TECHNET Archives

1995

TechNet@IPC.ORG

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
[log in to unmask] (Jerry Cupples)
Date:
Thu, 12 Oct 1995 16:05:30 -0500
Content-Type:
text/plain
Parts/Attachments:
text/plain (172 lines)
Mr. Norman S. Einarson said:

>In answer to:   David Bergman
>                      Jon Holmen
>                      Mike Cussen
>                      Jerry Cupples
>
>When reading the notes by the above, I felt it necessary to respond.

I think Mr. Einarson refers primarily to me....

>David Berman's response includes mostly the bare board process of
>baking, with some pre-assembly baking requirements.
>I must respond to to some of the comments and also give an opinion.
>
>First, why do people bake boards prior to assembly?  Baking may may,
>or may not, help solder joint pinholes, depending if there are
>plating voids present.  Solder joint blow holes are almost always
>caused by a wetting problem with the board or components.  Therefore,
>baking is a waste of time for these reasons.  Further, there is
>nothing that will diminish solderability faster than a bake cycle of
>the boards prior to soldering.
>
>Some people will bake boards to help prevent delamination.  However,
>when is the last time you saw either measles or delamination?  Not
>very often, that's for sure.  The process of board manufacture, in
>parallel with bare board materials has improved significantly over
>the years.  Then why a bake cycle prior to a soldering process.
>Because of the statement Jerry Cupples made!  If the boards
>delaminate, without first seeing a bake cycle, send them bake to the
>vendor and make him eat the components also.  He said they won't
>quibble.

We _do_ bake them, though I don't believe it is of real benefit. I
virtually never see boards delaminate, with or without benefit of a prebake
prior to reflow. We often send old solder sample boards or scrap product
back through the furnace here for profiling - sometimes that old junk board
may go through 4-5 passes with no bakeout, and IMO delamination is rare for
normal FR-4 MLB's going through a wave solder machine or an IR furnace at
typical process temps. We used to get a solderablility test sample with
every lot, which we would wave solder without any prebake. I have never
seen one delaminate at Interphase, based on probably 200 tests in the last
2 1/2 years. Consider this - That 14 layer board (and others) is a dual
reflow 2 side SMT design, and most of the time the board goes through a
water wash defluxer before it gets second side assembly. Within hours of
being bathed in a 140 F hot water spray for about 5 minutes, it goes back
into the IR furnace, certainly with no bakeout. I've had  0/~5,000 delam
rate on this unit; undoubtedly the most susceptible one we assemble.

>Let's look at that statement.  You chose the material to use, not
>your vendor.  You could have used a more expensive polyimide material
>and added another 30% to the price of your boards.

That would be highly unreasonable in a competitive market. Polyimides also
have adhesion problems! We specify conventional FR-4 like about 98% of all
PWB users, and we have MLB designs in regular production up to 14 layer,
and very commonly 8 layer. The stackups are generally left to the PWB fab
house, i.e. prepreg and core or foil as _they_ may choose to meet our final
thickness which is almost always good old 0.062". In my experience, it is
not typical to see an exact callout for layer construction, and is normally
left to the discretion of the fab house.

>Though it's not
>an industry practice to eat the cost of components, your vendor may
>possibly eat this cost.  However, this solely depends on how good of
>a customer you are, the total value of the loss and if they choose to
>continue doing business with you.  Many cases have been lost in court
>on the subject.

We have an agreement with our (first-class people represented here) PWB
sources to reimburse us for our loss of components due to a defect in their
fabs. We have not done this often. This is fairly common with board sources
in my experience. I have sent onsey-twosey completed units with confirmed
internal shorts or opens (found after assembly in our ICT) back for credit
- including components worth 5-10 times the PWB - prehaps 6 times these
past few years. They have generally not paid much to provide us this
assurance which we value highly. They have also never "argued" with my
opinions, i.e. I would not send them back were I unsure of the fault, and
this does not occur with regularity.

