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.

TMS320F28377S: Microcontroller stops working at cold temperature below -10C

Part Number: TMS320F28377S


What is the difference between Wafer Fab code "$4" and "YF"? We have the TMS320F28377SZWTS IC (NFBGA package) with Wafer Fad code and silicon as "$4C" and it shows problem operating at cold temperature below -15C and wonder if it's related to the Wafer Fab code. We don't has this problem with the Wafer Fad code and silicon revision as "YFC". How can we continue to purchase this IC with Wafer Fad code and silicon revision as "YFC"?

Thank you very much for your help.

  • Part Number: TMS320F28377S

    What is the difference between Wafer Fab code "$4" and "YF"? We have the TMS320F28377SZWTS IC (NFBGA package) with Wafer Fad code and silicon as "$4C" and it shows problem operating at cold temperature below -10C and wonder if it's related to the Wafer Fab code. We don't have this problem with the Wafer Fad code and silicon revision as "YFC". How can we continue to purchase this IC with Wafer Fad code and silicon revision as "YFC"?

    Thank you very much for your help.

  • Chung,

    Can you give more detail on the issues you are seeing with the device at cold temperature?

    This FAQ goes into detail on the changes made wrt fab code and requirement that a new Flash API be used, however, it is my experience that if this were the issue the CCS would not allow you to program the device.

    I have also attached the PCN that give more info on the fab code.  Note that other than the change required to the flash tools, devices from any fab should behave the same in your application.  As such TI will ship any of the listed devices interchangeably to fulfill orders.

    Best,

    MatthewPCN20200115000.2.pdf

  • It's the same thread as following:

    https://e2e.ti.com/support/microcontrollers/c2000-microcontrollers-group/c2000/f/c2000-microcontrollers-forum/964799/tms320f28377s-cold-temperature-operation

    We've narrowed down the root cause is that somehow the microcontroller switched to internal clock from external 20MHz clock at -10C causing the FW not working properly anymore. We did re-spin the PCB board but the problem is still there and we wonder whether any differences between the chip with YF and $4 since the board with YFC is working fine down to -30C. By any chance do you have documents specified the soldering process for this particular NFBGA ZWT0337A package?

    Thank you very much for your help.

  • Chung,

    I'll speak with Hareesh on the previous forum to see if there is additional data not captured on the post.

    This is the spec for the nFBGA package, let me know if Section 4 gives the info you need.  https://www.ti.com/lit/spraa99

    For the 20MHz external clock, is this a X-tal or a canned oscillator?  Have you looked at the clock at cold temp to see if it is having any issues locally?

    I believe that the missing clock detect logic on this device will initiate an automatic switch(by default) to the internal clock if it detects a missing clock condition when the external clock is in use.  It will also issue an NMI, which if not serviced will eventually time out the watchdog, resetting the device(which may be what you are seeing as well).

    Best,

    Matthew

  • The external clock is a standard clock oscillator 20Mhz from Abracon, P/N: ASFLMB=20.000MHz-LR-T with a 33R2 resistor connected to X1 (G19) of the TMS320F28377S. We did try to capture the missing pulse of the output of the oscillator but we did not capture any when ambient temperature reached below -10C. What is NMI and whether FW can detect this before the device switched to the internal clock?

  • Chung,

    Do you mean that the oscillator output was flatlined/dead below -10C or that you could not capture any issues actively with the scope?

    In terms of the behavior of the missing clock detect logic the TRM on page 111 gives the flow diagram of what will happen:

    https://www.ti.com/lit/spruhm8

    This logic can be disabled, by writing bit 2 of the MCDCR register.  This is on page 353 of the TRM link above.  If you believe the logic to be falsely detecting a missing clock this would be an option. 

    If this is left enabled, if the logic believes there is a missing clock, the switch to the internal oscillator and high Z of all PWM outputs is done in HW.

    I would look at the section 111 to understand how the logic works, and what the device considers/detects a missing clock to understand if there is a potential issue in normal operation of your system.

    Best,

    Matthew

  • Thank you so much for your info and the document about missing clock detection logic. Recently we have changed the oscillator from 20MHz to 10MHz on our board and tested with a simple basic flash program and we found that it can run down to -40C without any problem. However, it can only restart when ambient temperature >-30C. This could reveal that the external clock was falsely detected by the MCU. We'll try with the existing FW and 10MHz oscillator to see if the MCU can run at -40C and let you know the outcome.

  • We've found the root cause why the missing clock detection logic flag is triggered. We realized that VDDOSC (H16 and H17) needs to be connected to VDD (+3.3V), according to errata doc, SPRZ422I page 29, otherwise internal frequency can be drifted. Once H16 & H17 are tied to VDD, the MCU can operate down to -20C without any problem. Definitely, the part with wafer FAB code YFC is different than $4C.

    Thank you so much Matthew.