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.

TMS320F2809: Thermal Cycle Possible Failures

Part Number: TMS320F2809

I have a number of pcbas with "TMS320F2809ZGMS" installed on them that went through a thermal cycle test. The parts should not have experienced more than 120 C or less than -40C. The parts appear to be somewhat functioning but seem to have problems accessing some of their internal flash memory sometimes or possibly only the OTP memory.  Some of these units seem to be fixed by replacing the crystal on the pcba, but some do not.  Also sometimes a unit might start working for a little while and then quit working.

Does this make that the flash or OTP memory could be degraded such that we could have intermittent operation of the units at times.  We should not have been writing to flash throughout the thermal cycle test. I noticed the flash has a lower operating range specified for writing to it for this part (0C to 125C). Could the flash still be degraded if you are operating close to -40C?
Some other pcbas with "TMS320F2809ZGMS" may have been exposed to temperatures down to -50C and had failed with a similar failure mode, but earlier in the test.
Could the flash fail if it is powered up and the ambient goes below -40C without writing to it?

Also your datasheet indicates that the write life of the flash could be lowered by writing to it below 0C. Do you have any data on how many writes are likely at various temperatures below 0C?

  • Glenn,

    Thanks for reaching out to the TI E2E forums.

    I don't anticipate the temp cycling itself or the operation of the device inside its temp range(including programming the flash) to have this kind of effect.  I would also not expect the flash to degrade permanently if only reads occurred outside of the temp spec.  Programming or erasing the flash in this region may be a bit more uncertain, although I would think this could be recovered with a reprogram after the device is back in spec.

    Thinking about the X-tal comment, if the device attempted a program and/or erase when the input clock was potentially out of spec could cause issues; but those should be able to be rectified with a reprogram once the clock is good.  For reads, if the device was out of spec for the clock, this would fix itself when the clock is fixed with no damage to the flash internals(I think you are seeing this).

    I'd like to take this in a slightly different direction, since we are talking about temp cycling.  Since these are BGAs, have you tried to reflow the PCB post temp cycling to see if the failure goes away?  I'd like to rule out any kind of mechanical failure at the connection from the C2000 to the PCB before we debug any potential electrical failure inside the device.  Especially with the observations that the unit can work for some time but then fail.