PX4NuttX: setup I2C digital filter

[tridge]

This may reduce the impact of I2C bus noise, which could help with compasses and with airspeed sensors

[OXINARF]

From a logic perspective looks good, but I can't say much more. I'll only ask if this was tested in all PX4 boards (and maybe Aero)?

[tridge]

only tested on FMUv2 so far

[WickedShell]

The comment for the table number should be annotated with the datasheet version you are using, because the RM0090 I used it was a very different table number.

[WickedShell]

The digital noise filter makes a significant improvement to the readings of the data off the bus. Far fewer bit flips on sensor data. (It doesn't entirely eliminate them, you can see a single temperature spike on the log with the noise filter, but compared to the other log file it's a massive improvement.

This data is collected from an extremely noisy system, that has large grounding problem on the I2C signal lines, crosstalk, and is generally quite a horrible RF enviorment. There remains a bug where the MS4525DO driver can accept data that is invalid, but @tridge and I think we can manage this by double reading the sensor, and rejecting any min/max sensor output values (the most common failure remaining).

With the filter:

Without the filter:

EDIT: Other then firmware, everything was identical, nothing was moved or altered between the two tests.

[WickedShell]

With:

Without:

[tridge]

excellent results Michael!
@lucasdemarchi, can you test on aero-fc?
@rmackay9 can you do a flight test on a copter?

[tridge]

@rmackay9 randy did a successful test flight, so I will push now

[WickedShell]

Flight tested this as well at this point, no issue, cleaned up a noisy system quite nicely.