Other Parts Discussed in Thread: MSP430FR5969
Team,
We've been running into a MSP430FR5959 issue recently, and we were hoping you might be able to help us out with it! The gist is that at low temperatures (starting at -20C or so), some MSP430FR5959 chips appear to be sporadically locking up.
- The lockups are relatively infrequent, occurring every few days or so of runtime on affected chips (but see note on temperature-dependence below). We have so far observed this on about ~10 chips out of a few hundred.
- It appears to be temperature-dependent. The effect typically begins around -20C. The rate appears to increase as temperature decreases. At ~-35C the rate appears to be significantly higher than at ~-25C.
- It also appears to be chip-dependent. The chips which exhibit this behavior exhibit it repeatedly, while others running the same firmware do not exhibit it at all. The temperature dependence also appears to be per-chip. Some chips exhibit the issue at ~-20C, while others only start exhibiting the issue at ~-30C.
- All the affected micrcontrollers have previously been exposed to both hot and cold temperatures for extended periods of time
- We have the WDT_A watchdog timer enabled on these chips, and the watchdog does fire after the expected time (~130 ms in our case) and the chip recovers. Watchdog intervention is confirmed by the chip's reset reason (SYSRSTIV = 22).
Some other details about our specific situation:
- We run the MSP430FR5959 off its internal DCO at 16 MHz. There are no external clocks or other clock domains, though some (but not all) affected firmwares do operate as a SPI slave with asynchronous SCLK.
- The firmware runs exclusively out of FRAM. Program code and data including all variables and program stack are stored in FRAM. We are not using the chip's SRAM.
- We configure FRAM NWAITS=1
- We make continuous use of all 3 DMA channels on the chip
- We do not use any LPM modes (we always stay in active mode)
- This has been observed in at least 4 different firmwares (though they all share the above properties).
- These symptoms (chip-dependent, cold-dependent, and resulting in lock-up) are similar to wait state issues we have encountered on different chip technologies.
Anyway, thank you for helping us look into this. We were hoping you could let us know:
- Are there any known issues or quirks with the MSP430FR5959 operating at cold temperatures?
- Can you confirm that FRAM NWAITS=1 should definitely be sufficient for a MSP430FR5959 running off its internal DCO at 16 MHz across the operating temperature range? Does TI recommend extra FRAM wait states?
- Is there any other information that comes to your mind that may help us with this problem?
Thank you very much,