MR NORMAN S EINARSON

>Response to Jerry Cupples:

>The following
>should summarize most of this discussion.
>
>This discussion began with the following statement from Mike Cussen:
>
>"I ran into a delamination problem with an individual supplier on (2)
>different multilayer boards (FR-4) after the IR process in our SMT
>assembly.
>
>My question is:  "Is it necessary to bake the blank boards prior to
>the SMT process and if so, what are some of the guidelines I may use.

>Response to Jerry:
>
>We both agree that baking boards diminishes solderability.  We also
>agree that delamination problems are seldom an issue when not baking.
> We both agree that today's boards should not require baking prior to
>any soldering operation to prevent delamination.  However, we are at
>issue on liability and who is responsible when not baking.

Yes. I do say that if I did have problems with delamination, I would insist
that the baking I do (or not) is irrelevant, and for technical and
practical reasons, the functional equivalent of beating drums to keep
elephants from stampeding your office. Very unlikely to happen regardless
of the type of drum you have or the rythym of the beat. False logic may be
used to defend the practice.

>You stated that when a vendor insists on baking prior to IR reflow,
>you would no longer do business with that vendor.  Well Jerry, your
>present vendor (first-class people represented here) has nothing to
>worry about because you already bake your boards at 110 degrees "C"
>for a minimum of (4) hours (and usually overnight).  I only refer to
>your vendor as "first-class" as this is how you stated it.  You infer
>that anyone that insists that you bake may not be first-class.

No, Norm, you inferred that, I may have implied it, but the intended
meaning is that if I had problems with delamination, and the vendor told me
that I should be baking the boards to prevent it --> therefore I was
responsible; I would question that vendor's veracity. We use 2-3 vendors
for PWB's and in my mind they are quite competent. It is partially my
responsibility to qualify these sources, and mostly my headache when we
receive bad product. Hadco and Zycon are our current primary sources. IMO,
they are near the top of the heap in technology, quality and any other
measureable rating factors I'm aware of. Since 1978 I have participated in
buying and assembling PWB's from about 20 sources, and feel qualified to
comment on the relative abilities of vendors. Forgive me if this sounds
boastful, and it is not an attempt to downgrade any other suppliers, but
you seem to repeat several times my comment about the "class" of source we
use. It is also true that neither verbally nor in any other way has a
request, suggestion, or requirement been given to us that we bake our
boards prior to reflow.

>Though I agree baking may not be necessary, it is an added insurance
>factor against delamination.  When boards go through the HASL process
>at your vendors plant, they are submerged in 465 degree "F" solder
>for 5 to 6 seconds.  This is an excellent thermal stress (not thermal
>shock) test.  If delamination was going to happen, it should show up
>here.  However, it must be remembered that vendor always bakes the
>boards just prior to the HASL process.

I doubt this is "always" done, at least not the kind of slow bake with
individual units separated as you describe later. In fact I am positive I
have seen boards go into a vertical HASL machine with no baking. Of course,
they effectively bake the heck out of them in the laminating press. The
matter of thermal stress imposed at HASL is something I pointed out.

>When you mass solder your
>boards in an IR oven, they are in the tunnel for a lot longer than 5
>to 6 seconds.  Therefore, when a vendor requests a bake cycle before
>your IR, just as they have done before HASL, I don't feel this to be
>an unreasonable requirement.  Of course, no one can force you to bake.
>  However, who does the liability belong to?

I contend the thermal stress is greater when a board is being submersed in
liquid solder than it is for IR/convection reflow with appropriate preheat
profile, even considering the somewhat longer exposure in a furnace.You
specify "when they request...", but I say this is uncommon, it has never
been asked of me. If it were asked of me now, I would resist. It may appear
as a recommendation in an IPC document but that imposes nothing on my
practices.

>When you bake boards, 10 - 20 in a stack, it must be assumed that
>baking is done before the assembly of components.

I don't think it must, but you could assume this correctly.

>This infers that
>there is a reasonable time frame between baking and soldering.  This
>also infers that the only accomplishment in this case would be
>diminishment of solderability.  When baking boards in a stack, it
>would require considerable for the boards in the middle of the stack
>to come up to temperature.  In addition, where does the moisture go
>when stacking? To the next board in the Stack???  Whenever baking
>boards, they must be separated to allow them to rapidly come up to
>temperature, with free air flow around each individual board, to
>allow any moisture to dissipate.  This only makes good common sense.
>When baking populated, or unpopulated, they must always be separated
>when in the oven.

I disagree. It is true that the separation is better, and convective
heating occurs faster, but if they are stacked the moisture will still
evaporate and vapor can move between the boards and out of the laminate.
Another board in close proximity is no vapor barrier. However, I point out
that doing a bake on stuffed boards in carriers or racks would be highly
impractical and I have never seen it done anywhere.

