Magnetometer failsafe: YES (1 events)

PX4 Autopilot
1

I’m testing a few Black Cube Autopilots and sometimes I see “sensors status” return a failsafe event on the magnetometers.

Can anyone explain what might be triggering this? Is an internal mag sending bad data or did it timeout?

error count is 0 for all 3 mags.

This happens consistently every boot.

INFO  [ecl/validation] validator: best: 0, prev best: 2, failsafe: YES (1 events)
INFO  [ecl/validation] sensor #0, prio: 75, state: OK
INFO  [ecl/validation] 	val:  -0.4198, lp:  -0.4237 mean dev:   0.0000 RMS:   0.0078 conf:   1.0000
INFO  [ecl/validation] 	val:   0.1628, lp:   0.1620 mean dev:   0.0000 RMS:   0.0078 conf:   1.0000
INFO  [ecl/validation] 	val:   0.7605, lp:   0.7501 mean dev:   0.0000 RMS:   0.0166 conf:   1.0000
INFO  [ecl/validation] sensor #1, prio: 50, state: OK
INFO  [ecl/validation] 	val:   0.0226, lp:   0.0240 mean dev:  -0.0000 RMS:   0.0034 conf:   1.0000
INFO  [ecl/validation] 	val:  -0.2786, lp:  -0.2904 mean dev:  -0.0001 RMS:   0.0122 conf:   1.0000
INFO  [ecl/validation] 	val:   0.4233, lp:   0.4158 mean dev:  -0.0000 RMS:   0.0086 conf:   1.0000
INFO  [ecl/validation] sensor #2, prio: 75, state: OK
INFO  [ecl/validation] 	val:   0.4310, lp:   0.4318 mean dev:  -0.0002 RMS:   0.0357 conf:   1.0000
INFO  [ecl/validation] 	val:  -0.7049, lp:  -0.7078 mean dev:   0.0001 RMS:   0.0122 conf:   1.0000
INFO  [ecl/validation] 	val:   0.0135, lp:   0.0298 mean dev:   0.0001 RMS:   0.0266 conf:   1.0000
2

@rroche? How can we help on this topic?
No error count, yet a failsafe is strange.

3

Digging through the code it looks like it may just be a reprioritization of the 3 internal mags based on confidence level - not that any of them has had a failure. I think “failsafe” may not be the best term but the same reordering of priority also occurs when a mag fails.

This doesn’t occur when an external mag is used.

4

Very good! Good to hear there is nothing actually wrong here, some improvements to messages may make this less confusing, maybe you may be able to think of a better wording and do a Pull request?

5

Thanks for the reply @proficnc :tada:

@dlwalter I agree with @proficnc if you have an a few extra minutes to spare a PR would be much appreciated.