Would like some input from legacy dudes :)

I've been tasked with designing a small educational computer that has
to contain entirely "macroscopic" components (i.e. DIP). Further, the
memory must be external so that the address and data busses may be
"blinkenlighted". I will want about 8K of ROM and 8K of RAM.

Volumes will be small (batches of 30 or 40 at a time), but I want the
producible lifespan to be approximately four or five years, without
necessarily holding large stocks of the special parts.

I really want to stick with an architecture I know, since I also have
to write the documentation and a specialized programming language (of
the order of integer BASIC in complexity) for the board.

I see WDC still sells the 65C02, and it seems one can also still buy
the Z80 in 40-DIP. Does anyone have a cogent reason to believe either
one would be better than the other? Also, if I design a board around
NOS legacy 6502s (cheap cheap compared to current-production WDC
parts), will the W65C02 drop in? Are there other people making cheap
pin-compatible 6502s?

I guess I could use an 80C31-class part at a pinch, though I won't
enjoy it. But are they still in production? I see them everywhere but
don't know how much I can read into that.

On Wed, 25 Mar 2009 09:38:00 -0700 (PDT)
larwe <z...@gmail.com> wrote:

> Would like some input from legacy dudes :)
> 
> I've been tasked with designing a small educational computer that has
> to contain entirely "macroscopic" components (i.e. DIP). Further, the
> memory must be external so that the address and data busses may be
> "blinkenlighted". I will want about 8K of ROM and 8K of RAM.
> 
> Volumes will be small (batches of 30 or 40 at a time), but I want the
> producible lifespan to be approximately four or five years, without
> necessarily holding large stocks of the special parts.
> 
> I really want to stick with an architecture I know, since I also have
> to write the documentation and a specialized programming language (of
> the order of integer BASIC in complexity) for the board.
> 
> I see WDC still sells the 65C02, and it seems one can also still buy
> the Z80 in 40-DIP. Does anyone have a cogent reason to believe either
> one would be better than the other? Also, if I design a board around
> NOS legacy 6502s (cheap cheap compared to current-production WDC
> parts), will the W65C02 drop in? Are there other people making cheap
> pin-compatible 6502s?
> 
> I guess I could use an 80C31-class part at a pinch, though I won't
> enjoy it. But are they still in production? I see them everywhere but
> don't know how much I can read into that.
> 

Consider extending your definition of "macroscopic" down to SO
packages.  I personally find 50 mil pitch surface mount to be quicker
to hand solder than DIPs.  Or is the concern that you want something
that can be socketed for when the students let the smoke out?

-- 
Rob Gaddi, Highland Technology
Email address is currently out of order

larwe wrote:
> Would like some input from legacy dudes :)
> 
> I've been tasked with designing a small educational computer that has
> to contain entirely "macroscopic" components (i.e. DIP). Further, the
> memory must be external so that the address and data busses may be
> "blinkenlighted". I will want about 8K of ROM and 8K of RAM.

[...]

Obviously you want a fully static design so you can step it cycle by 
cycle and/or insruction by instruction. Also, it would probably be good 
to have tons of legacy software, such as debuggers, monitors, 
assemblers. For those reasons, I suggest Z80 and making the computer 
compatible either to CP/M 80 or ZX Spectrum.


Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com

On Mar 25, 12:44=A0pm, Rob Gaddi <rga...@technologyhighland.com> wrote:

> > to contain entirely "macroscopic" components (i.e. DIP). Further, the
> > memory must be external so that the address and data busses may be
> > "blinkenlighted". I will want about 8K of ROM and 8K of RAM.
>
> Consider extending your definition of "macroscopic" down to SO
> packages. =A0I personally find 50 mil pitch surface mount to be

Socketing is the issue, not so much for the magic smoke as for
plugging in daughterboards underneath the chip (not the way I wanted t
do it, but not my choice, and yes I know there are other ways of doing
daughterboards if daughterboards it must be).