>Baking will only serve its intended purpose, if done within 3 to 4
>hours just prior to any soldering process.  Otherwise, it is a wasted
>effort.

I have read the number is more like 12, but I agree that it should be done
soon if it is to completely dessicate the laminate.

>All board material has very high moisture absorbtion
>properties - some more than others.  Let's assume that you bake all
>of the moisture out of your boards and then return them to your store
>room or assembly area.  All of the moisture in the specific
>environment will be absorbed back into the board material in much
>less than 24 hours.

Perhaps so. The boards will reach equilibrium with the water vapor in the
air. They do not continue to adsorb any moisture after equilibrium is
reached.

If your environment is 30% RH, then this will
>represent the moisture factor.  If it is 70% RH, then this would be
>the moisture factor.  If you store them in a bucket of water, this
>would be the moisture factor, etc., etc., etc..  A simple weight loss
>test can be performed to determine exact moisture factors.

Not sure of your point, but we do have humidity control in our SMT assembly
area, and try to maintain about 45% +/- 10%. This is of some significance,
and a good practice, IMO. On the other hand we deflux with a double hot
water recirculated pressure spray and rinse. They are grossly dried with a
turbine blower feeding an air knife, and allowed to set in racks in the
room environment for several hours. On two-sided boards, we print paste on
side two, pick and place, and IR reflow once more with no bake. If moisture
absorption was the death factor some seem to believe, we might see some
delamination...

>You stated in your earlier response that you have an agreement with
>your first-class vendor to eat your components when failures occur.

What I said was that if we receive defective PWB's, the vendors will
reimburse Interphase for the value of the material lost when we assemble
components to the defective product. We do have such an agreement.

>In today's business climate, many vendors do this either to win your
>business or hold your business.  You stated that you have only
>returned one or two boards about six times over the past few years.
>If this is the case, your present vendor has made a pretty good
>business decision.  Not too many dollars lost!  However, though the
>vendor is eating a limited number of boards with components, what do
>you think the response would be with 100 - 200 boards, six times over
>the past few years. My money would go on them dropping you as a
>customer after the 2nd or 3rd time.

Bad bet, Norm. We would disqualify them, and move our business to another
vendor. We make a high end product. Components are much harder to buy right
now than PWB's. It is common for us to have a $20 board with 4 or more $40
ASIC devices which have 26 week lead times, SRAM and so forth that you
_can't_ buy at any price. The value of the components is significant, but
we can't afford to lose a board even if we get full reimbursement for the
material. Our customers want their orders on time, too. Every assembly we
make is pretty much a sold product we can't ship. We have little motivation
to stiff our PWB sources.

>Just wait until the industry
>gets busier and deliveries get extended to 12 to 16 weeks (and it's
>coming).

I agree that now is not the best time to exercise strong arm tactics with
any good sources. We have about three excellent ones who still want our
business enough to give us good service. I think the effect might be seen
in pricing as well, depending on how the supplier rates the risk of
offering such warranty terms. Our vendors knew the terms when we were
quoted and before orders were placed.

BTW, just today I spoke with someone from Merix who called me after seeing
a posting I made here. He wanted to see if we would like to discuss the
need for additional PWB sources. He confirmed that Merix also has
agreements to reimburse customers for the value of components when PWB's
are later found to be defective, and that they do not require any baking
prior to reflow. Based on what I hear, Merix is an "up and comer" in the
trade. The willingness of a supplier to offer such warranties is probably
related to confidence in their own abilities.

>My money must again go to no one eating your components.
>After all, why should they?  It requires over 100 processes to
>manufacture a multilayer board.  Though an attempts is made to
>control all of the processes, by all first-class vendors, from time to
>time.  There is absolutely no way that a vendor
>can 100% guarantee the quality, 100% of the time.  Even with
>destructive testing on every board, there is no 100% guarantee.
>Therefore, shared liability.  The bare-board to the vendor,... the
>components to the user.

Now here I truly disagree with you. We _do_ expect a 100% acceptance level
on PWB's, and can tell you it within sight. "#%&@ does happen" just won't
illicit any sympathy from me. When I worked at Fujitsu, we got PWB's mostly
from the Fujitsu Limited Kanuma and Nogano (Japanese) facilities. We
assembled thousands of boards per day. I saw _one_ defective board (by my
own definition, not theirs) in about three years. I sent it back to Japan,
and got an almost embarrassing letter of apology. The better US suppliers
are drastically improving historic quality levels to world standards. I
certainly expect a 100% _guarantee_ of quality if not 100% perfect boards.
And delamination of MLB's is IMO a gross defect. By the way we did no
pre-baking _whatever_ at Fujitsu.

>You suggested that the delamination problem may have been the
>material.  Why then would the vendor not go back to the material
>supplier and make them pay for your loss, plus the cost of the board?
> Becausethe laminate supplier will only give a credit for the cost of
>raw material and nothing else!

