[rescue] Slightly OT: Bad Cap Saga

[Curious George]

--- On Mon, 8/18/08, Geoffrey S. Mendelson <gsm at mendelson.com> wrote:

> On Mon, Aug 18, 2008 at 12:54:04PM -0700, Curious George
> wrote:
> > And that rate is determined historically?  Or, just "wishful
> > thinking"?  I.e., if you don't have any recourse with your
> > supplier, how can you come up with a practical business
> > plan?  "Well, *hopefully* we won't have too many defects
> > and then we'll make X% profit..."
> 
> It's often historical data, but how do you rate a new product, or
> a new company? How many companies had big problems because they
> assumed the new company's caps would fail at x% per year, and it
> turned out they were way off, it was 100%. :-)

Exactly.  That was the issue I was alluding to when I said
I wouldn't know how to design in an environment like this.
 
> > I.e, does Dell not care what's going on inside their "warranty
> > failure" machines?  (I've never dealt with a Dell machine
> > that had to be returned under warranty so I don't know if
> > they expect you to return the unit or just ship you a new
> > one, etc.)  How do they refine their statistics to decide
> > which of their vendors are giving them problems (i.e. warranty
> > costs) if they don't examine the product to determine *where*
> > the failure was?  (I assume Dell contracts a given machine out
> > to several different suppliers)  Don't they care about their
> > *process*??  Or, are you saying the nature of the business is
> > such that they *can't* care (i.e., that it is largely out of
> > their hands -- they are just MARKETERS)
> 
> I don't know. I think that Dell would get statistical reports
> on computers sold in the U.S.,

So, "they" take back warranty repair machines and don't just
"discard" them?

E.g., I think HP "fixes" many of its crapjet printers by just
removing the electronics assembly at the rear and dumping the
rest in a box of (hopefully recycled) plastic parts.  If so,
they have no idea what sorts of mechanical failures they are
trashing.

> but I have no idea of how detailed the would be.
> It may just be main board failure, or video sub board
> failure, or so on.
> 
> Do they troubleshoot each one to let's say go down to
> "c47 failure"?
> 
> My guess is that they have a limit. If a particular model
> goes beyond that
> limit they start to look at them. Or maybe not, they just
> close them out on their website. :-)

<frown>

My first job was with a company that, among other things, did
some subcontracted manufacturing for an IBM division (who, in turn,
was working for someone elese, etc.).

The design was only "nominally" complete... a bit beyond the
"prototyping in foil" stage.  So, troubleshooting failures
was a bit above the skill level for a technician so that job
fell to Engineering.

Each time I "fixed"/found something, I had to fill out a 4-part
carbon (NCR) form on which I listed the symptoms, justification
for the action I took (reasoning) and the action itself.  The
last "page" of the form was actually an envelope into which
any replaced (defective) components were placed.

I also had to record the time to identify the problem, determine
the repair and *implement* the repair -- to the tenth of an hour.

<frown>

As a youngster, this seemed a colossal waste of time to me!
It didn't take long to figure out what sorts of things would
go wrong and just go fix them!  All this (needless?) paperwork...

One day my boss relieved me of the form-filling duty -- had
someone else do that for me.  Apparently, someone at IBM had
been a bit annoyed seeing my pencilled in addition to the form:
"Time to fill out this DAMN form:  0.1hrs" and commented on it.

<grin>

Of course, in hindsight, it made perfect sense!
 
> > So, they operate like a Bank Robber, Confidence Man, etc. -- i.e.,
> > reaping whatever gains they can *while* they can yet, all the
> > time *expecting* to get bit sooner or later?
> 
> It's called self insured risk analysis. If you figure that in the
> end 10% of the units you sell are going to have to be replaced,
> you charge 10% more to break even. If you can cut that down to 5%,
> you charge 5% more.

Yes, but that only *really* works when you have a good idea of
what your actual exposure will be.  If you don't have reliable
historical data to base those assumptions...

> > And, you have to reflect that in how you approach your business.
> > If you don't care if you are DBA <name> two years from now, you
> > can probably make a killing screwing over customers.  That may,
> > in fact, be what's hqappening currently as there are lots of
> > new brands in the marketplace now -- many of which I suspect
> > will not be around for very long (either by design or by lack
> > of concern)
> 
> You should see it here. DVD players change brands and due
> to the falling dollar, prices each week. :-)

So, there is no effort made to defend your "brand" -- just
pick a new name and start over... (?)

> > Yes.  But I don't want my inattention to what's happening in
> > The Industry to leave me any more exposed than I need to be...
> 
> True, but as a design engineer, you should leave that to
> the business people. :-)

For the most part, I have found those folks to be quite
ignorant of technical issues.  I let them "educate" me as to
the needs of the user/consumer/market.  But, they rely on
engineering (and manufacturing, etc.) to educate them
to the realities of our respective fields/capabilities.

<shrug>