Charles Barker@IO-US
10/17/97 10:13 AM

Thanks to all who shared their knowledge and experience!

1.  DO YOU KNOW OF A SOURCE FOR DUMMY PARTS OTHER THAN TOPLINE?

a)   Soldering Tech Int'l in Madison, Ala. 205-461-9191, FAX 205-461-9566.
b)   We had better luck asking for mechanical samples from our supplier.
c)   Watch out that dummy BGAs have size / presence of the die in the BGA
same as real thing. Otherwise, all bets are off for soldering reliability.
d)   We typically have to purchase parts (either empty packages or
electrical fall-out) right from the component supplier.

2. COMMENTS ABOUT OSP PAD FINISHES? WILL THEY TOLERATE ALL THE PROCESS
STEPS?

a)   HASL works great. BGA will probably lengthen the useful life of HASL
process.
b)   HASL gave us a bad experience at the beginning. We use OSP now.
c)   OSP may give you problems in harsh environments.

"We went exclusively to OSP finishes that can tolerate multiple reflow
operations after having disastrous results with HASL finishes with CBGA
devices early in our work. HASL finishes might have improved since then,
but we have no reason to try them again.  We don't have huge quantities of
boards in production, so assembly tends to not stretch out that long, and
we haven't really seen problems with the OSPs in such applications."

"We have made some tests of OSP on one of our products (backplane module).
We tested the boards in a harsh environment (Batell 3) and none of the
boards passed the inspection due to corrosion. This board didn't have any
BGA but I wouldn't recommend OSP on any board in an uncontrolled
environment."

3.  WHAT ABOUT OTHER PAD FINISHES? PALLADIUM, GOLD, PLAIN OLD HASL?

a)   Couldn't see anything to be gained to justify the cost of using Gold.
b)   I am against OSP. Instead of OSP rather go for Electroless
Nickel-Immersion Gold. Control the Gold thickness between 4-8 micro inch.
Suggested via size 0.013" + 0.000", -0.008" for BGA and ask for one pad to
one via ratio. I have seen problem by not having that ratio and had to put
extremely difficult/costly jumper wire.
c)   Looked at e. Palladium and i. Gold, and they both did fairly well in
assembly trials.  Soldermask compatibility with the aggressive e. Nickel
bath that precedes the final finishes wasn't all that great.  There were
some combinations of masks and bath types that had some problems.  We may
in the future go to these finishes as these problems are worked out.

4.  WHAT STENCIL THICKNESS FOR BGA? DO YOU FIND .006" SUFFICIENT? HAVE YOU
EXPERIENCED BRIDGING WITH .008"?

a)   No problems with .008" think stencils
b)   BGA is pretty tolerant on paste thicknesses.
c)   We use .008, have used .010" without bridging on components with sn10
solder balls.
d)   We have used 5 and 6 mil thick stencil.

5.  DID YOU FIND THAT YOUR OVEN PROFILE CHANGED DRASTICALLY?

a)   Thermally heavier devices will need more careful profiling. Must
insure that pads near center of the device get up above reflow for
sufficient length of time, etc.
b)   What you need to control is peak reflow temp. Try not to exceed 210 C
about 55 sec else you might see bigger size voids. If you will be doing
first time and you want to do right take profile using existing setting on
solder sample loaded with scrap/dummy/wrong size components. Try to make
thermally equal as possible leaving chips. For BGA, we had dedicated
profiles.
c)   The larger--in a thermal sense--a component is, the slower it will
reach solder reflow temperature, with the solder joints closest to the
center of a BGA heating up the slowest. The PCB will typically reach
soldering temperature before the component, but that depends on the mass of
the BGA. The only way to determine an adequate reflow profile is to
instrument an assembly with thermocouples and measuring the temperature
profiles at various locations. Your slowest heating soldering pads need to
reach minimum acceptable reflow temperature with some margin to assure
consistent properly wetted solder joints.

6.  WHAT ABOUT SOLDER MASK DEFINED PAD?

a)   Solder mask defined (SMD) solder joints show shorter cyclic lives than
otherwise equal non-solder mask defined (NSMD) solder joints. The SMD
solder joints contain stress concentrations which have been show to reduce
to between 70 and 33% of the NSMD solder joints, depending on severity of
the test conditions as well as the severity of the stress concentration.
b)   Per Tandem study, your pad should be same as that of BGA component pad
for optimum reliability. If BGA's ball is with solder mask define then go
for solder mask otherwise copper define if BGA is copper define.

7. WITH THIS LARGE OF A DEVICE (40MM X 40MM) WHAT ABOUT CTE MISMATCH
......?   USE HIGH TG .....?

a)   Since you're using a plastic part, the TCE mismatch shouldn't be that
great between the component body and the board.  You will have a local TCE
mismatch between the solder and the component or board, though.  You really
need to do some testing (thermal cycling) or finite element analysis to
determine the reliability of your design.
b)   We found it necessary to go from the "regular" (i.e., difunctional)
FR4 to the multifunctional blends.  Right now we use the 170-180 C Tg FR4
as well as some polyimide blends on certain products.  We found that the
weak link in this technology wasn't the solder joints (for our
applications), but rather the array vias within the board.  For adequate
routing, narrow vias and thicker boards are often necessary, and this is
where we found failures early on.  By going to the better dielectrics, we
found far fewer failures.  A recent paper by Greg Martin from IBM at SMI 97
echoed these results.
c)   I gather that plastic parts on FR4 do not pose much trouble on modest
thermal environments such as your first range. The next step appears to be
to use an underfill to relieve stress on the solder joints.
d)   I recommend high Tg as BGA rework require multiple heating at higher
temp.
e)   The concerns about the reliability threat from a CTE-mismatch can be
assessed using the information in ANSI/IPC J-STD-013 "Implementation of
Ball Grid Array and Other High Density Technology", and IPC-D-279 "Design
Guidelines for Reliable Surface Mount Technology Printed Board Assemblies".
The Tg of the BGA and PCB materials comes into play only during processing;
from first principles, the closer the BGA and PCB materials are in Tg and
CTE above Tg the better.

8.   WHAT ARE YOUR RESPONSES TO THE QUESTIONS ABOVE FOR MICRO-BGA?

a)   None received.

##############################################################
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.
For the technical support contact Dmitriy Sklyar at [log in to unmask] or 847-509-9700 ext.311
##############################################################