We have been wavesoldering rigidflex motherboards for seven years.  They have
proven tricky, to say the least.  The most common failure is a phenomena we
call fiber crystalization.  It does not appear as a measle, craze, or delam,
but as a single white fiber 0.003 wide or so, running for 0.030 to several
inches long, through the pattern of the drilled holes for a daughterboard
connector or across the pattern of round I/O connectors. The lines do not
appear in undrilled areas, and not in low density areas.   Our recipe calls
for a 3-4 hour thermal pre-conditioning at 225 F.  Unlike a bake, we must
follow the pre-condtion within four hours with the soldering operation, even
if the pwb is stored in a nitrogen chamber.  Wavesoldering is done on 490-500
F solder pots with 200-220 F topside temp.  Current contract specification is
MIL-STD-2000A.
 The questions are:
a.  Has anyone seen this before?
b.  Has it been examined to determine the mechanism, such as rosin
starvation?  Would your explanation fit with our preconditioning?
c.  Has anyone tested it to determine the likelyhood that this propogates
into dielectric breakdown or other sources of field failures?
d.  Is there something my pwb supplier should be doing differently?
e.  Is there a better proven receipe you are willing to share?
Any and all help is much appreciated and will be reciprocated, Thanks again.
   Steve

Steve Mikell ,   Soldering & Cleaning Shaman     SCI Systems, Plant 13
Huntsville, AL
e-mail  [log in to unmask]