I was really impressed with my first flight . I could take off and fly in stabilize mode so easily I decided to test alt-hold and loiter which were almost perfect until I switched back to stabilize then alt hold again.
The quad could not stay at altitude and climb and went down rapidly, it was not high but I had to switch quickly to stabilize to recover almost free fall, after few mode changes from alt to stab to loiter trying to understand, the flight ended by a crash on loiter mode (motor stops, free fall) .
I do not understand because the first part of the flight is successful and the behaviour is so different in the second part.
After repair I did another flight which was very similar to the first one .So it seems it is not related to calibration, I also checked that cables were correctly attached.
APM 2.6 3.1.5 with LEAH 6H GPS
I have checked the forum and have not found similar issue.
Do I have a wrong parameter ?
first part of the flight is in video (loiter mode) https://www.youtube.com/watch?v=uha02CAIWqA&feature=youtu.be
thx
Replies
After several days of intense search, I have the solution which is in fact the one that everyone had to face and I share it again for others.
I bought my APM already in its case and I wrongly thought that the barometer was already protected...
Also the case is in transparent plastic which is perfect for the sun to directly light the barometer micro holes ..
Sometimes the learning curve is a little too long ;)
So I painted in dark black the inside of the case = zero light from direct lighting but I have to check closely to see the inside leds. I will add external leds later maybe
I also added a compressed black foam on top of the barometer.
And for vibration dampening, I use the famous moongel in 2 layers
APM now works perfect as expected from it .. alt hold and loiter are now very precise : https://www.youtube.com/watch?v=hVUL9aGJ1f0&list=UUztTZ_HPaePBz...
I still have an issue to solve with autotune . each time I start the process, at the first or second roll, the quad flips and fall. but I guess this one is related to ESC sync despite the fact they are simonk flashed . I need some time to find this one.
I'm no expert but your log seems to indicate a voltage sag right about the time the trouble starts. The sag drops down to about 10.1 volt which would normally trigger low battery conditions.
Have you programmed your ESC's to disable their battery protection routines? I wonder if one or more started a throttle-down or similar....
Thanks for your help.I tried to focus on voltage and current issue.
Droneshare analysis says I have FS_BATT failsafe event and also no imu data due to vibration.
For the FS_BATT , I have this value as parameter in failsafe :
where it should have been 0 not to trigger failsafe
also
I have a warning for
where value should be between 20 and 80 . I dont know yet what this value is for.
For the vibration , I'm really surprised, because loiter and alt_hold where possible for half of the fly . and if you look at the video , the quad seems to be very stable.
By checking others users logs, I have found that their Vcc value is stable where mine is oscillating around 4,7 v.
I have not programmed ESC, but I had the same behaviour on another flight with full battery at the first 2 min of the flight so , it should'nt be related to empty battery.
Any other idea is welcome to help me.
Stephane, what ESC's do you have?
I'm still learning my way around log analysis but the DroneShare suggestion of vibration interfering with IMU data does make sense. You can see that the positioning data in NTUN gets locked and flatlines at the time I presume you started to experience issues.
Vibration is a big problem so I'd recommend you first investigate a method to reduce them. You can start with enabling detailed logging here: http://copter.ardupilot.com/wiki/ac_measuringvibration/
And start looking at methods to reduce vibration here: http://copter.ardupilot.com/wiki/vibration-damping/
HTH.
Thanks Matt for guiding my search
I have RCTIMER ESC :http://www.rctimer.com/product_1059.html
the quad is a RCTIMER Spider frame, so it is not high quality and may have some flex but I have seen videos of it on youtube and they seem to fly correctly with the original kit components.
I have changed the motors for sunnysky 2212 980
Apm is on top of this : http://copter.ardupilot.com/wiki/vibration-damping/#An_Excellent_3D...
Propellers are cheap 10*4.5 as i keep good and expensive carbone ones for the moment I'm confident with the quad stability ( I'm happy i did not started with them as I already broke 12 props)
I try to balance them correctly.
I need to wait for new arms to be delivered from china to start the next fly . I will log the IMU as stated in the vibration dampening wiki page to check the Gs.
Thinking about the flights, as I have behaviour changing during it , I should also check if the micro vibrations are not playing with the screws attached to the arms. It might be tighten at the beginning of the flight and progressively loose, explaining why ALT_HOLD and LOITER were possible at the beginning and going bad to crash at the end.
What I still do not understand is why motors stop completely to free fall.
I have found other threads related to my issue :
I don't know if it is related but it might also help others
http://diydrones.com/forum/topics/various-alt-loss-scenarios-on-alt...
Looking at the graph, it seems my BarAlt is totally wrong at the moment where my flight becomes erratic.
2 other flights :
1st = only STAB mode and 12 mn of flight , no issue, perfect flight in confidence
2nd = I climb to 20m and switch to ALT_HOLD, seemed to be ok so I started autotune, the process started and 20 s later all motors stop and free fall big crash .
Any idea of this strange behaviour ?
Thanks