Re: Apple IIgs random hardware crashing - any ideas?

[mjmahon@aol.com (Michael J. Mahon)]

ToasterKing wrote:

>Okay, as a follow-up to the totally bizarre behaviour of my ROM 3 IIgs
>as of late, I've stumbled upon something interesting.  Last night I
>completely removed the IIgs motherboard from its case, and reinstalled
>the cards and reconnected the cables with it sitting unhoused on my
>wooden shelf.  Ever since then, it has worked fine.  I used it for a
>few hours last night with SCSI and full acceleration and never had a
>problem.  Since then I've left it repeating the built-in hardware
>self-test for 24.5 hours, and it hasn't failed once.  (Formerly it
>would fail within 15 minutes of repeated tests.)  The funny thing is I
>see no way any part of the board could possibly be shorting to the
>ground shield in the case -- it has more than enough clearance
>everywhere.  I've also tested to be sure that the add-on ports
>installed in break-out panels in the rear of the case (connector for
>SCSI card, jacks for sound card, hyperstudio sound-sampler add-on card)
>and the power supply are not causing ground loops, both by checking
>with an ohmmeter and by disconnecting/removing them.  The GS still has
>crashing problems only when the board is seated in its case.  (The
>board has been removed/reinstalled several times and still this remains
>true.)  I've also tried poking/prodding/thumping the board while out of
>the case and while performing intense operations to see if I could
>invoke a crash, which would indicate I had a cracked trace, cold solder
>joint, or other loose connection, but still it refuses to crash outside
>of the case.  Also, the video corruption I mentioned while running NFC
>Megademo disappears when the GS board is removed from the case.
>
>Now at least I know my board is okay, but this is driving me crazy! 
>Does anyone have any bright ideas I neglected to consider??

Wow--nice problem.  ;-)

Seriously, intermittents are the worst problems to troubleshoot,
for all the reasons you have found.  There are three basic approaches
to fixing them, each with its advantages and disadvantages:

1) Fix things that might be wrong and then test extensively
   to see if the problem is fixed.

2) Monitor critical signals to see what is failing when the
   fault occurs, permitting the fault to be localized.

3) Find some way of making the intermittent fault occur at will,
   permitting the fault to be localized by ordinary methods.

You've done a lot of inspection and fixing (a la 1) without success.

You've found a variation of 3) in which the fault does _not_ occur.

(This is where some smart alec usually suggests running it
without a case.  ;-)

The only effects that the case could reasonably (?) have are
1) raising the temperature locally (not very likely), or 2) putting
mechanical stresses on the board that you have not been able
to reproduce manually (a bit more likely).

I'd suggest suspending the board outside the case by several
points (perhaps resting it on 6-12 erasers) and then--while
running a diagnostic that usually exercises the fault--pressing
locally with a pencil eraser to produce some board bending.

Work your way around the board, watching for a fault.

If you don't find any pressing on the top of the board, turn it
over and repeat the test pressing on the bottom (this will
flex different traces).

You probably don't want to hear that problems like this can
often be caused by almost invisible cracks in traces, often
near solder joints or vias.

Some folks might just bite the bullet and resolder everything
in sight (a method 1 overkill), but "fixing what ain't broken" is
a risk I would rather avoid.

I repeat, you have a worthy problem, and when you find and
fix it, you can be proud of yourself!  (And let us know, so we
can be proud of you, too.  ;-)

-michael

Check out amazing quality sound for 8-bit Apples on my
Home page:  http://members.aol.com/MJMahon/