6
\$\begingroup\$

Sophisticated PCBA, with a DM6446 using two 16 bits DDR2 devices to implement a 32 bit memory space. On some boards, we see a stuck bit on either bit 24 or bit 8, which map to the same BGA ball (C8) on the two different devices. Of the known-bad boards, 12/12 have this failure mode (either bit 8 or bit 24 stuck).

Sent one board out for cross-section & SEM. Reliability lab says the solder ball did not wet the pad, so electrically open.

Did a die and pry on a second board with the same failure mode. All balls on the two devices are good, except for ball C8 on both devices. On one device, the joint is half failed; on the other device, full failure.

Anyone ever see anything like this?

\$\endgroup\$
3
  • 2
    \$\begingroup\$ Is C8 tightly connected to the local ground plane? It sounds like you have something preventing proper reflow on C8. \$\endgroup\$ – Connor Wolf Aug 17 '13 at 1:10
  • 2
    \$\begingroup\$ We would need links to the datasheets for the RAM, PCB layouts, and a whole bunch of other things. \$\endgroup\$ – user3624 Aug 17 '13 at 4:39
  • 1
    \$\begingroup\$ Could you post images of the board layout for each IC, cropped a few mm outside the footprint, with C8 indicated? It would be particularly interesting if any geometry is in common. \$\endgroup\$ – Chris Stratton Aug 17 '13 at 15:50
1
\$\begingroup\$

Perhaps a long shot, but maybe inspect a few dozen of the RAM chips under scope for any sign of C8 pad defect?

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.