BGA voids

Subscribe to Thread

Good morning. We started our first mass production of pc... - Apr 23, 2009 by joseph

Every paper I've seen published shows that BGA voids either ... - Apr 23, 2009 by ScottE

During initial profiling we shoot for 15% or less voiding on... - Apr 23, 2009 by Cross

That you have one BGA of 30 BGA that has voids, makes us won... - Apr 23, 2009 by davef

CHANGE SOLDER PASTE, AND ADJUST PROFILES. Nolasco ... - Apr 24, 2009 by nolasco

Remember that the 25% is a reduction in cross sectional area... - Apr 27, 2009 by SteveK

joseph

#58644

BGA voids | 23 April, 2009

Good morning.

We started our first mass production of pcb assembly with BGA component recently. During X-Ray inspection, we found only one ball in an affected board having > 25% void (total pcba inspected is 15 boards, each board has 2 BGAs and each BGA has 64 balls).

Referring IPC-7095 stated that there is no evidence or empirical data that indicates that voids within ball will cause failure. We would appreciate if you can advise us for how you approach this, whether to consider it as process indicator or reject ?

Thank you again for your time and effort.

reply »

ScottE

#58645

BGA voids | 23 April, 2009

Every paper I've seen published shows that BGA voids either have no effect or actually give a slight improvement. The one exception are voids with a root cause in bad incoming boards, such as champagne voiding where nearly the entire interface is gone. I've done thermal cycling of BGAs(-40/+125C and -40/+150C) and found voids had no measurable effect on reliability, including voids up around 35%.

I have no idea what the words "process indicator" means in this context and wish IPC-610 hadn't used them. If voids aren't a defect, what are they indicating? Why would I monkey with my process and potentially create another problem to "solve" something that's perfectly OK?

The one place where I'm not yet willing to discount voids is in high power applications where significant voiding might effect current carrying and/or heat transfer. This is mostly with QFN style packages rather than BGAs.

reply »

Cross

#58647

BGA voids | 23 April, 2009

During initial profiling we shoot for 15% or less voiding on our BGAs. During production runs we consider voids under 15% to be ok. Voiding between 15%-25% we string tag and pay more attention to during test. Over 25% we consider a reject.

reply »

davef

#58650

BGA voids | 23 April, 2009

That you have one BGA of 30 BGA that has voids, makes us wonder what's the deal with that one BGA. Especially given that this is the initial run of a product with BGA at your plant, it really makes us want to understand just what is driving the voiding in that part.

reply »

nolasco

#58651

BGA voids | 24 April, 2009

CHANGE SOLDER PASTE, AND ADJUST PROFILES. Nolasco

reply »

SteveK

#58663

BGA voids | 27 April, 2009

Remember that the 25% is a reduction in cross sectional area. A void in the middle of the ball that reduces the cross sectional area by 25% is about half the diameter of the ball. If it is closer to either termination, the corresponding size would be less.

You really should put together some test boards and characterize your process by cross sections. Also, take note of your x-ray settings and measurements before x-section. That will let you "callibrate" what you see in the x-ray to real life.

reply »