RE: 264/TED/Plus4 Story

From: Bil Herd <bherd_at_mercury-cg.com>
Date: Sat, 20 Aug 2011 17:58:05 -0400
Message-ID: <243fa36576751ec256449f2c5ed94583@mail.gmail.com>
The game was Wizard of Wor :)

-----Original Message-----
From: Bil Herd [mailto:bherd@mercury-cg.com]
Sent: Saturday, August 20, 2011 12:22 PM
To: 'cbm-hackers@musoftware.de'
Subject: 264/TED/Plus4 Story

Hi Everyone,

Anyone that knows me knows I love relating a story when I can about the
little details that still make me laugh and occasionally cry these many
years later.  CBM's engineering department was relatively small, it
fluctuated between 20-30 people INCLUCING the technicians, engineers
programmers, IC designers, IC layout people and PCB designers.  Needless to
say since we all pretty much fit on a volleyball court at the same time we
knew each other real well and the best stories assume you knew how so-and-so
acted when drunk, etc.

So Gerrit left an opening for the keyboard story on the TED/264/Plus4, that
really has the back-story of when I, being the 23 year old unschooled long
haired kid finally took ownership of the TED project as opposed to being
just it's custodian.

When I got to CBM as related in the commodore.ca article, I was supposed to
be a programmer for the disk drives, and in Bagnals book I was supposed to
be a technician, good thing I didn’t know that at the time. These were the
days when CBM would add someone to the ranks without knowing where they fit
in, they were just on a constant hunt for talent in what we call the 202
area of Philadelphia.  I ended up the lead on the project because the guy
that was lead was leaving, he didn’t really like TED or understand it, (he
was afraid of the intricacies of the chip)  and he was the third or forth
engineer to have cycled through on the project.  This was the tail end of a
mini-exodus of engineers that weren’t quite suited to CBM style designing,
they could probably have designed test gear for Boeing and been right at
home but at CBM you need to move fast and live with the fact that the
products will never be perfect, in some cases fairly flawed.  I found timing
charts where the original “TED” engineer had set out to prove that the TED
timings didn’t work, he also had a set of "poo poo" docs on the C64, Vic and
PET, even though they were in production and selling he claimed they didn’t
work (okay he was right but for the wrong reasons. >:) ) This guy did things
like add a 22 ns delay for the database capacitance which is unrealistic.
The good news is by studying his work I developed my own style of timing
analysis for adding min/max timings and reference points, being completely
unschooled I had to teach myself everything that was common sense to a grad,
such as Boolean tables, etc.

This 2nd-to-last TED engineer had a nickname in my group, we called him
"shooting star" since he was obviously so much better than us, his time at
CBM was a mere waystation on his astronomic trip that was to be his career.
I would trip over his work again as he had "designed" the CPM cartridge for
the C64, I still have in my files the schematic I found in his stuff a
couple of years later, he clearly had lifted the design from the CPM card
for the Apple.

I got put on TED because I was the only guy left that understood that they
were “sharing” the bus with video and that they would ”steal” more time for
the refresh, etc etc.  I spent my spare time standing next to the prototype,
so I became associated with it.  When I was officially put in charge of the
TED the first thing I did was throw the prototype in the trash pile as it
was based on a VIC chip and had nothing to do with  TED… they were just
killing time by wiring up a smaller C64 to the point where people kept
stealing the R7 Vic chip out of the socket, they would even insert their old
R5 sometimes!  I fought this tendency by ordering a whole tube of R7’s and
put them on the desk with a sign that said take one.  After everyone got the
personal C64’s updated the Vic chip thefts stopped.

So there was ONE RULE when I got the Ted project, it has to have xxx number
of chips (I think it was nine chip) and no more, signed Jack Tramiel.  Well
the TED’s reset circuit consisted of an RC circuit and a 7407.  I proved it
wouldn’t work but I didn’t think about what this really meant as to who had
designed the computer (the answer was no-one had designed the computer at
that point)  It was an easy argument to make that the computer would fail en
masse, we HAD to add a chip for reset, didn’t matter which chip, we simply
could not do it with a transistor or a 7407 gate. (yes I tried feedback to
create hysteresis  )  The answer came back from the mountain that it was
okay to add a chip.  I remember that people were stunned that I had
challenged this (some thought I would be fired) and won.

