Dave,

It is hard for me to determine from your statement whether the topside of
the via is masked ("tented") or not, or whether you are just masking between
the vias and BGA pads over the "dogbone" connection and any other routing
that may be present on the topside between the BGA pads.

I am going to assume that the vias are not "tented", and that when the
solder wicks up thru the via in the wave process that this is where are a
few shorts are occurring.

I am also going to assume that you are using a bed of nails tester that
tester (or even  flying probe type) that wants to probe (contact) its via at
the center.

Firstly, it appears that you may not have the proper clearances in the
solder mask itself, in that it may have too large an opening around the BGA
pad itself, and thus allow for shorting to an adjacent trace or via pad that
is not properly masked. Depending on just what is preheated to what, and
where it is preheated enough to allow the molten solder of the wave to wick
and flow, there could be some capillary wicking that is not only sucking
solder up the hole and allowing it to form a bulge or raised area on the top
pad of the via, but also to possibly to flow down a few conductors on the
top side of the board, which could allow for the shorting at a point that is
exposed by an opening in the mask. I would think that this could happen even
where there was no shorting after only the original installation of the BGA
and subsequent X-Ray. In this scenario, please not that at some specific
temperatures many masking materials become plastic and will allow solder to
flow or bulge along copper underneath the mask. I would think that this
would be especially true with a significant amount of flux present.

Secondly, I would question whether there was enough clearance in your design
between the BGA pad and the pad of the via itself. please remember that the
actual ball of the BGA is not the same size as the pad for the BGA ball. The
BGA ball, which is basically spherical, with flat spots on the top and
bottom where it rests on contact pads, actually overhangs the pads
considerably. If the pads for the BGA ball and the pads for the via are too
large, then when a via sucks up a little too much solder to its top pad by
capillary action, and forms a bulge, that bulge may just be large enough to
bridge to the BGA ball, which is overhanging the BGA pad. This could
especially happen if the balls of the BGA have also actually themselves
reflowed, and especially if there is much movement or vibration or bumpiness
in the conveyor which could cause the BGA itself to move around slightly and
elongate or change the shape of the molten balls enough to cause a short to
an adjacent large solder bump on a via.

Thirdly, I would address the question of reflowing the actual BGA balls
themselves during the wave solder process itself. While this may be a
longshot, how thick are the  "dogbone" traces between the vias and the BGA
pads? The "dogbone" trace between the via and the BGA pads should be thick
enough to carry the current required by the connection to the BGA (too small
a trace could become a "fuse"), but at the same time it should be thin
enough to thermally isolate the pad for the BGA ball. This needs to be
looked at from 2 perspectives: Firstly, the BGA pad needs to be thermally
isolated enough from any planes that it may be directly connected to, so as
to allow the solder to properly flow on the pad to combine with the ball
which must itself reflow (at the time of attachment of the BGA); and
secondly, any soldering of the via, such as is occurring in your wave
soldering, should be thermally isolated to some extent from the from the BGA
ball and pad (by this I mean that a simple pass of the board thru the wave
soldering process should not necessarily reflow all of the solder balls of
all of the BGAs on the top surface of the board). If you have large traces
in the "dogbone" connection between the via and the BGA pad, you will have
very little thermal isolation, and you will force the BGA pad to reflow, and
in the process you may either wick solder away from the BGA ball (most often
the case) or in some instances you may wick solder to the BGA pad, which may
allow the BGA ball to grow (enlarge), thus causing a short.

Fourthly, I would address the question of having too large a solder ball on
the BGA pad itself, which could allow for bridging to any solder bulging
that might occur on the top pad of any adjacent via during wave soldering.
This could occur by having too much solder paste on the BGA pads themselves,
which will enlarge the BGA ball during the initial attachment of the BGA,
causing a larger overhang of the ball in relation to the original BGA pad
(thus reducing clearance to any solder bulge on any adjacent via). This
could also occur by having too large a BGA pad in the first place, which
will subsequently generate a larger ball by having much more solder paste on
the larger pad than is necessary. In this respect, I would question whether
you can actually measure the final size of your BGA solder balls in your
X-Ray photographs to see just how large they actually are in respect to
their original size. What is the original size of the BGA ball versus its
final size? What is the size of your BGA pad versus the manufacturers
recommended pad size? Is it possible you can (or need to) adjust your solder
paste screen to facilitate less solder paste volume on the existing BGA
pads, or possibly reduce the size of the BGA pads themselves?

Fifthly, I would question the size of the vias and the size of the via top
and bottom pads in the area of the BGA. If the pads on the bottom are larger
than necessary, then they will absorb and conduct more heat to the barrel of
the via itself during the preheat and wave soldering processes. If the hole
and barrel of the via is too large, it will conduct more heat, and flux, and
specifically more solder by capillary wicking to the top side of the via. If
the top pad of the via is too large, it will allow the formation of a very
large solder bulge, which could cause solder to flow or bridge to an
adjacent solder ball on the top layer.

