This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

TMS320DM6467T: Errata and BGA quality

Part Number: TMS320DM6467T

Team, 

My customer has the following question: 

We are investigating some issues we have been having on a product that incorporates the TMS320DM6467T DSP.  We have 3 DSPs on one board in the system and all three are tied to similar power controlled by another board.  We able to get serial logs out of the DSP and monitor the powerup sequences when the boards are power cycled.

Somewhat randomly, we have an issue where one DSP will delay 2 minutes before booting compared to the other 2 DSPs.  This has happened on several boards and not one specific board issue.  Sometimes the part doesn’t powerup at all until a new reset.  Typically this is all happening while at temperature extremes (sometimes -40C, sometimes 70C, but occasionally at ambient).

When we get these DSPs, we buy them with unleaded solder.  We reball them so that they use leaded solder to match the rest of our board.  I guess we can’t buy leaded solder BGAs and this is our only way to get leaded solder balls.  Then the parts are installed on the board and conformal coated.  So by the time we install the boards into the final product, the DSPs have been reflowed at least 3-4 times (repair might cause a couple more cycles). 

 We are working a wide scope fishbone investigation that is looking at software, BGA solder profiles, connectivity, etc… but I’m hoping you can help me with the DSP itself.  I have two main questions:

1)      Do you know if any errata or failures that might be aggravated by the somewhat unusual processes we are putting the DSP through?  Anything that would cause a powerup delay or a complete lockup?  Possibly in the HW itself or the ROM Bootloader?

  1. If a part starts failing due to being reflowed too much are there any telltale signs to indicate that?

2)      We have some confirmed soldering issues with the BGA DSPs and I’m wondering if TI has any best practices for using the TI BSDL file to improve our boundary scans?  Our supplier is currently using the BSDL for low speed connectivity.  But I’m wondering if there are practices that will better identify cold solder joints or other soldering issues where connectivity is weak and not purely disconnected.

 Regards,

Aaron

  • Hi Aaron,

    Sorry for the late response on this - I only have limited guidance on this as this is a legacy device.

    >>When we get these DSPs, we buy them with unleaded solder.  We reball them so that they use leaded solder to match the rest of our board.  I guess we can’t buy leaded solder BGAs and this is our only way to get leaded solder balls.  Then the parts are installed on the board and conformal coated.  So by the time we install the boards into the final product, the DSPs have been reflowed at least 3-4 times (repair might cause a couple more cycles).

    Talking to my quality team , the above is concerning - a reballed BGA is not recommended for long term use. This would violate the warranty. I do not know if that itself is manifesting into the issues your customer is doing. This will be difficult to support.

    I did look through the errata and don't see anything spike out that talks about a complete lock up , except for the DSP bugs listed in the errata, which IMHO do not look to be the cause, given the issue is at boot up etc. 

    Given issues happen at temperature extremes, it would be good to run some diagnostics on board if the board support to ensure clock/power supplies and peripherals used for boot are working. 

    It would be good to understand if ABA swap experiments are done - to see if the failure is following the device or the board etc. 

    ON 2, I do not have any application note etc that I can point to for this. 

    Regards

    Mukul