But here’s where I screwed up, I missed the opportunity to understand why
the reset circuit sucked and consequently why the un-buffered data lines
scanned the keyboard and ultimately the joystick.  The problem was that the
TED design was set by the chip and consequently the chip designers.  They
did an amazing job for a single chip approach, especially for back then,
again it was a workmachine specification for the cheapest computer we could
build, not a game machine, specifically DON’T compete with the C64.  The
chip designers were  Bruce Ahearns, Dave DiOrio and Eric “Chow Yan” Yang,
and they nailed it in my book.  (they had no electrical rule checking, the
chip was hand laid out and hand checked using small rulers and pen plots)

To a chip designer it would make perfect economic sense to scan the concept
of a keyboard with the datalines…. Perfect sense.    I missed this fact
every time I stared at the schematic… _every_ time and I looked at it a
hundred times.  I was looking at the schematic from the eyes of someone that
cant make changes or add chips or impact the design other than a few support
items like reset circuits and some RFI suppression, even the connectors were
already designed by the case designer Ira Velisnsky who has since passed
away.

We never tested the joystick port until the new custom joystick arrived
hand carried by Ira.  (We didn’t even have DIN connectors in engineering at
the time) And boom, plugging in the joystick caused video data corruption.
I looked at it right away and figured that if this was real data corruption
then it should also crash the processor.  I held the joystick up to the
monitor and sure enough it locked up.  Within minutes a crowd formed
including my boss.  I looked at the schematics and finally realized that
indeed the DATA lines are being drug all over the keyboard and down a 4 ft
cable and I laughed.  I knew all the ways that was bad, the ringing and RFI
emission, the susceptibility to interference (w/o joystick) and plain and
simple you DON’T do that to data lines.  It pissed my boss off that I
laughed, I told him that I was laughing at my own oversite, I had after all
walked around with this schematic for many weeks.  He was not amused.  He
said in so many words, “well if you don’t fix it right away your fired”.  It
pissed him off more when I was still grinning and he did his version of
storming out of the lab.  Knowing him he was calculating who to replace me
with.

So once I finally “saw” the dataline problem I knew I didn’t want a hard TTL
part to isolate the lines, hard driven lines are RFI/FCC nightmares,
especially since they would be above the metal keyboard plate, not
below/inside.  TTL also doesn’t like to be shorted out, even via capacitive
loads.  I remembered seeing that MOS had a single port I/O and in my mind
this was perfect, it was ssssllloooowww on the output side (I/O speed
instead of databus), It was NMOS which was more forgiving for RFI and shorts
AND it was made by MOS meaning we could control how many are available
(there was an LS shortage at that time that plagued pipleines for a couple
of years).  In less than two hours the tech (amazing techs) had “dead
 bugged” the 6529 to the underside of the board and also made me up a
special joystick cable.

My boss, Joe, walked in and I was standing there playing a video game, 10
years ago I could remember which game, I have since forgotten. I didn’t say
anything, I wasn’t afraid, I was stone faced.  He thought I hadn’t even
started to fix the problem and raised his voice.  I just pointed at the
monitor.  He got more pissed, I pointed again.  Finally with that grin he
hated I showed him that I had taken the cover off the monitor and wrapped a
20’ joystick cable around the picture tube, right on top of the yoke which
looked scary to the unaccustomed (I had been a TV repairman when I was age
16-19).  I wasn’t playing a game so much as proving that no longer were we
susceptible to casual joystick, I had put 20’ right in the highvoltage
flyback section of what was basically a TV.   He turned and walked out,
total cost to fix was something like $0.24USD per unit.
More importantly though was I took ownership of TED that day, I HAD missed
this very serious flaw because I didn’t question the design and the
background of the origins of the design.  That’s what I had been laughing
about was I had understood that I now needed to do this.  From there we did
do some interesting things that the chip guys wouldn’t have known like
creating a 1/4th duty cycle Phi2 so we could attach chips like the 6551
UART. (Actually we created a Phi0 so we would have data hold time)

The end of the story was that I went to a CES show and woman pulled me into
her booth to show me the educational software she had written for the C64.
I remember her display had nasty lines in it, she didn’t yet have access to
the R7 chips. She told me how she was a small business and how money was
tight and then looked directly at me and told me that CBM had screwed her by
not being compatible with what had taken her a year to write.  When I got
back from CES I wrote a memo, misspellings and all (we got our spellchecker
later that year in the form of a VAX) entitled “Yes Virginia, there is
compatibility”.  That memo in a way, laid the foundation for adding a C64
mode to our next/last 8 bit machine, not so much to compete or doublesell
but to support the developers that had supported us.  Simply put that woman’s
software would work on the next CBM machine, she could go back to worrying
about the other things a small business person does.

Bil

       Message was sent through the cbm-hackers mailing list
Received on 2011-08-20 22:00:17

Archive generated by hypermail 2.2.0.