Report on Meeting of Hardware Engineers on 12/21/85

Downend-gang, Moore-gang, Moncrief-gang and System-3D-gang were represented. 
Here is a synopsis of the discussions.

Digital RGB standard : Requested by Rich Moore that we come up with a
standard binary format for R-2R video output. 
Cyberball uses 5 bits Red, 6 bits Green & 5 bits Blue. The extra Green bit
helps with the grass playfield shading. Blasteroids used 5-5-5, using the
extra bit for a special function. Jeds circuit uses 8-8-8 (24 bit data
path). 
Conclusion. We can't propose a standard that would be acceptable to all
project leaders and animators. If they want to propose a standard as a
group we will implement it. 

Classes in the different hardwares in development. : Requested by Moore, 
VanElderen and Downend.
In general the hardware engineers did not percieve any productive use in
classes for each other. The purpose of classes would be to inform
programmers and game designers of what is currently available. The best
format would be - one class presented by Doug, Gary and Pat explaining the
main differences between the hardwares developed by each (to be presented
sometime early February), and one class presented by Max/Jed explaining
their system capabilities (time To Be Determined). System 3D wants to wait
until they have silicon before they do anything. 

STATIC RAMs 8Kx8. Some static RAMs are described as "internally syncronous,
externally asyncronous". Jed is concerned that these ICs may not work on
all applications where other static RAMs will work. Specifically, when the
IC is permanently enabled and the address lines are driven from a ripple
counter, the address lines will not change state at the same time. Will the
"internally syncronous device work". 
Conclusion : Erwin is to be asked to contact Toshiba and Fujitsu to find
out the difference between these devices and other devices by the same
manufacturers with similar specifications but no mention of the "internally
syncronous" operation. If there is no difference then the parts can have 
the same Atari p/n as the other static RAMs. If there is a usage 
difference then they should have a seperate p/n.

AVL: It continues to be very difficult (impossible) to obtain "current"
information from the AVL as to what electronic devices to call out for a
parts list for production. Part of the problem is - not being able to
determine recent and commonly used production parts from parts that were
approved years ago and probably never used in production. The
specifications of those older parts should be examined by the engineer for
suitability and cost-effectiveness before being used for production release
product. However one cannot examine every part on a parts list. 
A suggestion was made that "out of date parts" (the criteria for 
out-of-date to be determined) be moved from the current avl database to an 
"old parts" database on an ongoing (bi-annual or annual) basis. Those parts 
could be reactivated at any time if they became "current". The "old-parts" 
database could still be used by anyone who wanted, but the "current" 
database would be the first choice location. The components group are 
encouraged to comment on this. 

PALS : Even though recent costing indicated that Fairchild PALs were the
most cost-effective for production ($1.00), there is a problem using
Fairchild PALs for prototype development. They cannot be programmed by our
engineering DATA I/O logic programmer (unless we buy the adaptor for a few
hundred dollars). This does not affect their suitability for production
since we buy preprogrammed PALs for production. If you need to program a
Fairchild part you can take the code to WYLE and do it there. It is not
known if the National Semiconductor buyout of Fairchild will have any
effect on this situation. 

If you have any comments on these items, or wish to suggest future items 
of discussion, reply to this address. We hope to continue these meetings 
approximately once a month.