So, I've finally gotten back to the sage of the missing 8kB on my 102. The clues are as follows:
* 21446 bytes showing free after cold reset (so, 8k missing) * All ram chips populated * No sockets - all RAMs soldered * Good continuity across the address and data buses * Good continuity on the chip select (M5 pin 12 to M6 pin 20) * NO activity on the M5 chip select * Peeking from addresses inside the M5 RAM address space returns the lower 8 bits of the address that I'm reading * Pokes to addresses in M5 range never store - read out is always as above * M6-20 is not shorted to ground (meter shows a capacitor charging when checking that resistance) So, either I've got a bad decoder or a bad SRAM. My question to the collective wisdom of the group is: Which is more likely? Should I replace M5 (decoder) first, or M6 (SRAM)? I'd love for the answer to be M6 as that will mean I don't have to solder the SMT 40H139, but I really have poor gut feeling on which part is failed. My bet is on M6, just because I imagine LSIs to be more prone to failure than a simple mux. Cast your votes. I'll order parts sometime this week but I'll have to wait until mid-August before I have time to actually swap any parts. Thanks to everyone who's chimed in on my problem so far. I've learned quite a bit about the 102 from this experience. -Josh
