17 Dec
2001
17 Dec
'01
11:13 a.m.
On Sun, 16 Dec 2001, Richard Erlacher wrote:
The
peripheral chips in every 1981-82 PC, PC-Clone, etc, that I've got in the
basement (they're all in one box) are 5 MHz parts without exception. These were
readily available to mfg's who purchased the quantity, but weren't available in
surplus for a while.
--
Ben Franchuk --- Pre-historic Cpu's --
www.jetnet.ab.ca/users/bfranchuk/index.html
Peter Wallace
Mesa Electronics
Peter Wallace
Mesa Electronics
Peter, you're missing the point. However, first
of all, EP6xx an EP9xx were not
of the classic "PAL" genre. They were early CPLD's, since they contained,
the
equivalent of two or more PALs. Those were quite expensive, particularly in the
Altera cases, since, although Intel and TI made the physical parts for them,
they disguised it enough that folks thought of the Intel parts as being quite
different, hence quite expensive, thereby not placing cost pressure on ALTERA.
Since the TI parts were identical, their agreement with Altera apparently was
that they'd be noncompetetive. The OTP's which were quite inexpensive, were, at
least in the case of the EP6xx and EP3xx quite reasonable, though not as
inexpensive as the 16R/L/Xnn bipolar parts from TI, NS, MMI, Signetics, and AMD
among others.
The PAL16R/L/Xnnn series were quite a bit less costly, faster, and more
straightforward in their application than the Altera parts, however, costing <$2
in production lots.
I think you are missing the point, In 1981 PALS were expensive...
As far as the EPXXX Parts go, I was just mentioning when we first used
programmable logic. We made low power systems so we had to wait until CMOS
parts were available at a reasonable price. When we started using Altera,
there was no second source, Intel parts were available only a few years
later, and TI after that (and only a few parts)
You could without much ado accomplish the same tasks with TTL MSI/SSI. It's
still not rocket science, and still not expensive, except that more than one
component is involved. Several I/O board makers did that with 8250's as well.
----- Original Message -----
From: "Peter C. Wallace" <pcw(a)mesanet.com>
To: <classiccmp(a)classiccmp.org>
Sent: Sunday, December 16, 2001 1:17 PM
Subject: Re: MITS 2SIO serial chip?
Not quite, you need to generate a wait state so IORQ is more than 200 nS
long (since we lost a cycle generating the RD --> IORQ delay), so you need
a tri-state output to drive IOCHRDY, and theres that niggling timing
question with the DART's clock signal. Maybe it would work with the ISA
clock, maybe not... I think you underestimate how tightly integrated the
Z80 peripherals are with Z80 bus timing...
On Sun, 16 Dec 2001, Ben Franchuk wrote:
> "Peter C. Wallace" wrote:
> >
> > On Sun, 16 Dec 2001, Richard Erlacher wrote:
> >
> > > I disagree that it's a mess. I haven't looked at the requirements
for a
Z80
> > > peripheral since the early
'80's, but I can assure you that I'd dispose
of any
> > > 1st year engineering intern who
couldn't whip up a suitable PAL or
equivalent
> > > MSI/SSI logic to handle the
generation of properly timed inputs to the
thing in
> > > an hour or less.
> >
> > Sure its trivial to do now but we were talking 1981 when PALS were
> > expensive.
>
It was as trivial then as now, if you preferred TTL SSI/MSI logic. I think
you
could do it with a 74LS00 and a 74LS74. That's about $0.25 for the two back
then. I never heard about pal's until about 1990.
In some ways the peripheral
chips are in a really sorry shape. You have vintage slow I/O (2 MHZ?)
or PC motherboard chip sets. Nothing in between. On my FPGA I can run
with a 250 ns memory cycle, but need to stretch it to 625 ns for I/O.
PALs were certainly available earlier, just expensive,
non-reprogramable, and power hungry. I think we used our first
programmable logic in 1986 (Altera EP900s and EP320s - both low power) for
emulation of some PC motherboard stuff in our low power V40 based embedded
PCs. We never used PALs but have used GALs a lot for simple decoders and
random logic.
Bipolar logic was, indeed, power-hungry but that was the technology of the time.
EP???' were not bipolar, they were CMOS, and, in the case of the '80's
perhaps
even NMOS. I'm not certain about that. GALs became avaiable in '84-85 and they
were expensive for only a short time, quickly grabbing market share when folks
realized that the 16V8 replaced the 16L8, 16R4, 16R6, 16R8, and several others.
Once the programming algorithms were available in the common programmers and
there was affordable software with which to use them, and I frequently used
conversion software from PAL fuse maps to GAL parts, having developed them from
PALASM intended for bipolar parts, I never bought another bipolar PAL. For
obvious reasons, it wasn't long before the GALs had replaced al the 16L/R parts.
> At about $.50 now they are hard to beat.
> I think I would do most non-common (probably not Ethernet, USB or
> video) I/O these days with FPGAs. Its great to be able to change the
> function and pinout with just a downloadble config file. A 100K (Well
> maybe 15K if you remove Xilinx inflation factor) SpartanII chip is only
> $19.00...
> There are several free HDL's in Verilog or VHDL
for a fast ethernet interface
that will run entirely in a small Spartan or moderate 4000-series device costing
less than $25, but that's in quantity. Likewise, those can be implemented in
CPLD's for about the same cost. USB is better done in a dedicated (for the USB)
MCU, and at lower cost. Of course you have to license a Device ID. (or whatever
they call it)
Unless you have a compelling need to integrate the MAC part of
Ethernet into a FPGA, FPGA Ethernet is more expensive than an external
Ethernet chip. Also, It is not possible to integrate the PHY into a FPGA
at the moment (well maybe with some oddball mixed signal part) and when I
can get a 100BT Ethernet chip with integrated PHY for less than $5.00, it
would be silly to use a major portion of a $20 chip for a non standard
Ethernet interface that doesn't even have any driver support...