3 Hours of flawless flight testing using Stab, Stab Simp & Alt Hold, Try Loiter and Bam, Flyaway.
Well it could have been a lot worse.
I was doing some birds eye view shots of my parents house, Turnigy HAL Frame, converted to coaxial octocopter, 3DR APM2.5+ with the latest software.
Two successful flights before this event.
I should tell you about the only other time I have used a GPS related mode, which was RTL, the HAL came back about 15 deg off course and when it reached the takeoff latitude (or longitude) it seemed to realise it was off course and started to yaw wildly left to right about twice a second, i clicked back to stab and recovered it to a happy landing. That event was about my 3rd flight, I have since done dozens of flights accumulating around 3hrs with this frame and controller.
So this event in particular after the aerial shots I thought I would try loiter, the HAL started to climb up and move away from me, I wasn't too worried and let it carry on to see what would happen, it then seemed to change direction in a hurry and fired back towards the house and tree, in the time it took me to realise and flick back to stab it was too late, unrecoverable, took out a few branches and landed heavily on it's side. amazing to only bend the arms and break two propellers perhaps. images attached.
Attached (I THINK) is the log, I don't understand how to veiw or decipher this, but would appreciate any help in doing so.
Thanks.
NB File 116 is the crash log and the other two a couple of happy flights, well, in 115 it did a crazy yaw on the way down but I put that to my descending too fast. Given the narrow air frame but lots of power it is a bit of a beast to tune.
In case it helps, here is (within a meter or so) where it actually was when I engaged loiter
And it hit the tree about here
Replies
As a sidenote, I'm intrigued by your design.
The x8 I can understand, but I see you have all the ESC's out on the pods as well. Was there a reason you put them there, rather than inboard, near the centre of gravity/balance?
I only have a traditional quad, but I've been trying to keep my design as mass centralised and balanced as possible, with the aim of improving stability.
It's not a critism btw - I'm just wondering if there's an advantage to your approach is all.
New Zealand has a rather magnetic high declination at 24.5deg.
I had left the apm to auto configure the declination using gps... this failed though and set my declination to 0.405!
I think perhaps the copter failing to maintain loiter and being way off course the one time I tried RTL may have been related to the Dec being out by 20Deg!
Hey Charlie,
In addition to what Randy said, how did you have loiter configured? I made the mistake (and just once mind you) of leaving the loiter radius too large (30m instead of 3m). So whilst the quad was flying around in this circle it thought everything was fine - until it decided to get to low to the ground.
Charlie,
The 116 logs are almost empty so perhaps that's from a different flight or maybe logs weren't recorded for some reason. In anycase I had a look at the 114 and 115 logs and you roll-in vs roll and pitch-in vs pitch, are not tracking well. Comparing these two is a good indication of how accurate your roll/pitch is tuned.
Bad tuning could lead to bad loiter performance because the copter won't do exactly what the navigation controller wants it to do. I think you should focus on the Rate Roll and Rate Pitch P, I and D values.
Rate Roll/Pitch P: 0.14 <-- should be ok
Rate Roll/Pitch I: 0.03 <-- low, please try increasing to 0.1
Rate Roll/Pitch D : 0.008 <-- slightly higher than default
A few other things:
1. you're flying 2.9-rc5 but you might as well upgrade to 2.9.1. It's the same but it has a bug fix in case you want to use the throttle failsafe
2. try setting your INS_MPU6K_FILTER to 20. I think we've decided this is better than the default of 42hz which is what you get when that parameter is zero.