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.

TMS320F28062: Flash Memory Sectors Erased

Part Number: TMS320F28062


Tool/software:

I have found a recurring issue with the TMS320F28062PNT where a certain MCU memory sector is being erased while the device in in storage. It has a low incidence rate but has happened a few times now. We have identified MCUs from batch numbers 26AZDXW and 15CS9QW to have this issue and were wondering if this is a know issue for certain batches or a more general problem. A process issue is unlikely as the product has been in use for many years now and this issue has only been seen recently and from different batches internally.

Any help would be appreciated.

Many thanks,

Liam

  • Liam,

    I am not aware of any known issues involving flash retention.  We have 15 year retention spec(at Tj = 55C), but more than that your comment that a sector is being erased would not be something I would expect of a retention issue, i.e. I would expect a bit or perhaps multiple bits, but in a random fashion, and not an entire sector.

    The batch numbers would correspond to June of 2022(26xxxx) and May of 2021(15xxxxx) respectively, so much shorter duration than 15 years even assuming they were programmed soon after manufacture.

    Is it the same sector on both devices?  How long between when the devices were programmed and they were used again?  Can you comment on the storage conditions, like temperature, humidity, etc?

    Best,

    Matthew

  • Hi Matthew,

    Thank you for the response.


    We program 2 sectors on the device and it has been the same sector erased both times. The first 2 times we noticed this issue was after a high temperature process so we did initially attribute the issue to this (although this heat treatment is applied to all our devices and again, no issues previously). However, when investigating this issue we saw it occur in a device that was programmed and left in normal room temperature storage for about 2-3 weeks before it was tested again and the issue was seen.

    Many thanks,

    Liam

  • Liam,

    Let me follow up with some folks in the quality team, as this behavior is unusual and I have not seen something like this before.  We may need to have these devices returned to TI for failure analysis, but let me check with some others first.

    Best,

    Matthew

  • Liam,

    I am requesting that you return the devices to TI for failure analysis. 

    The returns process is documented here https://www.ti.com/support-quality/additional-information/customer-returns.html?keyMatch=returns and differs based on if you bought the devices from TI.com or from a distributor.

    You can reference this E2E thread for any authorization request/requirements you may get back from the submission.

    Please let me know if you have any questions.

    Best,

    Matthew

  • Liam,

                Before you ship the devices, could you please dump the data from the “erased” sector from a good device and a suspect device and compare them? You could save the sector contents as a hex file and then use software such as “Beyond Compare” to compare the contents. I am interested in determining if the sector in question is completely erased. i.e. every bit of every word is 1. It is also important to ascertain other sectors are perfect.

  • Hi Hareesh,

    When we previously checked, the entire sector was erased. We confirmed this through looking at the Flash memory via the JTAG debugger. Other sectors seemed to be unaffected. I will double check this with the other devices we have seen this issue with but have not reprogrammed.

  • Hi Matthew,

    I will begin the return process.

    Thanks

  • Liam,

                Just “looking” at a sector visually through the JTAG debugger would not suffice. Kindly do this before shipping the suspect device(s):

    1. Dump the entire Flash content of a good and suspect device as a hex file and compare them. This is to look for any bit flip anywhere in Flash memory.
    2. Repeat #1 for the suspect sector.

    In order to do a thorough analysis, it would help if you could send us the correct Flash image. You could send it as a private message to either me or Matthew. You need to initiate a friendship request to do this.

  • Hi Hareesh,

    I will follow your instructions and let you know the results

  • When you ship the devices to us for analysis, please refer to this e2e post.