previously, get_devname() was used as the I2C device path, but on NuttX,
get_devname() is the device file which the driver creates. This patch
changes it, so the sematics are the same as on NuttX: both now use _bus
to decide to which I2C bus device to talk to.
I did not see any other use-cases than the led on ocpoc.
2. Rewrite and rebase pca9685 driver
3. Try to fix issue when push the stick of channel 3 to the maxmum position, 0uswill be output to channel 1, should be maxmum pwm signal
4. Fix the code style
Both 8 channels PPM encoder and 8 channels revicer are required.
Before launch px4, ppmdeocde programe should be launched.
To download ppmdecode programe,
visit https://github.com/crossa/raspberry-pi-px4firmware.
Pxfmini and navio are not popular autopilot hardware in china,
I can handly to purchase it.
So that I use raspberry pi to build autopilot separately.
This dirver help us to decode ppm single to pwm and pushlish it
Use the size of each item type instead of the biggest one.
In AeroFC that runs is constrained mode it was using 7860 bytes
and now it uses 6930 bytes almost 1KB less.
This change first pushes out the _reset_wait by 100 Ms.
which is about 3 time longer then the code take to execute.
Then it does the reset of the accel, gyro and mag and
the ends the wait by setting _reset_wait to now+10 us.
The MPU9250 and MPU6500 buth support 1 Mhz and 20 Mhz. Buy upping the clocc we will get the maximum clock rate the driver
supports that is <= 20 Mhz. This will boost the FMUv4Pro SPI speed to 11.25 Mhz (it was half that)
Insure a 0.0 voltage initial condition on VDD_3V3_SENSORS
By starting the GPIO_VDD_3V3_SENSORS_EN, low and deferring
the GPIO init of the slave selects and drdy signals until
board_app_initialize. We get ~ 180 ms of power off time
with 0.00 voltage applied to the sensors.
This ensures that the massive (several hundred params if temperature compensation is enabled) param load is not done in flight. We also lower the priority to ensure that attitude controllers and estimators which might consume sensor data directly execute immediately, which should reduce their latency.
If we have not received an inbound heartbeat we are probably not connected and so user-facing output (which is even buffered) should go only out once we are connected.
