30 Jan
2003
30 Jan
'03
5:35 p.m.
Stuart Johnson wrote:
To make a long story short, the manager complained in
front of
his folks that it took too long to burn a set of EPROM's. One of the
technicians, being a bright fellow, took it upon himself to patch a copy
of MY EPROM for the EPROM programmer and shorten the timing so that it
would go faster. The manager was proud of him and thought that HE (the
manager) had SHOWN those folks in ENGINEERING that his folks had no
flies on them... and it worked, barely, until they bought a different
lot of prom chips. This was the last time I delivered a package of
documentation w/source code, listings, floppy media, etc. to
manufacturing! Guess whose ass got chewed for this?
The real problem in this case isn't that some of the EPROMs failed to
verify; that was a GOOD thing because it alerted you to the situation.
The REAL problem is that you've now shipped a bunch of units to customers
that are marginal and are likely to fail much sooner than they should.
If that had happened to me, I would have asked the VP of Engineering
to raise hell and prop it up with a stick. If people in Manufacturing
want to modify the production tools or process, that needs to be approved
by Engineering, no exceptions.
I've seen similar stupidity where a bean counter authorized the use of
an alternate vendor's parts based solely on the vendor's salesman
claiming they were identical, without bothering to get Engineering to
qualify the alternate part. Similar result: it seemed to work for a
while, but with a later batch we had problems, and the RMA rate for
the units that had been shipped was very high. Turns out that though
some of the timing specs for the alternate part were actually better
than the original, one particular timing spec was not as good, and
proved to be critical. The bean counter got fired, because this fiasco
cost the company about $35K. (The supposed savings would have been
about $1K.)