>1) Why should a vendor eat this cost?  2)You have a QC department to
>ensure that there will be no delamination.  3)You chose the materials.
>The process to manufacture boards varies very little from one vendor
>to another.

1) Only if they provided me with defective product.

2) Our "QC" department has no means of evaluating the internal layer
adhesion strength of a board, nor do we electrically test bare boards
(although we pay our vendor to do so). Routine thermal stress testing at
receiving inspection may be adviseable, although we no longer do this. For
a conventional HASL finish, the thermal shock the panel sees at fab gives
some assurance that the inherent cohesion/adhesion is at least OK. IMO,
this is a reason that delamination is rarely seen by the time I get a
board.

3) For lamination materials, I think not. Some fabricators use adhesion
promotors, some use vacuum lamination, some use black or brown oxide (I
have heard violent disagreemnet about this); bake and cooling cycle times
do change. You imply it is a function of our material specification as a
variable. I contend this is more a constant - simple FR-4 - than the
process choices which will be made by a fab house. Even the
_susceptibility_ to moisture is a function of the lamination process, IMO.

>Why should the vendor eat the cost of the components?
>They shouldn't!  They are only responsible to replace the cost of the
>bare board.....Period.  Therefore, when a vendor asked you how and
>when you baked your boards, you must tell them.

Virtually every circuit board made has an implied warranty called
suitability for intended purpose. Consequential damages may or may not be
excluded from such a warranty. In our case this is a matter of terms of
purchase by prior agreement. Again, we have chosen a top name fabricator
with impeccable abilities, who can accept such a risk with little concern.

>When a vendor
>insists that you bake your boards prior to a soldering process, then
>it is a must.  Though delamination is not much of a problem today, it
>is imperative to follow the vendors instructions or all resulting
>quality issues must be your sole responsibility.

If a vendor were to insist on such a process for the relatively common type
of board we specify, this vendor would not do business with my Company.

>Though I do not like any type of bake cycle for solderability issues,
>and further don't think that it is necessary, you must include baking
>if it is a vendor requirement.  The higher your layer count, the more
>expensive the board and the more necessary baking becomes.  All
>baking must be accomplished within 3 to 4 hours just prior to any
>soldering process.  Though it is difficult to remove all moisture
>from your multi layer boards, a 250 degree "F" bake for 3 to 4 hours
>will remove most of the moisture, thus, reducing the possiblility of
>delamination.  Some people bake all of their boards overnight.  The
>longer you bake, the more you lose solderability.  Everyone seems to
>have a different criteria.  People with extended bake cycles must
>accept all non-solderable conditions as part of their process.

Again, the real control is to ensure your storage conditions do not expose
the boards to excessive humidity (greater than about 60%) My contention is
that baking is generally a weak cure for several rather rare ailments. It
doubtlessly has an adverse effect on solderability and promotes oxide and
intermetallic formation. This is not to say that it has _no_ value, rather
that it is ineffectual in prevention of delamination. If we believed we had
moisture problems, we would store them in a dessicator with a lot of
Drierite.

I contend that the dessication afforded by baking is _not_ a cure for weak
adhesion, or separation due to oxide breakdowns. If one (of numerous)
contibuting factors is moisture within the lamination, it might help. My
experience leads me to believe that this is not common, nor is it a primary
cause for the precious few delamination cases which I have experienced
(_none_ in recent years).

>From the original post, I infer there was a materials problem. It is
possible that there is a process problem, or that the design (though he
said FR-4) is unconventional. What I implied was that baking is not a
cure-all, and that I suspect first a defect in the board. I may be wrong.
If this were happening here, my experience would trigger a quick FAX to the
fab house and an MRB form with my signature.


Jerry Cupples
Interphase Corporation
Dallas, TX
[log in to unmask]
http://www.iphase.com

These opinions strictly my own.




ATOM RSS1 RSS2