• State
  • Locked Locked
  • Replies 6 replies
  • Subscribers 30 subscribers
  • Views 1500 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

MTBF of solder balls on the TMS570

Cass Hussmann
Prodigy 40 points


Hello,

I have been looking for a while and cant find any means of evaluating the mean time to failure of solderballs on the BGA package used for the TMS570 given the temperature variance I know the component and the board will each experience. I would like to see if the MTBF of solder joints will exceed the mandatory life time of the product as the BGA package will remove cheaper expanded IO chips requred when using the PGE package. Is there any data or references anyone knows of that might help me? I cant imagine that standard practice is to just evaluate this again and again for each new product as it is something that should be consistent with all same chip packages irrelevant of the die inside.

Thank you,

  • 0
    TI__Mastermind 49120 points

    Hello,

    I have asked our functional safety experts to comment on your query.

    Regards, Sunil

  • 0
    TI__Mastermind 49120 points

    Hello,

    The best option is to download the Safety Analysis Report (SAR2), which is available after a signed NDA is in place. You can then model failure rates for package with the BGA or QFP under various temperature profiles. With the default assumptions and motor compartment use profile, you get:

    • QFP: approx. 0.4 FIT/pin
    • BGA: approx. 0.2 FIT/ball

    This is based on an IEC/TR 62380 failure rate model, which is primarily looking at wear-out due to thermal expansion/contraction of the IC to PCB connection.

    Essentially the BGA is better per connection because of the better thermal characteristics, but has almost the same overall performance due to the larger number of connections.

    Regards, Sunil

  • 0 in reply to Sunil Oak
    Prodigy 40 points

    Hello Sunil,

    I been told that as our group is associated with a University we are unable to have access to the SAR reports. Given that this data will not be used for any paper or thesis and only for component justification would it be possible to somehow request the FIT values for the BGA and QFP packages over a temperature range of 18C to 73C alone. My concern is simply that with 10-12 thermal cycles a day that if we decide to try and use the BGA package to remove expanded IO chips and allow for more RAM, a significant number of balls will fail before the intended mission time has been completed.

     

    Thank you,

    Cass

  • 0 in reply to Cass Hussmann
    TI__Mastermind 49120 points

    Hello Cass,

    I checked with our team responsible for the functional safety collateral. Unfortunately we cannot distribute the SAR reports without an NDA.

    Regards, Sunil

  • 0 in reply to Sunil Oak
    Prodigy 40 points

    Thank you,

    As I can not sign an NDA as a member of a university apparently I calculated the values based on IEC TR 62380 and get values closer to

    650.34 FIT for the PGE package (PQFP), (4.51 FIT / pin)

    554.771 FIT for the ZWT package (PBGA), (1.61 FIT / ball)

    For a mission profile of 18C to 73C (our worst case) for 4020 cycles per year using FR4 as the PCB material. Can you think of a reason why this would be so much higher than what you expect?

    It also supprises me a little that IEC TR 62380 will always give a lower FIT value for a BGA compared to a QFP I was under the impression that the pins would be able to take more strain from CTE mismatches than a solder ball connection.

    Thank you,

    Cass

  • 0 in reply to Cass Hussmann
    TI__Mastermind 23000 points

    Hi Cass,

    TI does not restrict university access with a valid NDA in place.  The challenge is that many students and faculty have difficulty finding someone with both authority and willingness to sign an NDA on behalf of the university.  Sorry this is causing you problems.

    With respect to IEC62380, the thermal characteristics in the IEC technical report are very conservative to cover a wide range of   package designs.  We replace the generic package values with the actual thermal performance of the TI packages which you should be able to find in the datasheets.  The TI package values should bring down the numbers.  

    Another point of difference is likely the number of cycles.  I believe your profile has roughly 2x the cycles of the auto motor profile we use as default, which would likely result in a higher failure rate.

    The failure model is IEC 62380 focuses on thermal expansion causing mechanical stress.  At least for this failure model the superior thermal characteristics of the BGA give it an edge in projected failure rate.

    Regards,

    Karl