Ioan,
One thing you can do to help reduce the amount of stress seen directly on the BGA solder joints is to underfill the BGA prior to depaneling. Once cured, the underfill will help alleviate the stresses seen directly on the solder joints/pads caused by board flexure. This is especially true when you have a thin circuit board, as you most certainly do. I am not talking about stresses causing the solder joint itself from cracking, but the pad attachment to the PWB from cracking.
Please note this will also help prevent cracking due to CTE mismatch, however, the underfill epoxy must be carefully selected such that its rate of CTE closely matches the rate of CTE of the PWB. 
An excellent article on this subject was written by Werner Engelmaier. I have attached it here. http://www.globalsmt.net/regions/japan/content/view/33/65/ 

I don't believe there is any point in performing dye and pry, etc., as I believe you have already determined the root causes of the cracking, which are essentially:
1. pad cratering due to mechanical stresses during depanelization.
2. Handling issues.
3. Possible CTE stresses during service cycling.

All of which are contributed to by the thin board.

Two other possible fixes are looking at the board material itself. Some circuit board materials crater much easier than others. It is also possible you can use an .062" board rather than the .028" thickness, but that depends on whether the board must fit into a sliding frame, whether it has edge connectors that must line up, etc.

So, you have three options you can try without any drastic changes to the overall design.
1. Epoxy underfill
2. Board material
3. Board thickness

If reworkability of the BGA is an issue such that you don't want to do a full underfill, then try a combination of cornerbonding and edgebonding of the BGA using Loctite 3609. While perhaps not as strong as a full underfill, this may be enough to prevent the cracking. If you need to rework the BGA you can heat the Loctite 3609 to 150 C and it will easily chip off with a small exacto knife, allowing you to remove/replace the BGA. See examples of this in the pictures in this article, also by Werner Engelmaier
http://www.globalsmt.net/regions/uk/index.php?option=com_content&task=view&id=1971&Itemid=131


-----Original Message-----
From: TechNet [mailto:[log in to unmask]] On Behalf Of Ioan Tempea
Sent: Monday, July 21, 2008 6:25 AM
To: [log in to unmask]
Subject: Re: [TN] Mechanical damage to BGA

I see your point Eric.

The size of the balls is actually a manufacturing error, the balls were intended to be 0.35mm in diameter, so the pad size was designed at 0.25mm on both the component and the PCB.

So, since nobody comments on the size of the traces and the pad size from the mechanical and stress and CTE point of view, I guess the design is OK...

Thanks to everybody that answered,

Ioan

-----Message d'origine-----
De : TechNet [mailto:[log in to unmask]] De la part de Eric CHRISTISON Envoyé : July 21, 2008 4:14 AM À : [log in to unmask] Objet : Re: [TN] Mechanical damage to BGA

Measure the size of the pad on the underside of the BGA device.

The size of the pad on the PWB should be about the same size to make the joint of equal strength at both interfaces.

0.25 for a 0.60mm dia ball sounds far too small to me.

Agreed though that you may still get failure at depanelisation due to board flexure. Best to avoid that.

Regards,

Eric Christison Msc
Mechanical Engineer
Consumer & Micro group
Imaging Division

STMicroelectronics (R&D) Ltd
33 Pinkhill
Edinburgh EH12 7BF
United Kingdom

Tel:	+44 (0)131 336 6165
Fax:	+ 44 (0)131 336 6001

---------------------------------------------------
Technet Mail List provided as a service by IPC using LISTSERV 15.0 To unsubscribe, send a message to [log in to unmask] with following text in the BODY (NOT the subject field): SIGNOFF Technet To temporarily halt or (re-start) delivery of Technet send e-mail to [log in to unmask]: SET Technet NOMAIL or (MAIL) To receive ONE mailing per day of all the posts: send e-mail to [log in to unmask]: SET Technet Digest Search the archives of previous posts at: http://listserv.ipc.org/archives Please visit IPC web site http://www.ipc.org/contentpage.asp?Pageid=4.3.16 for additional information, or contact Keach Sasamori at [log in to unmask] or 847-615-7100 ext.2815
-----------------------------------------------------

---------------------------------------------------
Technet Mail List provided as a service by IPC using LISTSERV 15.0
To unsubscribe, send a message to [log in to unmask] with following text in
the BODY (NOT the subject field): SIGNOFF Technet
To temporarily halt or (re-start) delivery of Technet send e-mail to [log in to unmask]: SET Technet NOMAIL or (MAIL)
To receive ONE mailing per day of all the posts: send e-mail to [log in to unmask]: SET Technet Digest
Search the archives of previous posts at: http://listserv.ipc.org/archives
Please visit IPC web site http://www.ipc.org/contentpage.asp?Pageid=4.3.16 for additional information, or contact Keach Sasamori at [log in to unmask] or 847-615-7100 ext.2815
-----------------------------------------------------