altitude shoots up in alt_hold

For anyone experienced at interpreting logs, see attached...

I'm assuming that I have a vibration issue, just wondering if anyone has any other opinions. Stabilize mode first, then switched to Alt_hold. There are times when the Acc plots are ok, but when it shoots up to full throttle in alt_hold mode, the Accz does get a little crazy, not sure if these vibrations are a cause or a result of the throttle going to max.

When I heard it go to full throttle, I brought the throttle down, but DAlt starts to climb??, even though the ThrIN is at zero. If I understand it correctly, DAlt is the desired altitude, so I don't understand why this would be increasing?

Board is APM 2.5

Firmware is Arducopter V3.2

Any thoughts?

Thanks,

Ben

2015-04-13 13-40-15 110.bin.log

Randy - Is there a way to tell the difference between an accel-z reading drops and accel-z readings that are going nuts because of sudden vibrations from external sources?

I never saw this problem until working with small ships (FYI - the dual sensor Pixhawk doesn't seem to alleviate the issue). But with small ships, accels can be in the Good to OK range, something triggers vibrations, makes the ship accel-z variance really high, and the result is a run-away. Is there was a way to look at the variance of accel-z and then automatically discount accel-z and weight barametric more?

The trigger can be:

- a ship with Good vibes at typical thrusts but poor at full throttle

- a blade that becomes unbalanced or weak

- a shift in ship mass making one rotor work harder

- ???

I agree that we don't want to encourage poor ship design, but would also like to answer questions about poor alt hold rather than runaways.

I'm thinking that you don't want to stop the runaway (that would dampen alt-hold too much) but with the appropriate algorithm, stop a continued runaway fairly quickly (faster than one could switch to stabilize).

Thoughts?

Replies

Developer

Randy April 16, 2015 at 10:40pm

Re the desired alt, we don't allow the desired to be too far from the actual altitude so the desired can be dragged up as the actual alt climbs. It depends upon the WPNAV_SPEED_UP/DN but normally I think they cannot be more than 7m apart.
- MR60
  
  Forrest Frantz > Randy April 17, 2015 at 6:37am
  
  Randy - Is there a way to tell the difference between an accel-z reading drops and accel-z readings that are going nuts because of sudden vibrations from external sources?
  
  I never saw this problem until working with small ships (FYI - the dual sensor Pixhawk doesn't seem to alleviate the issue). But with small ships, accels can be in the Good to OK range, something triggers vibrations, makes the ship accel-z variance really high, and the result is a run-away. Is there was a way to look at the variance of accel-z and then automatically discount accel-z and weight barametric more?
  
  The trigger can be:
  
  - a ship with Good vibes at typical thrusts but poor at full throttle
  
  - a blade that becomes unbalanced or weak
  
  - a shift in ship mass making one rotor work harder
  
  - ???
  
  I agree that we don't want to encourage poor ship design, but would also like to answer questions about poor alt hold rather than runaways.
  
  I'm thinking that you don't want to stop the runaway (that would dampen alt-hold too much) but with the appropriate algorithm, stop a continued runaway fairly quickly (faster than one could switch to stabilize).
  
  Thoughts?
MR60

Forrest Frantz April 16, 2015 at 6:38pm

tell me about the flight: modes, for about how long, did you take off right away, or hover on the ground, etc.

i see in the log:

- incredible stability and low vibes for the first 66 seconds ... so low that it looks like you are still on the ground.

- then followed by 77 seconds of all hell breaking loose.

This reply was deleted.

ArduCopter User Group

Replies