Puzzling Long Term Bugs

A list of bugs that we'd like help solving.

• TODO-984: Some REV7 and other V0p2 units seem to have problems with serial RX and TX, garbling some characters
  • Seems especially problematic for longer (16+?) blocks of characters such as setting a key
  • Have tried to ensure no sleeping or other fiddling with CPU clock during UART operation
  • Seems to apply to a few % of REV7, REV10 boards at least
  • In some cases it has been possible to load a large (30KB) app over serial with the bootloader
  • The internal oscillator is factory calibrated to within 10% (p304, table 29-9) but the max error the UART receiver can tolerate is about 4.5% (p185, table 20-2). Source (2016-12): AVR328P datasheet (2015-11)
  • App note on calibrating internal oscillator (2016-12): AVR053
  • Solved: The internal clock is calibrated against the 32.768 kHz resonator, fixing UART timing issues. The calibration does not persist across resets so in some cases the device cannot be programmed using the bootloader.
• SIM900 cannot restart PDP context if it loses signal while connected to GPRS.
  • The SIM900 enters an undocumented state with the same name as the 'PDP-DEACT' state, but cannot start GPRS from it.
  • As far as I can tell, this is undocumented and there is no way out.
  • Workaround: Reset the SIM900 if it responds with PDP-DEACT.
• REV10 occasionally stops sending GSM packets, although the firmware is working correctly and the SIM900 responds as if it is sending.
  • Workaround: This is a SIM900 bug. We reseting SIM900 every 255 messages in case we get stuck in this condition.
• Failure to program ATMega328p EEPROM on long (>8 bytes?) writes.
  • Solved: Details can be found here