3 Nov
2001
3 Nov
'01
5:01 a.m.
see below, plz.
Dick
----- Original Message -----
From: "Tony Duell" <ard(a)p850ug1.demon.co.uk>
To: <classiccmp(a)classiccmp.org>
Sent: Saturday, November 03, 2001 11:34 AM
Subject: Re: CPU design at the gate level
>
> Well, if you have to repair a terminal with those adders, carry generators,
etc,
Sure, or a minicomputer CPU, a control system, a machine tool, or
whatever else uses them. Believe me, I don't like having to hunt around
for a $1 part to fix a mutli-thousand-dollar (or pound) machine. I'd like
to just go out and buy it.
But, alas, the main market for ICs is not repairers. Nor is it
homebrewers (for whom a board of TTL chips is often a lot easier to debug
than an FPGA, and cheaper as well (a soldering iron is a lot cheaper than
the PC [1] to run the FPGA design tools, even if the latter are free). I
still believe it's educational to make at least one machine from MSI
chips. And also to make one using FPGAs.
[1] I am not sure of the current list price of a modern PC + monitor +
Windows, but in the UK, you can get pre-packaged semi-proprietry machines
for about \pounds 1000. And a bit more if you want soemthing that's
reasonably standard and which you have some hope of being able to upgrade.
I just got an unsolicited email from a vendor in Texas that's asking $279 US
for
what's really an eMachines "eTower-633" (with the trademark badge missing
form
the front of the box) with a 15GB HDD, a DVD drive, 128MB RAM a 56KB modem, and
the usual stuff along with a mouse and keyboard. No monitor and no OS.
Shipping to anywhere in the continental U.S. is $35. A friend of mine bought
one for his own use, and I helped him fiddle with it, and convinced myself that
it's a good and solid enough machine. His hard disk failed and he had a
replacement within a week, so the usual warranty mechanisms are in place as
well.
> in it, there's little you can do without them, aside, perhaps from building
a
> daughterboard with a bunch of programmable parts,
or at least one
significant
Yes, I've had to do that...
> one. The latter always produces the risk of not providing the races that
were
> designed into the original circuit and therefore
failing due to proper
design
practice.
Alas all too true. A lot of the TTL-based stuff from the 1970s has some
poor design practicies in it, but if it works, there's no reason to
attempt to redesign it.
-tony