AP_Compass: don't publish frozen data #9549
Conversation
|
seems to be a problem in the particular driver. What driver is that? |
@lucasdemarchi it is noticed with QMC5883L device. |
|
this doesn't look correct at all. It is quite possible for a compass to produce exactly static values while sitting on the ground awaiting takeoff. If this was used then the compass would show as unhealthy and the aircraft couldn't fly. |
|
@tridge unfortunatelly I've lost the log. Are you sure some compass normally could produce a static value without any noise? |
SergeyBokhantsev
force-pushed
the
AP_Compass_avoid_frozen_data
branch
from
2ee1b5c
to
b6e4549
Compare
|
If @tridge can confirm that a sensor could reasonably produce absolutely consistent values for seconds then we can close this. Marked DevCall |
|
See also #9525 - which seeks to remove the equivalent stale-data check from all barometers.... |
|
doesn't this mean if we have a very low noise compass it will be marked unhealthy? that seems like a bad idea |
|
If this is an issue with a specific compass driver then I'd like the fix in that driver. |
|
I don't think this is a valid change without a log to back things up, and even if found we'd want to enable this in one driver |
This is a try to avoid a situation described in
https://discuss.ardupilot.org/t/compass-stop-working-but-remains-healthy/33205
when compass hangs up for some reason. It still can be read but does not provide any changes. So it stays healthy all the time:
(plz never mind the peaks, they're likely caused by a long wires)
This fix does monitor the values published and if they are equal for a number in a row ( > threshold), then publishing will stop till the values be different again. Stop of publish will lead to a bad compass health, that is the goal.