A little more progress... Here is the BASIC startup code, which is jumped to by the kernel (in F000 ROM) if DIAG is not pulled low (Basically, it goes either here or the monitor): ; initcz Initialize BASIC RAM D3B6 iD3B6 LDX #$FB D3B8 TXS D3B9 LDA #$4C D3BB STA $51 D3BD STA $00 ; USR Function Jump Instr (4C) D3BF LDA #$73 D3C1 LDY #$C3 D3C3 STA $01 ; USR Address [4: C373] D3C5 STY $02 D3C7 LDX #$1C D3C9 iD3C9 LDA $D398,X D3CC STA $6F,X D3CE DEX D3CF BNE $D3C9 D3D1 LDA #$03 D3D3 STA $50 D3D5 TXA D3D6 STA $65 ; Floating -accum. #1: Overflow Digit D3D8 STA $10 ; 3: width of source (unused - from TTY) D3DA STA $15 D3DC STA $0D ; 3: Flag to suppress PRINT or PRINT# D3DE PHA D3DF INX D3E0 STX $01FD D3E3 STX $01FC D3E6 LDX #$16 D3E8 STX $13 ; Pointer Temporary String D3EA LDY #$04 D3EC STA $28 ; Pointer: Start of BASIC Text [0401] D3EE STY $29 D3F0 STA $11 ; Temp: Integer Value D3F2 STY $12 D3F4 TAY D3F5 LDA #$80 D3F7 BNE $D400 D3F9 LDA #$00 D3FB LDY #$B0 D3FD JMP $D41B D400 iD400 INC $11 ; Temp: Integer Value D402 BNE $D408 D404 INC $12 D406 BMI $D417 ; initms Output Power-Up Message D408 iD408 LDA #$55 D40A STA ($11),Y ; Temp: Integer Value D40C CMP ($11),Y ; Temp: Integer Value D40E BNE $D417 ; initms Output Power-Up Message D410 ASL D411 STA ($11),Y ; Temp: Integer Value D413 CMP ($11),Y ; Temp: Integer Value D415 BEQ $D400 ; initms Output Power-Up Message D417 iD417 LDA $11 ; Temp: Integer Value On the working PET, I can use 'G D3B6' and it starts BASIC, just as if it were running normally. On the bad PET, it hangs, and never prints the message which is output by the very next block (D417). [Interesting side note - this NEVER produces the startup "noise", so it is clearly not produced by BASIC either ... a mystery] Now, on the bad pet, if I do 'G D417', which bypasses this block of code, it issues the startup message and comes up in BASIC (somewhat weird because various things were not setup, but it's there). So - thats pretty conclusive proof that this is the block where the problem is manifesting itself. Looking at the block of code, it "shouldn't" hang ... There are only two loops, the first is counted by a register (as long as the CPU is good this should work, and I have swapped the 6502 with the other machine) ... the second goes until RAM fails to verify (this will definately happen by the time it reaches the ROM), or the $11,$12 location wraps to a negative value (32k max RAM). There are no subroutine calls in this block, so it should not depend on a valid stack. Using my "try BASIC then reset to monitor without interrupting power" technique, I see that $11,$12 almost always contains $01,$40, which means that it reached the INC $11 once (first time through). The only way I can see this loop crashing is if the $11 (or $12) location does no increment correctly (RAM fauly) - I've manually tested all kinds of values in them, or if the ROM "goes bad" during code execution. [If the $12 location was faulty, I should see "random" values in $11 depending on the timing of when I hit reset] I'm thinking ROM again (this was my "gut feel" when I started because I have seen so many PET ROM's go bad) - although it reads OK in my EPROM programmer, and also this block dumps correctly from the monitor, both of these accesses are fairly laid back - hitting the code is the first time we execute code from this device... Timing on execute could be tighter (I haven't checked the databook yet), also the frequency of accesses will be much higher ... Perhaps if the ROM is marginal, one of these factors is enough to cause a failure. I think a bad ROM is more likely than a single location in the RAM appearing to work for manual tests, but failing during this particular block of code, especially considering that the rest of RAM is good enough for lots of other stuff to run ... What do you guys think? Anyone have experiences with marginal PET roms? Regards, Dave -- dave04a (at) Dave Dunfield dunfield (dot) Firmware development services & tools: www.dunfield.com com Vintage computing equipment collector. http://www.parse.com/~ddunfield/museum/index.html