[sebhc] diskette drive

[Patrick Rigney]

> For instance, they got disk controllers to work 
> *without* DMA, on a very slow CPU.
> 
> I know, this makes me an old fogey. 

Not at all!  I agree.  The solution was brute-force, but it's still
nonetheless elegant in its execution.  The proof of that is in the one
measure that I consider to be valuable above all others: reliability.
Whether you were using the hard sector or soft sectored controllers, for the
standard of the day, the disk systems were as reliable as a thrown brick.
But, I think getting the controllers to work without DMA on a slow CPU was
partly handed to them by the relatively low densities that were being used
at the time.  Still, not everyone pulled off what Heath's designers did in
the same environment--I also owned several Commodore PETs in my childhood,
and I slaved at chores and saved for a year to buy the $1295 two-floppy disk
subsystem.  It cost more than the computer it was connected to, it's cost
inflated by the use of not one but TWO 6500-series CPUs to accomplish its
work.  In retrospect, I think that was an embarrassing design compared to
the equally (and sometimes more) capable products produced by Heath, North
Star, and Apple all around the same time.

> The modern design 
> philosophy is "who cares how much hardware it takes -- chips 
> are cheap. Just throw enough of them at the problem. Who 
> cares how much software it takes -- memory is cheap. Throw in 
> a few more megabytes. If it's too slow, raise the clock speed 
> -- who cares how much power it takes."

Lee, I share your opinion on this.  Although I'm probably a bit younger than
most of the folks on the list (I'm just 40), I'm old enough that my first
computer had less than 1K of available memory and six seven-segment displays
(a KIM-1).  I remember writing a lot of fun programs, and even a few useful
ones, and taking great satisfaction in finding two instructions that I could
combine to save the one byte I desperately needed elsewhere.  Today I run a
software company, and every day I wrestle with well-meaning but
inexperienced Java-jockeys whose answer to every performance problem is, in
a nutshell, use faster CPUs in the server, and more of them per server, and
cluster those servers, and...????  They can't add two integers in assembly
language for any processor, and whatever time they spent learning
electronics in school was promptly forgotten after grades were handed out.
(I could go on about the free Cokes, ping-pong tables, custom office
furniture, stock options, and so on).

> There is satisfaction in finding a simple, elegant way to 
> solve a problem. Rarely are such solutions found by using 
> whatever is expedient (already in my junk box), or whatever 
> is cheapest, or copying whatever everyone else uses. I enjoy 
> the challenge of finding elegant solutions; in fact, the 
> journey in finding them is probably more enjoyable than 
> actually using the product they produce.

I think we're on the same page.  I share Steve's view of elegance.  I happen
to have a couple of 179x and 279x chips in my parts box, as well as several
765s.  Of course it would be easy and consistent with the original design of
the machine to build and support a board with any of these devices.  But I
think it is again about goals.  Mine is less to stick with the original
parts and more to provide longevity of the machine as a whole while still
being able to use it.  I could keep that machine forever in mothballs, but I
want to use it.  My goal would not be to build a board that looks like the
old boards and could easily have been on of them--that seems fruitless to
me, because the parts will only get harder to find and more expensive, and
it will never be an original Heath board anyway.  I'd rather save my old
parts for repairing the original boards and keeping them alive as long as
possible.

> Yes, but they left out all the features of the 765 that 
> weren't used in the PC. For instance, they left out 
> single-density mode, which is the standard format for all 
> Heath distribution disks.

Within a short time, through our collective efforts, we're likely to have a
huge collection at our fingertips of distribution media in downloadable
form.  The original distribution media is going bad, and nothing will stop
that.  Whether you use single-density-compatible super I/O chip or an
original 179X/279X/765, the media will just be something to pass around the
room and look at in ten years--it's data will have gone into quiet repose.
So, for myself, I don't consider that compatibility to be a design goal.
The more we use our vintage floppies and drives, the more we shorten their
life, and degradation isn't linear.  Using a modern chip and a little glue
gives me options and alternatives, and that, in my mind, becomes part of the
preservation of the system as a whole.

> Yes, it really is a beautiful board. But it is also very 
> expensive, and beyond what the average hobbyist could build. 
> If this is what you want, and you have the determination to 
> stick to it to the end, by all means, go right ahead!

I probably will, I just don't know when! I have more projects than time. :-)
I don't mean any disrespect to Howard by saying this, but he's running a
business and the board is, by that necessity, priced accordingly.  I'm sure
the cost of manufacture is lower, but that board is part of a project that
puts food on his table, and he's clearly put considerable time into it, and
the price reflects that time as much if not more than the actual cost of
manufacture.  He also has the additional burden of implementing an
interstitial bus (IEEE-696/S-100) to very close tolerance to its standard,
while at the same time supporting the exceptions that were common at the
time.  Actually, had such a board been available in 1982, it would have
probably cost the same or more, so in today's dollars, I think it's an
excellent value.  I can tell you first hand that running a North Star
Horizon on an IDE hard drive is worth every penny of what Howard charges
(and I'm probably going to buy another).  As a hobbyist, my time is free,
and I know I'll learn a lot doing a similar project.  And besides, is any of
us an "average hobbyist?" :-)

That reminds me of a question I keep forgetting to ask: are the busses in
the H-8 and H-89 at all compatible?  Could a card be designed to work for
both?  I know that each had its own version of the H-17 controller, yet they
share the 444-19 ROM.  Aside from the difference in form factor, how
different are they?  (I don't have an H-8; yes, I'm a loser, I know. I'm
working on it.)

A final word on your "old fogey" comment: every time you "old fogeys" speak,
I learn something.  It is my earnest goal to be an "old fogey" some day
myself (in the eyes of the 20-year old software engineer sitting across the
room from me, I already am).  Whether or not we agree on every point, I
acknowledge and respect your considerable experience.

Patrick


--
Delivered by the SEBHC Mailing List