Respecting the question of testing and having to make contact with the via
itself, I would pose a few questions: 1.) Do you actually truly need the via
to be "solder plugged" to accomplish this? 2.) Can you change the type of
probe tip that you are using (say from pointed to spherical) so as to not
need the via to be "solder plugged"? 3.) Can you test the by probing the
annular ring of the bottom pad of the via itself as opposed to probing to
the center of the hole in the via itself? As some others have suggested in
response to your questions, can you provide a separate test point pad (or
are you constrained by traces / routing)?

Alternately to the above, if you are in fact having to plug vias so that a
vacuum can be created in your test fixture, where you would in fact be
trying to plug all of the vias on the board so as to be able to cerate a
vacuum to insure contact with your test fixture probes, let me offer a few
suggestions: 1.) Run the bare board (or partially stuffed board) thru the
wave solder to plug all the vias prior to assembly (or assembly of the
remainder of the board (especially the BGAs)) (i.e.: Put the BGA on
separately after the wave solder). 2.) Forget plugging the vias entirely,
and increase the vacuum in your test fixture (get a bigger vacuum pump) so
that you can still maintain the required contact for your probes in spite of
the leakage thru the unplugged vias. 3.) Mask the trouble areas from wave
solder under the BGAs, and live with a little leakage under the BGAs, and
apply some additional mechanical pressure to the top side of the board via
some mechanical method (fixture, foam, springs, your hand, etc.). 4.) If
your shorts are limited to the involvement of one or two specific vias, then
mask those few specific vias from the wave solder. The vacuum test fixture
should still work ok, and the amount of effort to manually mask only a few
via holes should be far less than that of reworking the BGAs. 5.) Once again
in the longshot department, depending on the size of your vias, and the type
of solder mask that you are using, you may be able to mask ("tent") all of
the vias on the top side of the board only (or at least those that you are
having trouble with under the BGA), while leaving the bottom side of the via
open. If the solder mask is not too viscose when applied, and the vias are
not too large, then you may not need to solder plug the vias at all (or at
least in those specific "tented" areas), if you can insure that the mask
does not flow thru the vias to the bottom side of the board. 6.) Have the
board shop specifically plug the all of vias by one of the several methods
available. 7.) If your manufacturing run is small (you mention only 16
boards with problems) maybe you could manually "tent" (solder mask) the top
of the specific vias that are causing you problems before installing the
BGA.

While I know that I have brought up a number of points here that are liable
to provoke all kinds of responses from people in the fourm, both pro and
con, I bring them up simply as food for thought in solving the problem you
describe.

I might add that you seem to be having to go thru a lot of extra garbage and
hell just to satisfy the requirements of your testing equipment. Maybe you
should consider a different kind of testing approach, or if you are actually
stuck with the equipment, maybe it is time to chain or handcuff the rep who
sold you the equipment to the actual testing machine itself, until he solves
the problems for you himself (such as by possibly providing a bigger vacuum
pump (as mentioned above)), so that you can live with a few holes that are
not plugged in certain areas such as under the BGAs.

Hopefully this will provide a few things that you can look at, and possibly
a few answers to your problem.

JaMi

* * * * *

----- Original Message -----
From: "Dave Chapman" <[log in to unmask]>
To: <[log in to unmask]>
Sent: Monday, April 28, 2003 7:02 AM
Subject: Re: [TN] BGA


> Happy Monday, I have a BGA 256 Tundra which has no shorts in SMT (we
x-rayed
> all 16 boards) When it goes over the wave we get a short or two. Please
> assume the wave is in control.
> We mask the topside of the BGA because there are a lot of vias which when
> they test it would leak and not get pulled down good in the fixture. I
need
> the vias on the bottomside under the BGA to be filled with solder for
> testing so I can't mask the bottom. What should I do?
>
> Dave Chapman
> Manufacturing Engineer
> Circuit Service Inc.

---------------------------------------------------
Technet Mail List provided as a free service by IPC using LISTSERV 1.8e
To unsubscribe, send a message to [log in to unmask] with following text in
the BODY (NOT the subject field): SIGNOFF Technet
To temporarily halt or (re-start) delivery of Technet send e-mail to [log in to unmask]: SET Technet NOMAIL or (MAIL)
To receive ONE mailing per day of all the posts: send e-mail to [log in to unmask]: SET Technet Digest
Search the archives of previous posts at: http://listserv.ipc.org/archives
Please visit IPC web site http://www.ipc.org/html/forum.htm for additional
information, or contact Keach Sasamori at [log in to unmask] or 847-509-9700 ext.5315
-----------------------------------------------------