So... CPU board matrix Errors as below, I have checked the obvious such as swapping out the mathbox PROMS and RAMS and verifying all the eproms on the main board. I had a quick poke around the clocks, matrix and accumulator hardware with a scope and ABI boardmaster but I haven't found anything yet (yes I may have missed something as it was a general 'shotgun' test strategy).
Now I am trying to make more sense of the messages it's actually giving me, so from the self test I get bad data on several tests and it looks like a bit is stuck high but am I right in thinking I need to select one of the hardware tests that fails (eg 16) using the diag test point and dip switches to keep it in that test loop and then look at the data that is sitting on the six data lines of the two matrix RAMS or the addressing on MPA0-9 as listed in the TS guide?
I was thinking it might be a dodgy singe wipe socket although I wouldn't see how an output line could be stuck high through bad contact. I have taken the liberty of converting it to binary and padding the leading digits with zeroes which I assume makes sense and does seem to suggest a stuck bit as the tests that pass would appear to leave that bit high anyway:
TEST DATA (expected and actual result)
16 AAAA 1010101010101010 (expected)
ABAA 1010101110101010 (returned)
19 2AAA 0010101010101010 (expected) (2AAA-0000) x 4000 = 2AAA
2BAA 0010101110101010 (returned)
20 2AAA 0010101010101010 (expected) (0000-2AAA) x C000 = 2AAA
2BAA 0010101110101010 (returned)
Accumulator
A walking ones test, should return 0001, 0002, 0004, 0008, 0010, 0020, etc
0001 0000000000000001 (expected on first increment)
0101 0000000100000001 (returned)
Block Index Counter & Matrix processor Indirect addressing
B 0000 (returned) - no value given in the in TS guide.
Accumulator Adder
C 3496 0011010010010110 (expected)
3596 0011010110010110 (returned)
I should probably mention that I removed all the socketed devices and the crystal before using the boardmaster so I suppose it is possible it reported a line as 'high imp' as apposed to short +5v as a result *if* one of the devices is behaving erratically.
cheers
Karl.
karlcdoe2012-09-06 14:37:50
Now I am trying to make more sense of the messages it's actually giving me, so from the self test I get bad data on several tests and it looks like a bit is stuck high but am I right in thinking I need to select one of the hardware tests that fails (eg 16) using the diag test point and dip switches to keep it in that test loop and then look at the data that is sitting on the six data lines of the two matrix RAMS or the addressing on MPA0-9 as listed in the TS guide?
I was thinking it might be a dodgy singe wipe socket although I wouldn't see how an output line could be stuck high through bad contact. I have taken the liberty of converting it to binary and padding the leading digits with zeroes which I assume makes sense and does seem to suggest a stuck bit as the tests that pass would appear to leave that bit high anyway:
TEST DATA (expected and actual result)
16 AAAA 1010101010101010 (expected)
ABAA 1010101110101010 (returned)
19 2AAA 0010101010101010 (expected) (2AAA-0000) x 4000 = 2AAA
2BAA 0010101110101010 (returned)
20 2AAA 0010101010101010 (expected) (0000-2AAA) x C000 = 2AAA
2BAA 0010101110101010 (returned)
Accumulator
A walking ones test, should return 0001, 0002, 0004, 0008, 0010, 0020, etc
0001 0000000000000001 (expected on first increment)
0101 0000000100000001 (returned)
Block Index Counter & Matrix processor Indirect addressing
B 0000 (returned) - no value given in the in TS guide.
Accumulator Adder
C 3496 0011010010010110 (expected)
3596 0011010110010110 (returned)
I should probably mention that I removed all the socketed devices and the crystal before using the boardmaster so I suppose it is possible it reported a line as 'high imp' as apposed to short +5v as a result *if* one of the devices is behaving erratically.
cheers
Karl.
karlcdoe2012-09-06 14:37:50