Hi Edward,

You are correct in that you must take the cost of engineering and test
materials before making the decision to go ahead with a new process.  And
that isn't always easy to determine up front.  Perhaps the best way of
looking at this is on a case by case basis.

By that I mean, if you are going to install an additional finish, say for
example immersion tin (a relatively new product on the market) there are
certain parameters and "output" criteria it has to meet.  It must allow for
some measure of storage and still maintain its solderability, it must allow
for so many thermal cycles and still maintain solderability, etc.  Once it is
determined what those parameters are it's a matter of running through
numerous mental exercises and models in order to project a cost of test
materials, time, engineering and other shop labor costs, outside test costs,
etc.  It's a good idea to have a formal program that forces you to address
ALL these issues and more such as cost of sending back chemicals that this
new stuff may be replacing, discarding old tanks and supplies, training new
operators, writing SOP's, among other things.

Now looking at something on the order of direct plate the picture gets
fuzzier.  For several reasons.  First, unlike the example above, direct plate
has no means for measuring "output".  Immersion tin does.  You can put
systems in place to test the product right off the line - tests that prove
with no doubt how the process is performing.  Direct plate doesn't allow for
that.  Sure, you can measure resistance with an Ohm meter - and that really
doeasn't prove squat.  Ask any direct plate supplier and they'll tell you the
same thing - resistivity readings are "indicitave" only.  Or maybe you can
panel plate (flash) test panels and perform backlights - better but still
takes an hour or so.  But even that requires you to produce panels similar to
the boards you are actually producing.  Testing double sided coupons when
you're building 10 layer blind/buried vias with liquid dielectric doesn't cut
it.  The coupons must be similar to product you're building.

So, installing something that has no sure method of measuring output is quite
hard and should be heavily engineered at the start.  You know, look around
your shop for processes you can measure output performance on.  Drill,
copper, soldermask, legend, dry film, lamination, oxide, Ni/Au, etch - all of
these areas are easily measured.  There are systems you can put in place that
lets you determine with no doubt the quality of the product off the line.
Now, look for areas that arean't so easy - like desmear, oxide replacements,
direct plate.  There are no good ways of measuring performance off these
lines.  There are indication methods.  There are ways of measuring process
parameters, sure.  But you need to measure OUTPUT.  So, engineering these
processes are quite difficult.

Edward, to some degree you must trust your supplier.  It would be nice to put
all you faith in them but of course you can not.  The case study I told about
in my report was a good example of trusting too much.  You know, had this
place survived there would likely have been a lawsuit against the supplier or
else large refunds made by the supplier.  Win, lose or draw there are
arguments for both sides.  The supplier has dozens of installations and most
of them have no problems.  But we did.  On certain product EVERYONE will.
There was no data to alert any of us to the potential problem.  We learned
the hard way.

Bottom line?  Try to put processs in that have a fool proof method of
measuring OUTPUT.  When a vendor walks in and tries to sell you a process
that doesn't have such a method be scared - very, very scared.  You know,
there are dozens of installations of oxide alternatives and many more going
in as you read this.  It's the new wave of "replacements".  You know how we
measure performance?  We LOOK at the panels for color or hue.  We measure
etch rate.  We might laminate coupons daily and do peel strength.  What's the
problem then?  This stuff works by not only etching, but it also applies an
"adhesion promoter".  Can't measure that - can't measure its weight gain -
can't measure its effectiveness.  Etch rate doesn't tell the whole picture -
the window for weight loss is HUGE.  And even if it weren't, etch loss is not
the mechanism that allows it to work.  Laminating coupons takes several
hours.  And many shops have stopped baking product as it exits, like you once
did with oxide.  So, now there are millions of boards being built that employ
this new technology and we've stopped moisture baking (cause the suppliers
have dozens of installations that don't - AND IT WORKS).  In my mind that's
scary - very scary.

I expect to get dozens of responses here telling me all about how this new
technology works well and how you've come up with a way that's foolproof.
Have at it - I'm all ears.

Edward, good luck in future engineering endeavors.  Make your shop
"engineering" oriented rather than "production" oriented.  If you do you
might manage to avoid costly errors.

Sincerely,

Mark Mazzoli

##############################################################
TechNet Mail List provided as a free service by IPC using LISTSERV 1.8c
##############################################################
To subscribe/unsubscribe, send a message to [log in to unmask] with following text in
the body:
To subscribe:   SUBSCRIBE TECHNET <your full name>
To unsubscribe:   SIGNOFF TECHNET
##############################################################
Please visit IPC web site (http://www.ipc.org/html/forum.htm) for additional
information.
If you need assistance - contact Gayatri Sardeshpande at [log in to unmask] or
847-509-9700 ext.5365
##############################################################