Building a better "old" computer

From: Richard Erlacher <richard_at_idcomm.com>
Date: Sun Jun 25 01:23:32 2000

Hi, Ernest:

You'll get a lot of answers to this one, but a lot depends on the definition
of "better" that you use. I've recently been fiddling with techniques for
taking what was, back in 1982, considered a pretty good implementation of
the general case of Z80 application, e.g. the Ferguson Big Board, which used
the standard Z80, and NOT the Z80-A which ran at 4 MHz instead of the
standard Z80's 2.5 MHz. I've got a number of these boards so I can test the
various mod's separately, and ultimately intend to use the on-board 20 MHz
clock to drive a faster Z80 with the existing peripherals, since they are
used relatively seldom, with the exception of the video circuitry, but that
doesn't use any of the sadly inadequate Z80 peripheral devices.

The Z80 SBC market was driven mainly by two factors, namely memory speed,
and device cost. The Z80 was pretty cheap in itself, but its peripherals,
which people in general had decided were pretty good, (I wasn't one of them,
by the way.) Because of the hype put into the support chips, they were used
in place of potentially better chips because they did make the design dirt
simple and the supported some features that other devices didn't support,
e.g. Z80 mode-2 interrupts.

With today's technology you can build a board with all the capabilities of
the Ferguson BigBoard, i.e four parallel ports four serial ports, local
video and using a parallel keyboard rather than a terminal, using a single
device, i.e. an FPGA or CPLD (take your pick)and one memory IC. When you're
done, you 'll have a CPU that operates at about 25 MHz, a double/single
density FDC, the parallel and serial capabilities and other features of the
Ferguson board. I don't know whether the result will be better.

At the one-of prices (typically 1000x the advertised price) for Xilinx
FPGA's capable of this, the device won't be cheaper and unless you go to
pretty high volume, that won't change significantly with other FPGA makers.
CPLD's are typically much less costly, costing maybe $350 _at_Q1 for a device
capable of doing the whole job aside from memory. If you roll your own, you
can probably improve somewhat on the 25 MHz, going maybe as far as 50, but
probably not with a Z80 architecture, and certainly not if you stick with
the rather complex logic provided but not needed in the Z80 SIO.

Since I was a logic designer back in the early days of the Z80, I have to
say we can design better circuits today even when we use the technology
available back then. My assertion, of course, is that we can, indeed, make
a better computer, regardles of what your definition of "better" happens to
be, provided that we know in advance what that target is. This is true
because using the newer technology has taught us a lot and because we've had
20+ years' additional experience.

If we capitalize on the later packaging technology we can make the devices
smaller, and using newer power-supply technology, we can cut down on power
consumption by using fewer external supplies and using the newer
manufacturing technology, we can reduce overall power consumption.

In 1979, we had all the LSI/MSI/SSI devices that were used in the best
applications of the early-mid '80's. We also had the small surface mount
technology, though not the memory density of the '82-83 timeframe. The
"fast" Z80 in 1979 was the 'A' version, while, by 1981, there was the 6 MHz
'B' version. Unfortunately there was no complete support peripheral set
until somewhat later. In '82 they came out with an 8 MHz "H" version. This
was never supported with peripherals until the CMOS versions came along.
Those 1979 models work as well today as they ever did. With I/O limitations
imposed on PC's, I'm looking to use my old S-100 boxes once more for the
measurement and control systems I build from time to time. Condensed
versions of these old timers may come back. . . . Probably not, but who
knows.

Now, what was it you wanted to "improve?

Dick


----- Original Message -----
From: Ernest <ernestls_at_home.com>
To: <classiccmp_at_classiccmp.org>
Sent: Saturday, June 24, 2000 10:56 PM
Subject: Building a better "old" computer


>
> Based on what the current computer industry knows about building
computers,
> would it be possible to build a better Z80 based computer today, using the
> same chips that the builders in, say, 1979 had available?
>
> I heard someone say that the manufacturers did the best they could with
what
> they had to work with "back then," and I started to wonder if we could do
it
> better today. Has our understanding of how it all works improved enough to
> do it better now, using the same chips, etc.?
>
> Just curious,
>
> Ernest
>
>
Received on Sun Jun 25 2000 - 01:23:32 BST