Date:  10-Oct-93 08:14 CET
From:  Bil Herd [71155,553]
Subj:  histry.doc

Thought I'd finish what I'd started back in January of this year. I
had been talkin 'bout how busted up the 8563, now we get to the part
about how it got fixed... well fixed good enough... well patched
good enough to give every possible attempt at the appearance of
maybe passably working... 

One of the things that got worse instead of better was something
called the back bias generator.  Now as much as I admired the blind
ambition (as opposed to unmitigated gall... no one ever said it was
unmitigated gall and I am not saying that here and now) of slipping
in a major change like that right before a CES show, it became
obvious that it needed fixed.  Now the back-bias generator connects
to the substrate of the chip and if you've every seen the ceramic
versions of the 40 and 48 pin chips you would notice that the pin 1
indicator notch is gold colored.  That is actually a contact to the
substrate.  I have never heard of anyone ever soldering to the pin 1
indicator notch but I had little lose.  At this point all I did have
to lose was a HUGE jar of bad 8563's. (One night a sign in my
handwriting "appeared" on this jar asking "Guess how many working
8563's there are in the jar and win a prize."  Of course if the
number you guessed was a positive real number you were wrong.) I
soldered a wire between this tab and the closet ground pin.  The
left column reappeared though still a little broken up!  The EADY
prompt now proudly stated that the machine was READY and not really
proclaiming it's desire to be known as the shortened version of
Edward.  To fix the remaining tearing we put 330 ohm pullups on the
outputs and adjusted the power supply to 5.3 volts.  This is the
equivalent of letting Tim-the-Tool-Man-Taylor soup up your blender
with a chainsaw motor but it worked.  The side effect was that it
would limit the useful life of the part to days instead of weeks as
was the normal Commodore Quality Standard.  I was afraid that this
fix might be deemed worthy for production.  (said with the kind of
sardonic cynical smile that makes parole officers really hate their 
jobs)

Remember the syncronicity problem?  Remember the revolving door
analogy?  We built a tower for the VIC chip that had something
called a Phase Lock Loop on it which basically acted as a frequency
doubler.  This took the 8.18 Mhz Dot Clock (I think it was 8.18
Mhz.... been to long and too many other dot clock frequencies since
then) and doubled it.  We then ran a wire over to the 8563 and used
this new frequency in place of it's own 16 Mhz clock.  Now this is
equivalent to putting a revolving door at the other end of the room
from the first door and synchronizing them so that they turn at the
same rate.  Now if you get through the first door and walk at the
right speed every time towards the second door you will probably get
through.  This tower working amounted to a True Miracle and was
accompanied by the sound of Hell Freezing over, the Rabbit getting
the Trix, and several instances of Cats and Dogs sleeping together.
This was the first time that making CES became a near possibility. 
We laughed, we cried, we got drunk.  So much in hurry were we that
the little 3" X 3" PCB was produced in 12 hours (a new record) and
cost us about $1000 each.  

A new problem cropped up with sparkle in multi-colored character
mode when used for one of the C64 game modes.  Getting all too use
to this type of crises, I try a few things including adjusting the
power supply to 4.75 volts.  Total time-to-fix, 2 minutes 18
seconds, course now the 80 column display was tearing again.
Machines are marked as to whether they can do 40 column mode, 80
column mode or both. We averaged 1-3 of these crises a day the last
two weeks before CES.  Several of us suffered withdrawal symptoms if
the pressure laxed for even a few minutes.  The contracted security
guards accidentally started locking the door to one of the
development labs during this time.  A hole accidentally appeared in
the wall allowing you to reach through and unlock it.  They
continued to lock it anyways eventhough the gaping hole stood
silent witness to the ineffectiveness of trying to lock us out of
our own lab during a critical design phase.  We admired this
singleness of purpose and considered changing professions.  

We finished getting ready for CES about 2:00 in the morning of the
day we were to leave at 6:00.  On the way to catch the couple of
hours sleep I hear the Live version of Solsbury Hill by Peter Gabriel,
the theme song of the C128 Animals and take this as a good omen.  
Several hapless Programmers are spared the ritual sacrifice this 
night... little do they know they owe their lives to some unknown disc
jockey.

Advertisements in the Las Vegas airport and again on a billboard
enroute from the airport inform us that the C128 has craftily been
designed to be expandable to 512K.  Now it had been designed to be
expandable originally and had been respecified by management so as to
not be expandable in case next year's computer needed the
expendability as the "New" reason to buy a Commodore computer.
That's like not putting brakes on this years model of car so that
next year you can tote the New model as reducing those annoying
head-on crashes. 