I personally have no problem with SOIC gull-wing, and little problem
even with TQFP and TSSOP (under magnification). Some parts of the
board will use SOIC glue logic.

On Mar 25, 12:48=A0pm, Vladimir Vassilevsky <antispam_bo...@hotmail.com>
wrote:

>
> Obviously you want a fully static design so you can step it cycle by
> cycle and/or insruction by instruction. Also, it would probably be

Yes, that's true...

> to have tons of legacy software, such as debuggers, monitors,
> assemblers. For those reasons, I suggest Z80 and making the computer

Not a major criterion for this application, and no native development
tools will be necessary on the device except the HLL I'm writing, and
a very simple monitor - for which I have many possible choices. I
can't see a CP/M machine with only 8K of ROM and 8K of RAM, anyway.
And I'm certainly not going to try and emulate all the goodies in a
Spectrum's ULA, the display interface on this device is specified as 7-
seg LEDs and a small text LCD.

z...@gmail.com wrote:

> On Mar 25, 12:48 pm, Vladimir Vassilevsky <antispam_bo...@hotmail.com>
> wrote:
> 
>>Obviously you want a fully static design so you can step it cycle by
>>cycle and/or insruction by instruction. Also, it would probably be
> 
> 
> Yes, that's true...

Perhaps the easiest solution would be the EMULATION of the hardware of 
either one of the legacy processors by means of the modern processor 
like AVR. You can select whatever architecture you like or even invent a 
CPU of your own, you have a lot of useful stuff on chip to assist the 
development.


Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com

Would something like these be acceptable?  It certainly opens up the
possibilities for more modern processors:

http://www.delorie.com/electronics/m16c-26-adapter/
http://www.delorie.com/electronics/r8c-27-adapter/
http://www.delorie.com/pcb/r8c-1b-adapter/
http://www.delorie.com/electronics/bin2seven/

I suspect you can find lots of vendors making adapters like those.

Also, some of the starterkits have headers with external busses:

http://www.delorie.com/pcb/m3a/

The left board is the starterkit, if you put in a 3x32 header you
could plug wires into it, or build a custom right-side board to break
out the specific signals you want in some convenient format.

larwe schrieb:
> Would like some input from legacy dudes :)
> 
> I've been tasked with designing a small educational computer that has
> to contain entirely "macroscopic" components (i.e. DIP). Further, the
> memory must be external so that the address and data busses may be
> "blinkenlighted". I will want about 8K of ROM and 8K of RAM.

http://www.opencores.org/?do=project&who=cpu6502_true_cycle
http://www.opencores.org/?do=project&who=t65

This seems to be a good starting point, and eol parts won't be a problem 
for you.

Otherwise, digikey still sells the 68k, which is an extremely good 
teaching platform due to its easy assembler language.

-- 
Mit freundlichen Grüßen

Frank-Christian Krügel

On Wed, 25 Mar 2009 09:38:00 -0700 (PDT), larwe <z...@gmail.com>
wrote:
><snip>
>I guess I could use an 80C31-class part at a pinch, though I won't
>enjoy it. But are they still in production? I see them everywhere but
>don't know how much I can read into that.

I have boxes of ancient (and working) 80C31's sitting on a shelf. (Not
that it matters to this question.)  But you can get the AT89 from
Atmel and the C8051F from SiLabs.  Both companies make them with
external bus capability, I believe.

I like the 65SC02 and WDC licenses the core in a variety of already
done ASICs as well as in synthesizable form, I think.  So that's a
good choice, too.

Jon

On Mar 25, 3:09=A0pm, Jon Kirwan <j...@infinitefactors.org> wrote:

> I like the 65SC02 and WDC licenses the core in a variety of already
> done ASICs as well as in synthesizable form, I think. =A0So that's a
> good choice, too.

I have a strong leaning to use the 6502, for other reasons. Do you
happen to know the answer to my question about whether a design can be
made to accept both regular 6502 and W65C02 as drop-in substitutes?

At a very quick glance it seems that at most I might need an external
jumper, but if anyone has actually tried this...?