly and
you get the opportunity to spend 8 hours at the office, usually more!

In this case, I really prefer doing something that puts my brain to work
like converting cars to electric or hybrid.

As far as I know, every year there are international Olympics in Math,
Physics, Latin and whatnot. My associate made it to the national phase in
highschool. Twice, if I'm not mistaking. I still had to explain twice or
three times the concept of making a Diesel-electric car: the Diesel engine
runs at all times, the battery capacity needs to be enough to feed the
electric motor and that's basically it. No gearbox, except for the
front-wheel drive cars, in which case we'll probably block it into the
gear where the ratio is 1:1.

So, I got used to people not understanding what I want to do or why. I
didn't want to bore you with such a lengthy e-mail, so I wasn't too keen on
explaining all this.

> Just one more thing.
> You have to promise to train your own apprentice, and to be patient with
> him.

The company re-making vintage computers would be established for an
unlimited amount of time and it would produce Altairs with Z80 CPUs, 286,
386, Pentium Overdrives and Pentium II Overdrives, K6-III+, Pentium III-S,
Pentium IV single cores (for the socket 478 and 775), Athlon 64 X2, Phenom
X3, Core 2 Duo and Quad, Turion 64 ML-44 and Core 2 Duo and Quad for mobile
computers, forever. Yeah, an apprentice is definitely necessary in this
case. But thank you - I haven't really thought of that. Even such a tiny
detail is important.

Thank you for baring with me and reading all this.

More information about the cctalk mailing list