13 May
2010
13 May
'10
10:16 a.m.
Well, we can put a bow on this one now. I've found the problem.
Turns out there was a damaged via on pin 40 of A15 (the optional ROM
location) that provides +5 to pin 8 of A14 (the monitor ROM socket).
The result was that pin 8 was floating. Pin 8 of 8355 or 8755 is I/OR-
and is not used in this design (as I/OM- and RD- are used instead) and
so needs to be tied high. With it floating, it was randomly gating an
I/O read onto the bus I suppose. When I was freezing the part, I was
obviously changing the switch point and causing it to mostly be high
instead of sitting at or below threshold.
The via was likely damaged when they made mods to the board for the
Philips logic analyzer demo application.
So, now both the 8355 and the 8755A work.
Nothing sinister going on _inside_ the devices after all ;-)
Thanks to all for playing along.
Chris
On Wednesday (05/12/2010 at 07:58PM -0500), Chris Elmquist wrote:
On Wednesday (05/12/2010 at 06:47PM +0100), Tony Duell
wrote:
yes... and to make matters worse, some of the devices were soldered
in without sockets too! I first suspected the 8279 display+keypad
controller and so I had to sacrifice it by cutting the pins off the
package and then desoldering each pin one at a time. Sadly, that did
not fix the problem and I wasted a nice ceramic packaged 8279 circa 1976.
This board likely had a different history than other SDK-85 since it
was built into this Philips logic analyzer demo system. It's possible
Philips bought the boards assembled from Intel or some middleman who
assembled them in this static configuration.
yes... or shorts. Do you call them shorts over there? aka, short pants.
I still have the scars on my leg from dropping solder blobs onto them
while wearing shorts at the bench as a kid. Not cool.
--
Chris Elmquist
--
Chris Elmquist
That's on the ToDo list. Unfortunately the
ROM extension location at A15
has no socket installed and all the holes are flowed over with solder. So
The PCB was probably wave-soldered with no socket in place. Seems
strange, on an experimental machine like this not to fit the expansion
socckets, but anyway... I need to
carefully solder-suck, solder-wick those clean before I can
install a socket. Boards of this age are quite fond of letting the pads
loose when you heat them up so I need to do this slowly and carefully.
The best way I've found to do this is to hold or clamp the board
vertically, melt th solder with an iron on one side and suck with a
solder sucker from the other. It's not too bad doing it for a 40 pin
socket, it gets boring fast when you have to clear out 16 or 32 16 pin
locations for a memory upgrade (been there, done that, don't wear
t-shirts ;-))