Dan,

Thank you for the nourishing additional mental morsels!  I agree with all
you say and we are doing much of this.  Where I hadn't 'gelled' my position
was with regard to the last point:

At what point/product type/product complexity does expecting the supplier to
provide 100% good parts fail or become unreasonable?
"As you might surmise from the previous comments, this is a very complex
question with the cost of detecting defective material and the cost of
shipping defective material becoming part of the answer.  This means that
the answer is individualized for each situation."

I have been working up different scenarios with respect to cost of returns,
cost of inspection, acceptable risk, etc.  I appreciate the discussion and
"sanity check" it provided me.

My thanks to all who commented on this thread ... including Earl!  Sometimes
this list is almost as interesting as my dog list!
Jana Carraway

  -----Original Message-----
  From: TechNet [mailto:[log in to unmask]]On Behalf Of Don Vischulis
  Sent: Thursday, January 03, 2002 4:55 PM
  To: [log in to unmask]
  Subject: Re: [TN] Final Audit/Inspection


  Dear Jana,

  I've been lurking long enough.  Just have to put my 2 cents in.

  What if:
  you are not in the commercial market, and
  there are few companies capable of manufacturing your designs, and
  the design specifications are required for the end product, and
  it's a specialty product, so it is not IBM or Cisco volumes, hence your
  vendor pool is small, and
  the supplier has not been able to meet the yields they initially projected
  for such designs, and
  then on top of all that...it's flex...
  Without regard to the board type - if you're trying to work with the
supplier (DFM/CE - partnership, etc.), you should sit down with them and
determine why the yields are below expectations.  Based on that review there
may be several outcomes.  Those outcomes include requests for design
changes, supplier process improvements, and/or an acceptance of the lower
yields with a possible price concession (if you want them to build more).
It's essential that one is aware of industry wide process capabilities
(benchmarking) for that product.  There is competition in the low
volume/tight tolerance marketplace as well as the high volume market, and
all suppliers do not have identical capabilities.

  A design can only be tweaked (real word? spelling?) or DFM'd within the
  required technology design limits.  If we designed for 100% yield, we
could
  not build the products.  I think that's real life for many people.  We
  accept, and the manufacturer accepts, that the product yield will be
greatly
  less than 100%.  Okay, that's life with these products.
  This is where the Moonman is coming from with DFM/CE.  The design process
is iterative.  That is, you work with your supplier, built a prototype,
evaluate the results, and make adjustments.  All parties must work to
understand the root causes for any problems.  If a design is pushing the
capability envelope, concessions must be made.

  So, then we go back to the basic question(s), what is reasonable to expect
  at incoming - a)100% good product in the door, b) perform an AQL, c)
perform
  100% inspection, or d)other...?
  From a quality perspective, the answer depends on what the end user finds
acceptable.  You can expect 100% good product at the door but somebody has
to take steps to ensure that it really is 100% - even 6 sigma quality has a
few parts per billion defective.  If you can't accept any defects, sampling
inspection is not acceptable.  You must 100% inspect/test the incoming
product.  Remember that AQL is the acronym for Acceptable Quality Level - by
definition the acceptable percentage of incoming defective product that is
acceptable to the production floor.  Formulas exist to estimate the
percentage of defective product that the manufacturing floor will see given
the percent of defectives shipped from the supplier and the inspection plan
used at incoming inspection.

  Perhaps something you've been saying has more impact than I realize, as I
  think about it - contract negotiations.  If the supplier signs up to meet
  design/product criteria and doesn't, my options are:
  a)change suppliers,
  b)if a is not possible, negotiate incremental improvements with the goal
of
  meeting the specifications?
  IMO one must do one or the other (unless the corporation enjoys
suffering).  The process where one determines the best practices available
is called benchmarking .  Once the best capabilities available are
determined, the choices are living with the status quo, working with the
current supplier base to improve capabilities, or changing suppliers.  The
caveat is that perfection does not exist - the lowest absolute purchase
price (not total cost of ownership) and the most capable process do not
often reside within the same walls.  When they do, the world beats a path to
their door.

  At what point/product type/product complexity does expecting the supplier
to
  provide 100% good parts fail or become unreasonable?
  As you might surmise from the previous comments, this is a very complex
question with the cost of detecting defective material and the cost of
shipping defective material becoming part of the answer.  This means that
the answer is individualized for each situation.

  I hope that these additional mental morsels provide nourishment for you.
(Translated - I hope that this gives you some food for thought.)

  Don Vischulis
  [log in to unmask]