🚀 go-pugleaf

Daniels Beverly wrote:

> Myth #2: I also recall a similar story concerning televisions.  The
> players are usually sited as being Sony and some American manufacturer
>
> (GE perhaps?)

This one is not a myth.  Genichi Taguchi reported pieces of it in
different books, including "The little Green book":  Taguchi, Genichi,
Introduction to Quality Engineering,  UNIPUB, Kraus International
Publications, White Plains, Asian Productivity Organization, 1986,
perhaps his magnum opus 2 vol. set., Taguchi, Genichi; Clausing, Don,
Technical Editor, Tung, Louise Watanabe, English Translator, System of
Experimental Design: Engineering Methods to Optimize Quality and
Minimize Costs, UNIPUB, Kraus International Publications, White Plains,
NY 10601,  1987  If it's in there, it's in the first volume.  Also in
Taguchi, Genichi; Wu, Yu-in, Introduction to Off-Line Quality Control,
Central Japan Quality Control Association,  1979.  There is a technical
biography of Taguchi which I believe has something on it.

As I tried to pull it out just now, I found it only in Wu, Yu-In &
Moore, Wille Hobbs, _Quality Engineering: Product and Process Design
Optimization_ (Based on the works of Genichi Taguchi), (c) assigned to
Ford Motor Company and American Supplier Institute, 1985.  I'm sure that
ASI could give you complete refs.

Yu-In Wu refs. the Asahi Newspaper, April 17, 1979.  I've never found
the exact ref. for myself, nor have I seen a translation.

Sony had 2 plants, which made the same picture tube.  A key performance
parameter, "color density," had a tolerance range from (coded numbers) 9
to 14, nominal 10.  Tokyo plant had a distribution with +3 - -3 sigma
inside the tolerance, clearly a normal dist.  What we would call a Cpk
ratio of 1.  3/1000 parts outside the tolerance.  Meanwhile, a Sand
Diego plant built the identical tube, assembled into identical sets.
Their distribution showed NO parts outside tolerance, but they used
nearly the whole range.  Now you have to realize that if you see a TV
set by itself, your eye cannot tell whether the color density is near a
10, or 9 without training.  You must put them side by side to tell, and
then you need some help, I believe.  So it is hard to detect consciously
the difference between a TV which is nominal and one near the tolerance
limit.  As it should be.  Which TV sets were preferred by the
customers?  Answer:  The Japanese ones, by a wide margin.  Even though
NO US tubes were out of specification, and 0.3% of the Tokyo sets were
out of spec.

Why?  For one, it turns out that the measure is actually the color
resolution.  Go talk to your local TV station engineer.  It is hard to
get, and they pay extra for high color res monitors.  People can
unconsciously tell something is different with the set, even if they
can't tell you what it is.  This seems to translate into concern for the
set's performance, which means more complaints.

Why?    An easier way to deal with this is to use Taguchi's loss
function to determine that the US made sets have a loss value some 12
times that of the Tokyo made sets.  Now tell the engineers, anything
they can do to reduce the loss value is a good thing, and they will
start to build better tubes.

Now the above is mostly what I found in various Taguchi refs.  The color
resolution I got from a TV engineer I met.  Amazing what you find out
talking to people behind the camera.

But the story isn't over.  In 1988 I was talking at a conference with an
employee of a large high tech firm that used a lot of computers.  He was
in charge of checking out the new models, responsible for many of them.
Some time before he had seen the first models of a new unit with a high
res screen on it.  Clarity was a big issue for this new model.  Sony was
to supply these monitors, with tubes from the US plant (closer to the
market, I suspect).  "It was obvious," he said.  "You set the thing on
your desk, and it looked fuzzy!"  Eventually, the computer maker issued
a letter, saying they would not accept tubes made from the offending
plant.

