The results of the last tests.
I measured the contacts on the G7273: all ok.
Next test.
Conditions: console in slot 40, G7273 in slot 41, M9302 in last slot.
slots 42 till last also have G7273's.
Result: hung.
I swapped (sorry Tony) the G7273 of slot 42 and slot 41. Again: hung.
Last test.
Moved G7273 from slot 41 back into slot 42 and put the RX211 in #41.
The machine runs fine. PROMs on M9312, console, CPU regs, all are
accessed without PAUSE going on.
OK, it appears there are 3 possibilities to check out
:
1) The G7273 card is faulty and/or not making proper contact,
and is thus not actually completing the NPG chain.
2) The device _before_ this slot (possibly the arbiter
itself) can drive the input of the RX211 card, but can't drive the
terminator. I have no idea if the latter is a significantly heavier
load, but it might be!.
3) There is a problem before this slot (fault on the arbiter),
which is genuinely asserting a grant signal, but the RX211 is
getting into a state where it doesn't pass on grants, so said
errant grant never gets to the M8302 to cause problems.
Yet again I have to ask you to stop swapping cards around
essentially at random, and to get out the test gear and make some
measurements. With the G7273 card in, look at the grant lines at
the teminator. Is one high? which one? Is it genuinely high, or
marginal? What are the grant lines doing at the arbiter? All low
as they should be? Can you get continuity along the grant chain
with just G7273 cards in the machine? If not, why not? And so on.
Actually get some evidence before trying to cure the fault.
-tony
Well, I am not swapping at random, but at this point the simple tests
are exhausted. With the tests/swapping I established good starting
points where to look. I agree it is time to pull out the "heavier"
tools. This means more than just a quick glance at the schematics.
I will post those results, but that might take a while ...
tnx all,
- Henk.