[rescue] Slightly OT: Bad Cap Saga

[Curious George]

--- On Mon, 8/18/08, Carl R. Friend <crfriend at rcn.com> wrote:

> > Geoffrey S. Mendelson wrote:
> >> Everyone used to [test]  memory chips. At one particular time, no
> >> one made 200ms memory chips, they made 150ms ones. The ones that
> >> passed testing at 150ms were sold as 150's., the ones that passed
> >> at 175 were sold as 175's, the ones that passed at 200
> >> were sold as 200's, and the rest were sold as 225's.
> >
> > Seagate used to test all their hard disks as RLL.  The
> > ones that passed were sold as RLL disks; those that failed
> > were sold as MFM.
> 
> This is normal procedure, but it depends on one thing that seems
> to be missing today -- testing.

Exactly.  How do you "qualify" a device if you don't know *what*
it's characteristics are?

> I have it on good authority that
> Intergraph did this with the Clipper chipset -- chips that
> passed the tests were sold as 10 MHz kit; those that fared worse
> were sold as 9 MHz.  I'm not sure what happened with ones that
> may have passed at lower clock speeds.  I have a 9 MHz specimen in my
> personal collection.

This *had* been true of most "integrated" components.  You'd test
devices and qualify them (or, their "lots") and then brand (and
sell) them accordingly.  As your process improved, the "lower
grades" would inevitably become scarcer.  So, as a customer, you
could often buy a "substandard" part and get virtually the
same performance as the higher priced part.

Of course, it eventually becomes impractical to sort out the
(diminishing numbers of) lower performing chips so you just
drop that grade from your line *or* adjust the pricing so
the differential between "Grade 1" and "Grade 2" is negligible
(since you end up selling the same devices as both grades!)

I can recall Motogorilla (?) testing 64Kb DRAMs as *32Kb* devices
(two different flavors -- one that you used with the MS address
line *hi* and the other with it *low* -- obviously telling you which
half of the die had the defects in it!)

Of course, that was soon impractical as yields improved.  But,
when the technology was new, you (as end user) would take whatever
you could get!