Last night I had some success with my asteroids board repair, so I thought I'd share it. I'm writing it up as an old incidence sheet I used to use..
Problem: Asteroids board watchdogs in play mode, runs self test and Catbox + Plus draw program fine. It also runs the 'Diag Step' Diagonal test fine. Fluke 9010 reports no Ram/Rom/Bus/IO faults. I can read/right to the Vector Ram fine. This board was VERY clean, so clean it looked brand new. I gave it the Flex'o test as well, game never worked.
Solution:So when all else fails, grunt work time. In isolating a Program Counter line in the Vector section that look weird on the scope (LOAD6), the line all of a sudden looked fine, reconnect the Pin, and the game worked fine.
Fault Analysis: I then removed the count chip (H6 from memory) and discoverd that a manufacturing flaw occured, there was a small zit of solder that was shorting load6 to gnd under the chip. When I cut the chips pin to fold it up during isolation, I scraped it free with the micro cutter. The reason the self tests were ok, is that I can only surmise that during the execution of the drawing code, that the address that was at fault in bit 6 was never hit. Obviously when it was hit, it sent the vsm into never never land, and thus, control was never returned to the CPU, causing a reset. I can only guess that this board was troublesome from the start and was probably put in storage from day one. There appears to be a few attempts at repairing it, the watchdog and clock sections had chips replaced, those repairs were done cleanly.
I really need to spend some time on the VSM commands and come up with a better test sequence for the Fluke, start out with a 3-4 byte command for just a line, and then offset that line and jump it to a new location so all address get exercised one at a time and under user control (via fluke or catbox) then this sort of thing would have been found faster, at least in this case...
Hope this helps someone..
TomW
---------------------------------------------------------------------------
** To UNSUBSCRIBE from vectorlist, send a message with "UNSUBSCRIBE" in the
** message body to vectorlist-request . Please direct other
** questions, comments, or problems to neil .
