I’ve now twice experienced severe instability during the descent phase of RTL on ArduCopter 3.6.10 on an oldish Pixhawk4 (fmu_v2). I haven’t managed to capture logs yet (will look into this), so at this point I’m just asking if anyone has experienced anything similar, and hints for troubleshooting beyond logs.
Background: hardware unchanged, previously flown with AC3.5, no problems whatsoever.
Since the firmware update to 3.6.10 I didn’t pay much attention to RTL, but once I did, it gave me a scare: the copter returns to launch coordinates nicely, starts descending, but after a few seconds (5? 10? A few meters down.) it suddenly starts flip-flopping violently, as if over-correcting, reaching almost 90 degrees. (I’m not certain at this point whether it was pitch, roll, or both.)
Both time’s I’ve managed to stop the instability after 2 or 3 movements by switching mode (once switching the actual flight mode, once deactivating RTL mapped to a switch). I think the drone was in GUIDED mode when RTL was switched on, but this is one of the things I suppose I’ll have to test for relevancy.
There are other questions about sudden instability here, but my experience seems to happen only in, and consistently in, RTL descent - no problems in loiter/stabilize/guided modes.
Appreciate your input, while I steel my mind to capture some logs without crashing the copter…
Might simply be power settling or vortex ring state as it’s know in the helicopter business. Coming straight down, the air frame is descending in its own propwash. You will see major control excursions in pitch and roll. I have seen a bit of this on my quads, but never to the extent I thought it was gonna loose control.
If possible, reduce the rate of descent in the params. Most helo flight manuals will state greater than 400fpm rate of descent and zero to little forward airspeed can cause this.
An additional test would be to use the same rate of the descent in RTL in another flight mode and if shows some instability, add a bit of forward speed and see if it doesn’t smooth out = flying into cleaner air.
@glacier051 The descent phase was pretty slow, though, and similar manual descent didn’t produce any visible turbulence. Still, I will check parameters and pay attention to this next time, last time was all eyeballs for instruments.
@xfacta would the FC encounter motor/esc/desync issues in one mode, but not others? I’m just thinking that (so far!) I’ve only seen this happen during RTL descent. I’m not familiar enough with the inner life of ArduCopter - might stabilization algorithms or inputs change for landing, and cause new effects?