>All,
>Does anyone have any data on the relationship between solder balls (no
>clean flux w/ N2) and various solder masks?
>Some of our lines run OK and some do not for the same mask. The hard
>part to understand is that a mask will run poorly on one line and OK on
>another. My feeling is that the mask is statistically significant, but
>not necessarily the major contributor. All DOE have shown no sweet spot,
>but of course going to aqueous or eliminating N2 have shown positive
>results.
>The other variable that has completely eliminated solder balls is to
>shorten leads to ~.080". This went over poorly with the connector
>people.
>Any help would be appreciated.
>Steve Joy
>[log in to unmask]

Steve, this was a major problem for us at a former employer. We had zero
tolerance for solder balls greater than 0.005" dia, and a chronic recurring
problem with them showing up after wave solder. This was using RMA "leave
on" wave flux for mixed technology boards (RMA paste, too).

We definitely had cases where severe problems vanished overnight when
trying different soldermask. IMO the soldermask is related, but finding the
causal relation is apparently escaping the best minds in the business.

There was a series of papers in one of the trade mags a couple of years ago
about this problem (four part, I think) which seemed to reach no conclusion
whatever.

Certain curtain coated masks have been called the culprit within my range
of earshot, but I don't know whether to believe this or not.

I saw an unpublished paper about 2 1/2 years ago by Dr. Hugh Cole and Eli
Westerlaken of Cobar. This paper had some of the most interesting ideas I
have seen. Email me your FAX number and I will send it to you.

It said in conclusion:

1) Solder resist - More than likely there are some resists that perform
better than others, but with every resist there is an optimum cure window
and an optimum process window. Find it and USE IT. Do not overcure,
overexpose, or overdevelop!!

2) Hot Air Leveling - The interplay between the resist surface and the hot
air leveling flux is complex. But we will summarize our thoughts in three
words - CLEAN IT OFF. Without a real test for incoming board cleanliness in
your QA acceptance regime, you are playing with fire.

3) Solder Flux - Choose a flux extremely carefully and be certain that it
is properly maintained with the manufacturer's thinner. In addition to the
most important parameters such as high surface insulation resistance, good
wetting, and bright solder joints after soldering (the flux must survive
the wave, shiny joints are an excellent indicator that it has), the flux
should be easy to maintain and should not cause noxious odors in the
workplace.

-------

The above is all a direct quote, from a company in the flux business.

Let me add that I have heard from David Dodgen (with Multicore in Dallas) -
certain assemblers have seen great results from simply dropping the wave
temperature by 10-20 degrees C.

Our best result at my former employer was in switching types of resist at
the PWB fab operation (a captive shop), and the effect was dramatic; but
not completely consistent from lot to lot. IMO this lends credence to the
fact that resist process variables in PWB fab (cure, exposure, development)
can affect solder balls as much as the fundamental chemistry of the
material.

If your data indicate the significance of solder mask variables, you may
wish to work with the fab house(s) to get sample boards which would allow
you to test for the effect of variations in morphology related to changes
within the process used to apply and cure the polymer at fabrication.

You of course need to use a soldering process with the greatest window
during assembly, as well. Good luck, I spent a lot of late nights grappling
with that problem myself. The whole story is too painful to relate. I wish
I had some data to share.


regards,

Jerry Cupples
Interphase Corporation
Dallas, TX
http://www.iphase.com/