Tom
22,8 kV/cm is a fairish old voltage gradient, albeit not excessive provided that it is
homogeneous, and this means chemically homogeneous. Any inclusion of any type or an
area of poorly polymerised resin, for any reason, could be sufficient to tip the
balance for what is obviously punch-through. Could it be that the latest batch or two
of prepreg has a slightly different resin formulation (remembering there are
manufacturing tolerances on prepregs as well)? This may require a different cure cycle
in the press. Does the bare board fab guy run tests on each batch to determine the
optimal cure cycle and apply them in the shop? Do microsections reveal anything (e.g.
low thickness or inclusions)? It is unlikely that a section of an actual punch-through
will reveal much, but may be worth trying, just on the off-chance. At least you would
see whether there were any signs of resin starvation in the surrounding area.
As for the responsibility, who specified the lay-up, the fab guys, your customer or
yourself? Only in the latter case would I feel concerned although, of course, you want
to keep the customer happy. If at all possible, I think I'd wish to increase the layer
thickness or, if not poss, try using three layers of a thinner style.such as 3 times
106 (slightly thicker result but more resin) or 3 times 104 (slightly thinner result
but even more resin and an incresed risk of slippage during pressing).
FWIW
Brian
Tikusis wrote:
> A customer of ours has experienced a very odd failure on 3 assemblies
> and we can not determine the cause for these failures. Perhaps, someone
> has had a similar experience, or can present a plausible explanation.
>
> The failure is a burning pinhole short from the surface layer power
> plane to layer 2 ground plane. The power plane on layer one is about
> 2.5" x 0.75" and is positioned at a corner of the board. The actual
> board size is 7.0" x 4.5." The board has ten layers, 2 ounce copper on
> all layers and 0.060" overall thickness. The assembly runs at 290 volts.
>
> So far, three assemblies have failed in this manner during the
> customer's testing of the assembly. The first two failures were from one
> lot of Rev H, the third failure was from a lot built 3 weeks previous
> and was Rev F.
>
> The dielectric between layer one and two consists of 2 plys of 1080 and
> measures approx. 0.005" thick. Microsections through the pinhole areas
> reveal considerable melting and vaporization of the epoxy, resulting in
> a void between layers, much larger than the pinhole through the surface
> copper.
>
> At first, I thought perhaps a void or hole in the prepreg was the cause.
> But on further thought, this seemed impossible considering the
> likelihood of voids in the prepreg occurring in the same general area on
> three different boards and the fact that this is two plys of prepreg,
> and that any holes or voids in one sheet of prepreg would undoubtedly
> fill in during lamination.
>
> We have supplied the customer with over 12,000 pieces of this part
> number, on over 6 revisions, with no similar failures. Since this
> problem, we have Hi-Pot tested the last two lots (600 boards) at 900
> volts, 10 micro-amps leakage, with no failures.
>
> The customer is convinced that it must be a bare board problem because
> they have made no changes in their processes. The customer is most
> concerned that a board will have this failure in the field, which could
> have very bad consequences. So, it is very important that we determine
> the cause of this problem. Any ideas?
>
> -Tom
>
> ##############################################################
> TechNet Mail List provided as a free service by IPC using LISTSERV 1.8c
> ##############################################################
> To subscribe/unsubscribe, send a message to [log in to unmask] with following text in
> the body:
> To subscribe: SUBSCRIBE TECHNET <your full name>
> To unsubscribe: SIGNOFF TECHNET
> ##############################################################
> Please visit IPC web site (http://www.ipc.org/html/forum.htm) for additional
> information.
> If you need assistance - contact Keach Sasamori at [log in to unmask] or
> 847-509-9700 ext.5315
> ##############################################################
##############################################################
TechNet Mail List provided as a free service by IPC using LISTSERV 1.8c
##############################################################
To subscribe/unsubscribe, send a message to [log in to unmask] with following text in
the body:
To subscribe: SUBSCRIBE TECHNET <your full name>
To unsubscribe: SIGNOFF TECHNET
##############################################################
Please visit IPC web site (http://www.ipc.org/html/forum.htm) for additional
information.
If you need assistance - contact Keach Sasamori at [log in to unmask] or
847-509-9700 ext.5315
##############################################################
|
