We are using the DM3730CBP in a small volume project and have not really solved soldering problems since the beginning.
More than 50% of the boards coming from reflow have short circuits or interruptions or even both. Usually it is possible to repair such a board by replacing the DM3730 + Micron PoP stack in a rework station but this is time an money consuming, i.e. not a good and sustainable production process. It also indicates that the board and solder stop mask are not the problem. We even have tried two different PCB manufacturers and no difference. Several different stencils have been tried and different flux. As well as soldering the SoC first and then the PoP. Or both together. Or building a stack first and soldering the stack. And, we tried automatic Pick&Place in the assembly line and semi-automatic (manually controlled robot with optical feedback) on a rework station.
Now we have analyzed everything and come to the conclusion that the CBB and CBP packages have some warp which appears to squeeze the center balls (VCORE - short circuits) and lifts off the balls in the corners (interruptions).
Here are X-Ray images of three different PCB layout series. The oldest one is with a OMAP3530CBB and the other two with a DM3730CBP. The latter show different short circuits.
http://download.goldelico.com/gta04/images/20101109-GTA04A2-overview.jpg
http://download.goldelico.com/gta04/images/20110707-GTA04A3-schlecht-xr002.jpg
http://download.goldelico.com/gta04/images/20120622-GTA04A4-schlecht-01.jpg
What are we doing wrong that the balls are getting more flat in the center?
Since there are millions of devices produced with the OMAP3530 and DM3730 we hope that someone can give us a hint where to look.
Thank you,
Nikolaus Schaller