From: Karl Guttag
Subject: Second Set of Documents
This is the second set of documents. They particularly related to the 9918.
I knew I had a cache of materials on the 9918 history that I hadn't
found and tonight I found were quite a few documents had been stored.
I figured I should scan them so they never get lost again. I have a
lot more stuff that I may scan later.
The attached are:
- One slide/foil hard copy from a presentation I gave around 1982
or 83. It should be noted that I finished working full time on the 9918
in 1978 or 4 years earlier, but kept up with the 9918 as a pet project.
In late 1978, I was put on trying to debug the 9985 and then late in
1978 I became the lead architect of the 9995 (aka 9985A). In late 1979
I started on the 99000 (aka Alpha) and finished it in 1982. It talks
about the checkered history of the 9918. What many don't realize is
that TI came very close to cancelling the 9918 program in 1980 or 1981.
They would have kept making it for the Home Computer but withdrawn it
from the general market. Mattel showed Interest in the 9918, but the
Home computer projections were so large that that TI was afraid to commit
to Mattel in case there were not enough chips. As it turned out Mattel
sold a lot of systems and Home Computer didn't. Imagine if a system
with Colecovision's capability had come out 2 year earlier.
- Just a year or so after the 9918 was almost withdrawn from the
market, in 1982 Colecovision came out and about the same time the 99/4A
took off. Attached is a Consumer Reports article. The 9918 was fully
function in 1978 and thus was about 4 years old. It is amazing that it
was still the best video game chip on the market.
- The large (B-Sized 11x17 inch) Timing Diagram was an original one
hand drawn by Sergio Maggie. Sergio designed all the high speed clocks
on the 9918 and had a programmable logic array (PLA set of NOR gates)
that generated a bunch of signals that kicked off various events.
The PLA had inputs for the various modes like Text Mode, Graphics Mode,
and the like. You might note that there are clock cycles across the top
that jump from 295 to 466; this was a common type of "cheap trick"
to save some logic, the horizontal counter was designed to with signal
that made it jump to 466 after 295. Note, I'm not sure but we may
have changed the total counts per line so this may not line up with
later versions of the 9918/9918A/9118.
- A few pages that I wrote up on the Pattern and Sprite processing.
The "neat" writing on the bottom of the page I think is Sergio's, the
messy hand print is mine. You will note that these diagrams found their
way into the documentation, but I don’t think I was the first one to
write up the addresses this way.
- An early 1980s memo I wrote on the 9918 memory interface. As I
told you earlier, the 9918 was originally defined "little endian" but
the documentation did a cosmetic swap to big endian. I must have gotten
at least 6 calls from support people trying to help a customer that had
laid out their PC board wrong. You will note in later editions of the
9918 family documentation there is an explicit table and warning as to
how to hook up the DRAM interface.
- Next is an early 1980s memo I wrote on how to put in a "Super Graphics
Mode" which I think ended up being Graphics Mode 2. Note it only took
50 transistors! I was trying to beef up the capability of the 9918 a
bit but there was not a lot of support. Unfortunately as I discussed
in my prior document, there were not enough memory cycles available to
support a simple bit mapped mode.
- Next is a memo I wrote in frustration to Wally Rhines (then VP
of our group and later/today CEO of Mentor Graphics) in 1981. It is
a rather blistering memo. I was working on the 99000 at the time but
still gave occasional support to the 9918. I should add, I was not the
only person helping keep the 9918 alive but I was frustrated that it was
still the best chip available and it had next to no sales after 3 year.
That all changed about 1 year later when Colecovision and the Home
Computer went to high volume for Christmas of 1982 and 1983.
- Attached is an advertisement that is mentioned in the 1982 or 1983
presentation (item #1 above). Basically TI was only selling the 9918
to industrial control system companies rather than to games companies.
The reason marketing gave me was that Industrial Controls is traditionally
where the 9900 family sold the most. I remember distinctly telling them
that the 9918 marketing "Was like a drunk looking under a streetlamp
for his keys, and when asked if that is where he had dropped his keys,
the drunk reply 'no, but the light is better here.'"
I would consider the ones above to be of 9918 historical interest.
I found a number of other documents including ones on video overlay and
Dual VDPs in a single system. Maybe I can send them at another time.
Karl