B Degnan wrote:
This is one of the original black connector versions of the IBM 1904057
XM 407 display cards. It has 9114 RAMs in it, not socketed of course,
so I think if I can probe each RAM chip first to ID the bad chip it'd be
more efficient. I would similarly have to check the 74L chips. oy!
Tony Duell wrote:
Sorry, there was a typo in one of my replies (as ever). I meant '2114'
RAMs (not 2144). The 9114 is much the same chip.
It's 1K*4 static RAM, in my experience a very unreliable chip...
Just a few weeks ago I had to replace two 9114 SRAMs to fix a video display
problem in a Xerox-820/FBB-based machine. The problem there was the wrong
characters being displayed, so it was a 'data-related' problem rather than
address.
I was trying to figure out the failure mode of the 9114's I removed, but
haven't discerned much yet. Nonetheless, I wonder if there was a common failure
mode on these chips, something like inadequate protection on inputs due to chip
geometry or such, so that a design fault might manifest itself in different ways.
Does anyone know if IC manufacturers shared the masks when they cross-licensed
common ICs such as these, so that the same fault may show up in units from
different manufacturers?