Perhaps true. We get reimbursement for our materials cost only. The
difference is that we have a much higher added value and the PWB's are a
custom fabrication for us, not a commodity. What arrangements a PWB
fabricator has with their laminate supplier is a separate matter.

>I would like to know what the term partnership truly means.  It's a
>nice term and sounds pretty good.  We all would like to think that we
>are good partners.  Watch the partnership vanish when the quote
>reaches your purchasing department.  Does your agreement with your
>vendor to eat components include a partnership?

It is a warm fuzzy term. I think it means we agree to cooperate and make a
commitment over a significant time period to do business in a mutually
satisfactory way. We agree to share certain information regarding
schedules, etc. and promise not to simply shop our business around with
each new PO issued. It is really a matter of trust. We trust them to give
us top quality and fixed competitive prices, they trust we will continue to
give them orders, and that their efforts to supply good product will aid
our success and reward them with growth in their business. In the case of
our Company, we do about $50 million sales of almost exclusively
multi-layer board based product, and the PWB's are pretty complex (read
that expensive). We do business with large sources who have the resources
to build difficult product and meet our needs. I do not believe they
consider it a risk outside their control, nor one that is financially
dangerous. They trust we would not abuse this arrangement by attempting to
collect money over a trivial defect. I have signed lots of MRB forms "use
as-is" in my day when minor or cosmetic defects could have justified a
return. You must choose your partners carefully if you intend to compete in
this business.

>Is this agreement
>with your top name vendor, with impeccable abilities (the way you put
>it) in writing?  It should be!  If it is in writing, I sure would
>like to see a copy.  Does this vendor give all of his customers the
>same agreement, or do they only consider some of their customers good
>enough for this agreement?  Those that don't have the same agreement
>must suffer the burden of the loss of your components.

In one case I participated in a qualification where I saw it in writing.
Two vendors I'm aware of have reimbursed Interphase for the material value
of assemblies in cases where clear defects were found after assembly. Other
board users I know have similar agreements. We specifically asked for these
same terms during negotiations for qualifying domestic PWB sources at
Fujitsu in 1992, and I heard three well known sources agree to such terms.
These were all large and well known fabricators.

Probably many PWB vendors could not afford the risk of loss associated with
making such a warranty. Some would consider this only for key accounts.
Some would have too much concern that they could not rely on their own
process control capabilities.

>With all of this in mind, what is wrong with a simple bake cycle when
>it is requested or insisted upon?  This is nothing more than what the
>vendor does in their own process prior to HASL.  I'm sure that the
>laminate supplier imposes the same requirement on your vendor.   At a
>minimum, it is an added insurance policy to help prevent delamination.
>  Though we both agree that baking is not necessary, it is only I
>that would be willing to to accept the consequences for not following
>the vendors requirements.

You agree that it is not necessary. It has known deleterious effects on
solderability and promotes the growth of CuSn intermetallics. In the case
of OSP coatings, it is probably seriously harmful, and such coatings are
becoming increasingly popular. It is not so simple to have to bring out the
PWB's from the stockroom the day before they are to be assembled and do
this bake/cooldown. Any process that takes 8-12 hours has an impact on the
total production cycle. Sometimes we change our schedules based on
component availability and have to issue a new kit to the line.

Now if I were offered a job at $150K per year requiring me to wear a pink
shirt to work every day, I might do so. I am not too proud, but if I have
my choice I will wear what I choose, sir. Someone may tell me pink shirts
set off my eyes, but I still don't like them.

>As you stated, baking boards is not a cure-all.  However, it is an
>insurance policy agianst delamination of a product that must face
>extreme temperature during processing.  It also eliminates a contest
>that may arrive from delemaination from vendors requiring baking.
>And it's not a big deal!

It would be a big deal for any high volume assembler, and is a noticeable
effort even for a medium volume producer, especially if the baking were
done as you suggest may be adviseable. Boards routed out of their panels
are difficult to handle and rack, and an oven big enough to hold racks for
many hundreds is a monster. I say many people have abandoned the practice.
We may do so as well. To me there is no waste like the cost of insurance
against something you can afford or which is unlikely to happen. I am
suspicious of insurance salesmen, and I'm a hard headed person when it
comes to being guided by my own experience.

>Sorry that this was so lengthy.  This would be an excellent subject
>for a debate.  This is my last response on this subject.

Seems to me this was sort of a debate. I appreciate your civility, and
while we don't fully agree there is always an argument to be made on such a
subject. I apologize for any offensive tone I may have used, and for taking
the last word, I guess - but you raised numerous questions on what I
thought was more than a rhetorical level.

I hope anyone who reads this will do what _they_ feel is best, and take my
comments at face value. Your mileage may vary, void where prohibited, the
opinions are my own, etc. etc.

regards,

Jerry Cupples
Interphase Corporation
http://www.iphase.com