When the nose of my quad points south, the Planner (or copertGCS) say North. - all directions are good, but wrong.
The APM2 is mounted so that the SD card is in front left corner. - and GPS up, of course, do please don't ask that :)
it flies well, - but needless to say is useless to position hold, RTH.
how is this even possible ? - I recalibrated Compass, and also flew with learning on - does not help - the directions are good, but rotated 180degrees.
Sounds like a defective compass. I'd contact the store customer service for a RMA: email@example.com
ok, calibrations are very random, after doing several calibrations, I hit one that is pretty useful, will try to use learning to fine tune it.
And now I lost it at sea ! (with a 3DR frame and Bluetooth adapter)
Yep, after a few minutes or reliable hovering, and position hold , it started to drift away, living on an island, it did not took very long time to get well over water. after a while, It increased forward pitch to a point where I were unable to level it off by holding max "back" on the stick, PH and RTL also failed to level it, and it hit the water some 100 meters from shore.
I do not know what went wrong, or why it happened, because that was way out of BT range.
Complete loss, 3DR, new APM2 , BT module battery ETC is on the bottom of the sea. - wish I could learn something from it.
Andke: very sorry to hear that! We're you able to get logs from at least the beginning of the flight when it started to drift away?
Hi, this - in "quadcopter" folder is unfortunately all I have, did not had chance to look at it, or see if Quadcopter logged more, see attachment
now I checked the files, they are useless. no tlog data. does not seem like "Copter GCS" is recording it.
Sounds like the AHRS drifted.
Was it windy?
I uncovered a bug a while ago in arduplane where I believe it is using an incorrect model to correct the accelerometers to obtain its orientation with respect to the earth, mainly manifesting in windy conditions. It caused me at least one crash that was similar to what you described (pitch went so far off that it hit the limit and dived into the ground.) I don't know if it applies to arducopter or not and I don't know what progress was made with it after I went away for a bit. I don't think any was.
However, I was flying with APM1 at the time, and APM2's gyros drift a lot less, which basically lets us limit the rate at which we make corrections more, so an APM2 should take a lot longer to crash. I still don't think it is advisable to fly in any kind of wind at all, though. Which is kind of sad.
Pity there are no logs.
Yes, it was windy.
And yes, I've earlier seen the same problem with APM1 (that I now use for a plane, and it works fine)
it seems as if the quad fights wind for a while, no matter what AP mode, - it will eventually get bad idea of AHI)
We think there's a startup bug with the mag on APM 2, which we're tracking down now. If you press the reset button after you plug in your battery each time, it should be fine. We'll push a fix with the next code rev.
great - I hope that it's rusted remains will wash up on shore after storm in the fall, then I'll post a picture of me pushing reset :)
"you bastards - you killed Kenny !" (is it too later to name it that ?)
oh cr@p - I just realised what must have gone wrong by looking at my last config file, + rethinking the events.
My config file is nothing much to speak of, default values on everything, as it was a standard 3DR frame.
I have long experience with autpilots, both MK, OpenPilot, RVOSD and APM1 (and now - catastrophic experience with APM2)
Important info 1: I've seen on earlier flights that "idle" position of roll/pitch stick (mode2) could "move" - and require bigger input if I flew manually and held pitch/roll over time to fight wind - no I did not use autolearn, but leveled the 3DR properly before.
Important Info2 : I "recalibrated" crazy number of times, and "by luck" struck "north-south-east-west" that was good. - most calibrations were mostly random - only APM2 USB powered, no other magnetic fields/power nearby by battery power (carefully verified that).
Then - most likely my doom - I enabled Compass_Autolearn
>3 minutes into flight:
I took off and it held position nice, I tested RTL and PH. and it somewhat circled
- I went back to Stabilized control, and held it in place with a little climb while fiddeling with my Galaxy note (CopterGCS) to see that was going on(telemetry) (checked AHI screen and gyro/accelerometer curves, did not see anything crazy) while doing so, I held it in place using right stick (mode2) contantly applying input to counter wind.
after ~20sec: I switched back to PH - the 3DR took off directly away from me ! (just like before - with inverted compass)
I switched to RTL, and back, it kept flying away from me at good speed. I hoped for it to circle (bad compass) or stop. Maybe 6-8sec leter, 200 meters away I switched to stabilized.
Now it did not want to level off !! (I could still see its orientation because of powerful LED's.)
I tried to apply full input in the opposite direction. - no luck.
I yaw'ed it's pitched side back to home, to get it back over land, (knowing that would obscure the sight of LED strips.)
By now my maximum stick input was not enough to counter it's will to roll/, I lost altitude too fast. The last meters above water, I released stick and applied full throttle. in some soft of hope that it would level off and climb - nope - at "full speed" - more forward than climb, it hit water.
So this is a combination of two disastrous errors.
-Bad compass, that moved it far away instead of "home" or "hold"
-some bizzare error that makes the APM2 translate it's attitude - or stick input differently after a while of a constant stick input. (like when you fight wind while hovering in stabilise mode)
3DR, brand new APM2, BT module, battery , Li-Po low-voltage beeper, and Graupner GR16 receiever. R.I.P.
Wish a had a good video of the splash.