Thought I'd share this, just so it at least gets
archived for posterity...
Idug out my Otrona Attaches this week and decided to try and fix
the16-bit (aka "almost PC-compatible") board in one of them. The
symptom was that the board would run correctly for maybe 90 seconds
before it would crash.
The built in diagnostics (for ROM rev. H) can test the RAM on the 16-bit
board (using the "[" command from the monitor), and doing so revealed
that the RAM test would pass for a little while and then start failing,
indicating "U66" as the fault.According to the service manual this
indicates the actual faulty chip on the board. Not so! Replacing U66
had no effect.
I then got out my Fluke 9010 w/8086 pod (I'm just dying for excuses to
play with this thing...) and had ittest the RAM (the 16-bit board
contains 256K). RAM from word 0x20000 to 0x3FFFF was fine, but the lower
half of RAM started showing bit 0x4000 stuck after warm up -- looking at
the Otrona schematic (
http://oldcomputers.net/Attache_Schematics.pdf)
this would point to chip U61 (which services bit 14 for the lower half
of memory), not U66(which is bit 14 for the upper half).
So: Moral of the story is the Attache's attempt at being helpful is
simply misleading, alas.Hopefully this helps someone else someday :).
- Josh
And the other moral is: logic analyzers rule.