I had a few FBWA/autotuning flights with this plane before which all went ok.
Yet today, when I switched to RTL, to test it for the first time on this plane, at about 10 secs in, this happened:
Does anyone have an idea what could have made the plane act like this - rolling like crazy and flying inverted?
Unfortunately there is no log (yet), since the plane is still stuck on the top of a high tree. Hoping to recover it in the next few days with the help of strong windsā¦
Since DJI OSD is not being recorded, itās hard to trace back what I did (except panicking). Are Yaapu messages logged somewhere?
Iām pretty sure I switched back from RTL to FBWA (the direction it took was not towards home in any case) and also unsuccessfully tried to stop the rolling by stick input. When it started to fly inverted, I couldnāt make it roll back.
Could it be that AHRS was re-initialized while the plane was upside down and therefore considered in the new normal?
In hindsight, I should probably have switched to MANUAL to get rid of the inversion - but I was simply too surprised. I never had anything crazy happen with my AP planes beforeā¦
The plane was running AP 4.3.0-dev (from July 1, forgot to update since then).
Thanks for any guesses, and I hope I will manage to retrieve the log laterā¦
Not possible, it was a very tight fit (ZOHD Drift). Unless of course the hatch came off and it was dangling out of the planeā¦ Vista unit was on top of the hatch. But of course I had it secured with additional glue, not relying on the magnet alone.
Still, maybe this theory shouldnāt be discarded too soon. Something heavy dangling from the plane could also cause some centrifugal force leading to an inability to stabilize after a roll? Just wild guessingā¦
What about a servo failure? Normally a flight controller wont command a roll, so if the attitude was being read correctly and your flight controller was still mounted, then its hard to see any other reason for the plane to roll inverted in rtl. I would normally say the direction of the stabilisation was reversed on the roll axis but you would have seen this probem in fbwa also if that was the cause. U need log data
Indeed I do. There is some wind outside now, so Iāll make an attempt to retrieve it later today.
Servo failure is also possible. I immediately swapped the stock motor and ESC it came with, but not the servos. Theyāre probably not of the highest quality either. But I never had a failure with stock servos before, so I assumed it would be ok this time also. There was no sign of any mechanical problems before this happened.
Looks like servo failure, the linkage or servo probably failed and it started to loose control, with elevons if you loose either servo you loose both roll and pitch control. It looked like it was trying to hold the nose up but it started rolling due to only one elevon acting on the plane, once its past 90 degrees its lost pitch control.
If it was some setting reversed you would have seen it roll inverted much faster than it did, but it slowly lost control makes me think either the servo failed, lost power or the linkage came loose.
Sounds very plausible. Unfortunately still no hard data yet, first attempt to recover the plane was unsuccessful. Wind was not enough to blow it down. Thunder & rain tomorrow, maybe thatāll helpā¦
By the way, it was still beeping high up there after 24hrs, thanks to the Vifly Beacon which Iād really recommend because of that.
Good news! Finally managed to recover the plane today. There was a huge thunderstorm in the area yesterday, it was even raining planes from the treetops.
I have not tried powering anything up yet, first want to be sure itās totally dry. I did eject the sd card and it contained a readable log of the incident, which is of course great. The log will be posted momentarily, itās just that I prefer posting anon logs and the function is still buggy in MP. Is there any other way of scrambling the GPS data before posting?
Visual inspection: Pushrods still attached, ailerons totally intact. I will try the servos once Iām sure theyāre dry. Of course, in case they donāt work, there will be no telling if they already failed during the flight or were damaged by rain later. If they do work, the mystery will deepen. One wing came off completely but that certainly happened during the fall from the tree. Also checked the aileron solder connections, all still attached.
Here is a preview, roll vs. desired roll. Maybe someone with more experience can already tell from this if it was servo failure? Anything else to graph to confirm it? Looks like there were no error messages at all from AP, so I assume the cause of the incident has to be something mechanical indeed.
That image youāve posted here sure does support the idea of a mechanical failure. The FC was commanding roll, but the plane was just doing itās own thing. The log will be interesting to look at.
(I opened and resaved the anon log with APMPlanner, now it also works in MP, hopefully itās complete.)
In other news, everything still works! In fact if there werenāt pieces of foam missing from a wing and a hatch, it would even seem thinkable to glue it and fly it again. Incredible, after 3 nights in a tree + thunderstorm. Especially great because this was the first craft I had put a Vista on.
But āeverythingā includes the aileron servosā¦ So now Iām really curious what the log reveals.
Could be something loose in the servo so the fault is triggered with vibration or high loads, have you checked all the connections to make sure none are easy to wiggle loose.
Now that i have seen the log and looked back at the video, the flight controller was actually doing a good job of still flying the plane pointed at the horizon even with hard over elevator, you just didnāt have any directional control.
I just made some experiments, shaking the plane hard, putting it upside down, even slightly blocking the servos with a finger, all while someone else was continously moving aileron left/right (not easy to use a servo checker here as everything is soldered). I also removed each wing to test the integrated connectors and put it back on. Not the slightest sign of a malfunction.
From the log, could it be possible that the BEC failed temporarily and surfaces were stuck in a certain position for a while? I had already pulled on the solder connections, nothing came off.
If the BEC failed and it lost elevator the plane would have immediately nose dived when it started to roll, but the flight controller was using the elevator to keep the plane pointing at the horizon the whole time so the elevator servo was operating the whole time.
Right, of course, the elevator had to move to compensate for the rolling. Well, then Iām at my witās end. Will this remain one of the great unsolved mysteries in RC aviation?
I guess itās not imaginable that the FC sent values different from the desired ones to the servos.
Iāll check all the connections on the FC once again (servo cables are soldered, no connector involved except for the one on the wings), but I had not found anything loose during previous checks.
Replacing the servos (I guess Iād better replace all 3 thenā¦) would give me some amount of confidence back while flying this plane, but of course I would prefer it if the cause could be identified 100%.