On 2025-07-03 16:12, Wayne S via cctalk wrote:
Yes, stats are kept about issues.
Someone should look at the stats and start to investigate when there’s a lot of
failures with the same issue. Explicit instructions should be sent to field
engineers to take extra steps to document what they found and how it was
resolved, and report their conclusions back to the investigation leader.
That’s how IBM did it.
I know DEC kinda did it for software on VMS though their “Software and
Solutions” database. I really liked looking at that.
Sent from my iPhone
On Jul 3, 2025, at 12:52, Paul Koning<[email protected]> wrote:
On Jul 3, 2025, at 2:26 PM, Wayne S via cctalk<[email protected]> wrote:
That’s a good business practice anyway. You want your high price system up and
running as fast as possible, so not having to do more than cursory diagnostics
is a good thing I think deck realize that with the VAX and it’s remote the
diagnostic capability as for the board breaks, IBM used to do that for all the
boards they replaced. They even had a special board breaking tool.
My CE from IBM said that it costs IBM more to diagnose a faulty board than it
does to make a new one so that’s why they do it. Breaking the board also
ensures that the engineers won’t get caught up in a side project trying to
figure out what went wrong.
That's true for problems seen occasionally. When people realize a particular issue
appears "too often" it does become an engineering matter, because then it
indicates an issue with design or manufacturing or part selection.
For example, I remember a product that had a memory backup battery issue, which turned
out to be a change in plating for the battery holder. For engineering it turned into an
exercise in learning what "electrovoltaic series" means -- not something
familiar to most digital logic EEs.
paul
Stats are very important if the manufacturer has a lot of their own
product under comprehensive maintenance agreements, especially fixed
disk drives that would require major time to recover from backups when
they fail. Control Data had a problem once, I believe it was with the
MMDs, where they noticed premature failures in the field. Because they
kept accurate records, they were able to trace it back to serials after
a particular date when a water filter had been changed at the factory
and the new one caused some sort of problem with the epoxy holding the
magnetic material to the substrate.
cheers
Nigel
--
Nigel Johnson, MSc., MIEEE, MCSE VE3ID/G4AJQ/VA3MCU
Amateur Radio, the origin of the open-source concept!
Skype: TILBURY2591
