Bad Feelings...

From: A.R. Duell <ard12_at_eng.cam.ac.uk>
Date: Mon Jun 23 09:28:43 1997

>
> On Sun, 22 Jun 1997, A.R. Duell wrote:
>
> > Yes, but it some cases the 'kludges' led to problems later on - 'trivial'
> > problems like a total incompatability between the Disk II and just about
> > any other machine in the world, 'mariginal behaviour' like a case that
> > overheated, a PSU that was beyond the design limit when running a system
> > board, language card and 1 drive, things like that that caused some
> > machines to crash after about 1 hour, that _crazy_ slot addressing scheme
> > and the saving on chips/PSU consumption by switching the power line to
> > the I/O card ROM - I may be old-fashioned, but I don't like driving input
> > pins past the supply rails..., etc
> >
> > Saving components is only 'good' when it doesn't affect performance. I am
> > not convinced that this is the case with the Apple ][
>
> OK, now you're just outright wrong. I've never had any of the problems
> you mentioned above with any of my apples. The only problems I ever had

OK, it appears that I had a very marginal machine. It happens. On the
other hand (and I _know_ others will disagree here) I _never_ had any
crashes from my TRS-80 Model 1. It was 100% stable.

> was when I pulled the disk controller card from my ][+ when it was still
> powered on (I was young and lame). As far as the system over-heating,
> nope. Never happened. PSU beyond the design limit when running a system
> board? What on earth are you talking about? Crashed after an hour?

When I was having serious Apple ][ problems I measured the supply current
drawn by my Apple ][ motherboard, language card and one disk. IIRC it was
slightly _more_ than the rating for the PSU given in the Techref.

> Mine never, ever spontaneously crashed, and I've owned several. Crazy
> slot addressing scheme? I think it wored rather well. And the language
> card was for the ][+ to have backward compatibility with the ][. That's
> bad design? Actually, that's called "customer friendly".

I never said the language card was a bad design. It was a useful piece of
hardware.

>
> Of course, you're entitled to your opinion Tony, but I think it's driven
> mostly by ignorance (I don't mean that in a bad way) or just plain bad luck.

I was trying to use an Apple ][ seriously in about 1983. I had no luck at
all with it. I read the TechRef, and the view I got _time and again_ was
that this 'feature' existed to save a cheap TTL chip. Now, IMHO that's
_not_ a good design.



I can name a number of things (not Apple, not even just classic computers)
that I've had to redesign because the manufacturers wanted to save a few
bucks worth of components. I'd have much rather payed extra in the first
place, but alas the design was driven by price.

>
> > Oh, please don't think I am picking on Apple. The PERQ (my favourite
> > machine) has a number of _very_ marginal timings. The various CPU board
> > clocks are delayed with respect to each other by a string of TTL inverters
> > and buffers. Some memory cards will only work in landscape mode - the
> > timing is 'on the edge', so that the portrait mode doesn't work. And for
> > those that do work in portrait mode as well you often have to change a
> > coupld of chips to slower versions (say a 74S157 -> 74LS157) to get it to
> > work correctly.
>
> If what you described above is correct, then did it work? If so, it

Yes, it worked. It still works 100% AFAIK

> wasn't such a bad design after all, was it? If not, then yeah, the design

A bad design can still work. It is just more likely to fail under strange
circumstances, it's dependant on component values, etc.

Apart from my bad luck on the PSU, the Apple ][ works. It still works. In
that sense it was not a bad design. But if you accept that things that
don't work are not simply bad designs - they're useless - then you have to
find some other way to judge 'goodness'.

> sucked. What you describe here, in contrast to the Apple design, is in
> fact poor design. The Apple ][, as designed the way it was, worked fine.

The PERQ worked as designed (or at least my 3 all do)

Remember that in many areas the PERQ was pushing a 1970's design to the
limits. There's a 100MHz clock on the Landscape memory board (at a time
when most other manufacturers thought 20MHz was fast). There's signals on
the CPU board that are skewed by 4ns - a delay that could easily have been
lost by bad layout or spread of chip timings. They were trying to design a
fast personal workstation, and managed to do it. The Apple ][ was not
particularly fast (in the signal timings) for the time, and IMHO there's
no excuse for some of the tricks that were pulled.

> Sam


--
-tony
ard12_at_eng.cam.ac.uk
The gates in my computer are AND,OR and NOT, not Bill
Received on Mon Jun 23 1997 - 09:28:43 BST

This archive was generated by hypermail 2.3.0 : Fri Oct 10 2014 - 23:30:30 BST