[Coco] Hidden 256-color mode
jdaggett at gate.net
jdaggett at gate.net
Thu Jul 28 17:09:38 EDT 2005
going to the 87 GIME chip may have done two things.
1) made tweeks to the design to center performance.
2) cost reduction by doing a die shrink and or wafer size increase.
If my memory is correct around 1986 time frame is when the sub 1 micron barrier
was broken and 6 inch wafers were introduced. Prior to that was 4 inch wafers and
1.25 micron technology was used. For comparison the MC6809 was done on a 4
inch wafer and about 3 micron technology. 0.75 micron and the same die size as
the MC6809 would allow 4 times more transistors. OR conversely taking a large die
and reducing it's size by 1/4th. This would reduce cost of the die.
Engineering and tooling cost for doing improvements would be amoritized over the
first year of production. The only real costs really is that for generating a new mask
set. That probably was $25,000 and spread across say 500,000 parts for the first
year would add a nickel to the cost for the first year. After that the full cost of
changes is realized.
Finally the costs for a second pass of the GIME chip could have already been
planned. In the 1980's it was extremely rare that the first pass of any IC worked with
the first silicon. My guess is that 1986 chip is not the first silicon. Instead is maybe
the third pass and the 1987version was a planned cleanup pass.
On 29 Jul 2005 at 5:53, Nickolas Marentes wrote:
Date sent: Fri, 29 Jul 2005 05:53:18 +1000
From: Nickolas Marentes <nickma at optusnet.com.au>
To: CoCo List <coco at maltedmedia.com>
Subject: [Coco] Hidden 256-color mode
Send reply to: CoCoList for Color Computer Enthusiasts
<coco at maltedmedia.com>
request at maltedmedia.com?subject=unsubscribe>
request at maltedmedia.com?subject=subscribe>
> Why would a company invest a lot of money to release a second revision
> of a custom chip such as the GIME unless the problems were truly
> severe enough to warrant it? The CoCo3 was aimed at being a low cost
> computer and lowering costs was a high priority in order to remain
> competative. As we can see, the 1986 GIME was not so bad that the
> computer could not be used. Many of us are using '86 GIME chips today
> with no real problems. Certainly, no problems that can't be gotten
> around externally via software or via specific brands/speeds of RAM
More information about the Coco