sealhat / sealhat_firmware Goto Github PK

Implement a counter that keeps track of fixes without the "valid" flag and compare to a VALID_TIMEOUT define. If the fixes consistently not valid then the conditions are poor and the device should sleep for a while.

Also, consider logging the satellite information (quality, number) if this occurs.

USB should not be active when not plugged in. there should be a interrupt that triggers off of VBUS (refer to schematic for pin) and enables the USB module and associated clocks. The device should go back to the original mode once USB is unplugged.

This is already written for simple debugging. We should make it actually recover and log an error to FLASH.

the idle task has been re-implemented to put the processor to sleep but in IDLE. systick must be connected to a clock that goes to sleep in STANDBY mode since if we use STANDBY mode in the idle task the processor never wakes up.

seal_IMU.c has consecutive function calls that return errors and the errors are not checked. These need to either be checked before continuing or documented as not capable of returning an error with the given parameters.

applies to GPS, but should verify later

A function is needed to set the RTC counter value and therefor the current time. Preferably from gps?
We are not usign the START driver for our timestamps but maybe they could be used here?

see: /tasks/seal_CTRL.*

use Atmel smart to add the hardware instances for our custom boards.

Our processor has EEPROM NVM module in the chip we can use for persistent settings. This requires setting the proper fuse to set the EEPROM size to 256 Bytes and also drivers in Atmel start.

description of the hook

the RTC is a 32 bit counter tracking seconds since epoch. Our timestamps use this value, but the START calendar driver uses a struct of actual date values. Hopefully we can pull the function that does this out from the depths of START somewhere.

Function to set the epoch, the calendar module can have a custom epoch year.

currently if a task fails to start the system goes into a while(1) loop and dies. The system should instead log the failure and continue on with the remaining sensors. the only show-stopper task is the control/logging task.