Early SS Stern theory of operation document

[Duncan Brown]

OK after an extended grief-related delay, I'm slowly getting back to
munching through the Kordek archives. No idea why he had this... unless
it was from Tom Cahill's files, but even then...

The title of the document (for future Google searches) is

Theory Of Operation Stern's Microprocessor Controlled Solid State Games

Obviously this info would apply pretty well to Bally early SS games too.

I don't think there's exactly anything new and earth shattering in here
that hasn't already been figured out by all the dedicated EE-heads in
the pinball hobby... but it's interesting to read it in the words of the
folks who designed it and knew it intimately.

Enjoy!

http://backglass.org/williams/kordek_archives/stern_ss_game_theory_document_100.pdf

Duncan

[Duncan Brown]

Post by Duncan Brown
OK after an extended grief-related delay, I'm slowly getting back to
munching through the Kordek archives.  No idea why he had this... unless
it was from Tom Cahill's files, but even then...
The title of the document (for future Google searches) is
Theory Of Operation Stern's Microprocessor Controlled Solid State Games
Obviously this info would apply pretty well to Bally early SS games too.
I don't think there's exactly anything new and earth shattering in here
that hasn't already been figured out by all the dedicated EE-heads in
the pinball hobby... but it's interesting to read it in the words of the
folks who designed it and knew it intimately.
Enjoy!
http://backglass.org/williams/kordek_archives/stern_ss_game_theory_document_100.pdf
Duncan

John Robertson helpfully Acrobatized it into a searchable version. Here
that is:

http://backglass.org/williams/kordek_archives/stern_ss_game_theory_document_100_searchable.pdf

Duncan

[Kerry Imming]

Post by Duncan Brown
Theory Of Operation Stern's Microprocessor Controlled Solid State Games
Obviously this info would apply pretty well to Bally early SS games too.
I don't think there's exactly anything new and earth shattering in here
that hasn't already been figured out by all the dedicated EE-heads in
the pinball hobby... but it's interesting to read it in the words of the
folks who designed it and knew it intimately.

Thanks Duncan, this stuff is always interesting to me.

Looking at their RAM test, they waste time checking all 256 binary
values. It would be more efficient and provide almost the same fault
coverage if they shifted a 1 bit through the 8 bits. That is, 9*128
patterns instead of 256*128.

Also, they don't claim to do an address test, so if an address bit is
stuck the test will still pass. It's common to write different values
(i.e. the address) into different locations and read them back to make
sure they indeed writing to unique RAM locations.

- Kerry

[seymour.shabow]

Post by Kerry Imming
Looking at their RAM test, they waste time checking all 256 binary
values. It would be more efficient and provide almost the same fault
coverage if they shifted a 1 bit through the 8 bits. That is, 9*128
patterns instead of 256*128.

They did it the exact same way bally did.... because they copied bally's
software. Until MPU200 time with their own multitasking system.....
where they still test the ram the same exact way.

[c***@gmail.com]

Post by seymour.shabow

Post by Kerry Imming
Looking at their RAM test, they waste time checking all 256 binary
values. It would be more efficient and provide almost the same fault
coverage if they shifted a 1 bit through the 8 bits. That is, 9*128
patterns instead of 256*128.

They did it the exact same way bally did.... because they copied bally's
software. Until MPU200 time with their own multitasking system.....
where they still test the ram the same exact way.

It feels like I see more sterns with bad 5101's than Ballys - where the Bally would fail the test and the Stern still passes... Again probably adding emotion to a "scientific" test.

[seymour.shabow]

Post by c***@gmail.com

Post by seymour.shabow

Post by Kerry Imming
Looking at their RAM test, they waste time checking all 256
binary values. It would be more efficient and provide almost the
same fault coverage if they shifted a 1 bit through the 8 bits.
That is, 9*128 patterns instead of 256*128.

They did it the exact same way bally did.... because they copied
bally's software. Until MPU200 time with their own multitasking
system..... where they still test the ram the same exact way.

It feels like I see more sterns with bad 5101's than Ballys - where
the Bally would fail the test and the Stern still passes... Again
probably adding emotion to a "scientific" test.

Kerry was referring to the 6810 test. The theory is the same for 5101's
except that since it's a 4 bit device, you only test whichever nibble
you don't have tied high (or low.... williams ties their 5101's low, so
the unused nibble is 'read' as a 0, bally and stern tie their unused
5101 nibbles high, so they 'read' as an F (1111) - except for mpu200
games of course, where both nibbles are filled by the dual 5101s on the
board. They're not really 'tying' anything on chip itself high or low,
it's the data lines when the chip select is selecting the 5101, and how
the board would treat those lines when nothing is on the bus....)

The stern mpu200 5101 test holds the combined u13/u8 byte (one nibble
from each ram) in the 6800 B register while it shifts all possible
values though both chips. Unfortunately, if ONE of the 5101s is bad,
there's no way to know which one it is.... so a lot of people would most
likely shift them around, etc. handling them more and static zapping them.

It didn't help that stern used a LOT of AMD 5101's which seem to be more
prone to static zap than others.

Best thing to do, IMO with 5101 rams? Pull them out and store them and
replace with your favorite flavor of NVRam replacement. Solves a couple
issues - gets a modern reliable part in there, and gets rid of the need
to have potentially corrosive batteries leaking.

While writing this it occurs to me that you could stick a logic probe on
the 5101's and be able to figure out which chip has data registers that
NEVER change.... that chip's bad.

[Kerry Imming]

Post by c***@gmail.com
It feels like I see more sterns with bad 5101's than Ballys - where the Bally would fail the test and the Stern still passes... Again probably adding emotion to a "scientific" test.

Depending on the failure mode it could be related to power supply
voltage. Higher voltage results in faster RAM.

- Kerry