Weird S100 memory board problem

[Tony Duell]

> 	I have a CompuPro/Godbout RAM14 (16kx8) board, arranged in 4 banks
> of 4k starting at 0. I ran a memory test program on it and it shows address
> 0x3000 as failing; when you look at data written, it appears that BIT 0 is
> bad.

Can you check that you can change and read back at least one bit in this 
block of memory?

> 
> 	So, I ordered some 2147 chips from JDR and swapped the bad chip. It
> still fails the same bit. So, I swapped it with another chip in the row, and
> it fails again. It passes at address 1000 and 2000, so it can't be an
> address bit buffer chip.

I don't quite understand what you are saying here. If you swap the bit 0 
chip with, say, bit 3 of the same block of memory, does bit 0 still fail, 
or does it now work and bit 3 fail. In other words is the fault in a 
particular chip or a particular location on the board?

If, as I guess, it's a particular location on the board, you should see 
if there are any buffers that only apply to that location (e.g. data in 
or out buffers). Also, check the IC socket, and for cracked traces on the 
board going to that chip

-tony