Upon arriving at the hotel we find that out hotel reservations have
been canceled by someone who fits the description of an Atari
employee.  Three things occur in rapid succession.  First I find
the nearest person owning a credit card and briskly escort her to
the desk were I rented a room for all available days, second, a
phone call is placed to another nearby hotel canceling the room
reservations for Jack Trameil and company, third, several of those
C64's with built in monitors (C64DX's??? man it's been too long)
are brought out and left laying around the hotel shift supervisors
path accompanied by statements such as "My my, who left this nifty
computer laying here... I'd bet they wouldn't miss it too much".

The next day we meet up with the guy who developed CPM (Von) for
the C128.  As I mentioned earlier, someone forgot to tell him about
the silly little ramifications of an 8563 bug.  His 'puter didn't
do it as he had stopped upgrading 8563s on his development machine
somewhere around Rev 4 and the problem appeared somewhere around
Rev 6.  As Von didn't carry all the machinery to do a CPM rebuild
to fix the bug in software, it looked like CPM might not be
showable.  One third of the booth's design and advertizing was based
on showing CPM.  In TRUE Animal fashion Von sat down with a disk
editor and found every occurrence of bad writes to the 8563 and
hand patched them.  Bear in mind that CPM is stored with the bytes
backwards in sectors that are stored themselves in reverse order. 
Also bear in mind that he could neither increase or decrease the
number of instructions, he could only exchange them for different
ones.  Did I mention hand calculating the new checksums for the
sectors?  All this with a Disk Editor. I was impressed.

Everything else went pretty smooth, every supply was adjusted at
the last moment for best performance for that particular demo.  One
application has reverse green (black on green) and the 330 ohm
pullups won't allow the monitor to turn off fast enough for the
black characters.  I had had alternate pullup packs made up back in
West Chester and put them in to service.  On the average,2 almost working 8563's
would appear each
day, hand carried by poeple coming to Vegas. Another crises, no
problem, this was getting too easy.  If a machine started to sparkle
during the demo, I would pull out my ever present tweak tool and
give a little demonstration as to the adjustability of the New
Commodore power supplies.  People were amazed by Commodore supplies
that worked, much less had a voltage adjustment and an externally
accessible fuse.  I explained (and meant it) that real bad power
supplies with inaccessible fuses were a thing of Commodore's past
and that the New design philosophy meant increased quality and
common sense.

I'm told they removed the fuse access from production
units the month after I left Commodore.




The C128 design team:
SYS32800,123,45,6

Bil Herd       Original design and Hardware team leader.
Dave Haynie    Integration, timing analysis, and all those dirty
,              jobs involving computer analysis which was something
,              totally new for CBM.
Frank Palaia   One of three people in the world who honestly knows
,              how to make a Z80 and a 6502 live peacefully with
,              each other in a synchronous, dual video controller,
,              time sliced, DRAM based system.
Fred Bowen     Kernal and all system like things.  Dangerous when
,              cornered.  Has been known to  brandish common sense
,              when trapped.
Terry Ryan     Brought structure to Basic and got in trouble for
,              it. Threatened with the loss of his job if he ever
,              did anything that made as much sense again.  Has
,              been know to use cynicism in ways that violate most
,              Nuclear Ban Treaties.
Von Ertwine    CPM.  Sacrificed his family's popcorn maker in the
,              search of a better machine.
Dave DiOrio    VIC chip mods and IC team leader.  Ruined the theory
,              that most chip designers were from Pluto.
Victor         MMU integration. Caused much dissention by being one
,              of the nicest guys you'd ever meet.  
Greg Berlin    1571 Disk Drive design.  Originator of Berlin-Speak.
,              I think of Greg every night.  He separated my
,              shoulder in a friendly brawl in a bar parking lot
,              and I still cant sleep on that side.
Dave Siracusa  1571 Software.  Aka "The Butcher"

Not to mention the 8563 designers who made this story possible.


The names of the people who worked on the PCB layout can be found on
the bottom of the PCB.  

"RIP: HERD, FISH, RUBINO"

The syntax refers to an inside joke where we
supposedly gave our lives in an effort to get the FCC production
board done in time, after being informed just the week before by a
middle manager that all the work on the C128 must stop as this
project has gone on far too long.  After the head of Engineering got
back from his business trip and inquired as to why the C128 had been
put on hold, the middle manger nimbly spoke expounding the virtues
of getting right on the job immediately and someone else, _his_ boss
perhaps, had made such an ill suited decision.  The bottom line was
we lived in the PCB layou area for the next several day.  I slept
there on an airmatress or was otherwise available 24 hours a day to
answer any layout questions.  The computer room was so cold that the
Egg Mcmuffins we bought the first day were still good 3 days later.




Distribution:

To: Helmut Jungkunz > [100024,1545]

Receipt of message acknowledged