In 9 or ten years the plant had not learned to make its own product.
Now do you see why I am so frustrated?  I know exactly the logic, the
procedure, to follow for product improvement, yet managers will tell me
that 5% rejects is OK, that a little fuzz is OK, that customers don't
want machines to last over 3 years.  Then when the rejection letter
arrives, we wonder why.  Can you say 'system'?  There are causes for
effects.  We _can_ make improvements.  Engineers don't have to know how
to achieve the improvements.  Certainly managers don't need to know
how.  Manager only have to know that the procedure for getting there
exists.  The engineers always get excited when I and they together work
through to a solution.  We don't need motivation.  We need permission
and support.

Sorry, I got on my soap box again.

While I was digging out the refs for the tale above, I found what may be
the ref for the first 'myth.'  The question went as follows:

Myth #1:  For years I have heard reference to a "famous" study done Ford

(?) on transmissions(?).  One version of the story goes:  Ford and Mazda

were both using an identical (in design) transmission.  However, Ford
had far more complaints about the performance of their transmission than

Mazda reportedly did.  Some at Ford thought the difference was due to
perception; consumers just thought that Mazda had a better transmission
and so they didn't complain as much.  Others thought there might be a
physical difference.  So they obtained several of each transmission and
proceeded to measure various dimensions.  To Ford's surprise, their
transmissions had all measured dimensions in tolerance, however, the
variation took the entire tolerance range in every case.  The Mazda
parts were uniformly distributed very tightly around the target - the
Mazda parts showed very little variation.  Ford thus concluded that it
wasn't simply conformance to tolerances but reduced variation around the

target that caused high quality and happy customers.  (This story is
also promoted as one of the seeds for the six sigma movement)

Although I have heard this story innumerable times, I have never seen
the published study.  I have seen references to references, etc.  My
question is:  Is this just one of those "myths" or "parables" that gets
handed down and around or did this study ever actually occur?  If so,
where was it published?

In Ross, Phillip J.,  Taguchi Techniques for Quality Engineering,
McGraw-Hill Book Company, New York, NY, 1988, p 16.  Ross refs. Ford
Motor Company, 1987.  His relating of it is a little different.  The
product was automatic transmissions, Mazda's were "made more
consistently..."  Apparently Mazda had a stdev about half that of Ford.
Mazda achieved the improvement by use of a more expensive grinder.  The
added cost in mfg was recovered in reduced warranty claims.  Ross
concludes, that thanks to this information, the Ford plant (Batavia)
improved their "quality substantially and, in the first quarter of 1987,
surpassed the Mazda level."

Ross uses this example to do some hypothetical calculations designed to
show how the Loss function works.  Which it does.  It can be well used
to translate the distant, vague customer into dollars (or yen) related
directly to product dimensions and performance.  This makes it much
easier to determine which changes are improvements, and by how much they
improve.

Now, were either of these items precursors to six Sigma?  I doubt it.
Neither involved a product Motorola manufactured.  The person who told
me the tale of the TV's was not in at the inception of 6S, I think.  6S
development preceded the time frames cited here, no?  Most of the 6S
justification is made on piece parts and % rejects, is it not?  The
examples above get below the % rejects, to the distribution, and how
adjusting this controls the % rejects.  Can we say 'system' again?

The concept of variance reduction, which is one of Taguchi's major
points, serves equally well when dressed up in 6S.  Or A2Q, for that
matter (see yours truly web site, address below).  I can easily see that
a presenter of any methodology would like to claim these examples as
sources of inspiration.  So long as no one is fooled or hurt, I suppose
it is OK.  But what we really need to work on is getting down to the
core logic, apply it, and find the solution.  If you buy a program that
is not applied well to your plant and product (i.e., is applied
mechanically), or that your managers do not want to adapt/adopt the
system, you will not get the sizzle you paid for.

Jay
--
Jay Warner
Principal Scientist
Warner Consulting, Inc.
4444 North Green Bay Road
Racine, WI 53404-1216
USA

Ph: (414) 634-9100
FAX: (414) 681-1133
email:  quality@a2q.com
web: http://www.a2q.com

Power to the data!