Please help diagnose my APM 2.6 flyaway

zub January 18, 2015 at 9:23am

Thanks everyone for your help and advice on this.

I have now:

- updated to arducopter 3.2 firmware

- changed the anti-vibration mounts, with 2 foam mounts in each corner, and have one of those suspended platform on order

- enabled IMU logging

- clearly I needed to re-do the accelerometer and compass calibration, but have also spent a lot of time on the power monitor calibration, just in case the APM thought there was low battery voltage and caused an RTL command.

I have just done an extended hover in very calm conditions and the IMU data seems good: the X and Y accelerations are between +2 and -2, which is within the +3 and -3 in the manual, and the important Z accelerations seem to be -7 to -11, which is within the -5 to -15 range referred to in the measuring vibration wiki

http://copter.ardupilot.com/wiki/ac_measuringvibration/

However I would note that in the vibration damping wiki here:

http://copter.ardupilot.com/wiki/vibration-damping/

... it says

"Vibration needs to be less than 0.3 G in the X and Y axes and less than 0.5 G in the Z axis."

I don't know what units the ACC_X, Y and Z are, if they are in m/s/s, then -5 to -15 sounds like c. 0.5 - 1.4G (if G is 9.8 m/s/s), which is a bit confusing.

Anyone shed any light on this? Does my ACC_Z look ok? I have attached the log if anyone is interested, but recognise lots of people have helped me already, so thank you.

Anyway next on the list is to do the autotune.

2015-01-18 16-43-56.log

zub January 10, 2015 at 3:29pm

Thanks so much for your input, it is really great having help like this. How to diagnose a flyaway is painful, as I have lost a bit of confidence in the quad because of this!

Running through the answers:

Mike, pyrosamz: I am not sure I had a GPS glitch, as no. of sats and Hdop look ok. I have attached a screenshot of the graph of sats, hdop and altitude, where you can see that sats and hdop were steady (and good) whilst altitude was increasing.

Also there were no errors. I searched for ERR and ECode in the log and there was nothing to suggest the GPS signal was lost, which caused a switch to RTL. On the contrary, looking at the logs I go into LOITER at about 2,500 on the x-axis, and then at 3,000 it starts the flyaway, with me hitting the RTL at 3,150 to get it under control.

AndreK - thanks for your advice. I don't have IMU logging, I will now! I will also update the firmware. However I am not sure why v.3.01 that I was running would have caused this whereas the latest firmware is ok. It is the cause of the problem I am struggling with. However as well as updating the firmware I will change the anti-vibration mounts to help with z-axis vibration as you suggest.

I have also removed two of the props which I could feel a bit of slip, and will re-mount them.

Thank you guys for your advice and help. This has been instructive - if painful! I have a Blade 350 QX2 quad (amongst others) which just flies out of the box, which I have confidence in. My APM 2.6 has caused a wobble... but I'll get over it, not least I have a tri-copter build coming up on an APM 2.5, and fly my plane on an APM 2.6 too!

Anyone else got any ideas?

Zeke Zero > zub January 11, 2015 at 7:24am

Zub,

I have suffered from the same issue with APM and Pixhawk (which I thought would be immune because of fancy dual sensors and clever EK Filter..).

I would strongly suggest checking your IMU Z vibrations as Andre says. Check the Arducopter instructions for reviewing vibration. Z axis should be between -5 and -15 in a hover: http://copter.ardupilot.com/wiki/ac_measuringvibration/

Also have alook at http://copter.ardupilot.com/wiki/vibration-damping/

All the best,

Zeke,

Worcester,

UK.
- zub > Zeke Zero January 11, 2015 at 1:44pm
  
  Zeke
  
  Thanks for your advice. I have now mounted the APM on two foam pads, plus a third "softer" pad. Reading through the info on vibration damping, I'm not sure what to make of it! It goes from one school of thought that says moongel is great, to another which says mount directly on the quad! I have now enabled IMU logging, I will take it for a flight this week and have a look at the data.
  
  I really appreciate everyone's help here, thank you.
  
  Zub
  - Zeke Zero > zub January 12, 2015 at 2:27am
    
    No worries Zub (Zeke / Zub lol..)
    
    I found these to be great: anit vibe mount
    <http://www.goodluckbuy.com/glass-fiber-flight-controller-anti-vibra...>
    
    mounted with this between airframe and mount and between mount and
    
    controller.
    
    As with all things APM there are a hundred different ways to do the same
    thing!
    - zub > Zeke Zero January 12, 2015 at 2:46am
      
      From one "Z" to another...!
      
      Perfect, thank you. Just bought a few of those anti-vibration mounts, and I have some foam pads already.
      
      Much appreciated.
      
      Zub

pyrosamz January 9, 2015 at 3:57pm

In advanced parameters list what's your rtl hight set at? I think default is 80m.if you have a rc failsafe set up correctly the first thing that will happen is the copter will rise to that altitude in the event of failsafe be initialised.

Andre K. January 8, 2015 at 10:29am

you have no IMU logging, but behavior like this can be trigged by too much Z vibrations.

At least you can forget GPS problem. You can clearly see the compensated GPS.RelAlt pretend to drop, even when CTUN.Baralt shows that you are climbing.

Also, it seems you should upgrade, I think you may have very old ArduCopter.

Mike Kelly January 8, 2015 at 7:43am

Since no one else has answered....

If you had a gps glitch or something else that triggered an RTL failsafe it can appear as a fly-away. It shoots up to reach the safe altitude to start it's return.

Discussions

Please help diagnose my APM 2.6 flyaway

Replies

Activity