Althold issues

ArduCopter Copter 3.1
1

Guys,

I need some help. Flying 3.1.3 on an APM2.5 board. Althold works reasonably well as long as I keep the quad hovering. As soon as I move forward with moderate speed (shallow pitch angle) the quad decends. As soon as I stop it climbs to the original altitude again. Throttle stick is a mid position all the times. I also noticed that this behavior is less prominent on fast forward speed. Please have a look at the first minute of the log, especially WPAlt, BaroAlt and ThrOut. Any ideas?

Thanks,
Michael

[attachment=1]althold.jpg[/attachment]

2

viewtopic.php?f=3&t=4576

3

What is missing?

4

Hardware setup infos.

5

You mean the quad? It’s a rctimer X650 with 10x5 props and SimonK flashed ESC’s.

6

Genuine 3DR APM 2.5 with enclosure or RCTimer clone naked, without enclosure?
If the latter, that’s likely the reason for your problems.

Edit:
Vibrations might play a role too here.

7

Correction:
Your vibes definitely play a role here! Y vibes (green) are way over the limits. Also X (red) and Z (blue) are bad! You should get some anti-vibration mount!

[attachment=0]vibes.png[/attachment]

8

rctimer APM with enclosing and foam on the baro. Vibrations: well, I had setups with good vibration levels in hovering. But on both of my quads (rctimer and a TBS Discovery with 3dr APM) vibrations are a completely different story in forward flight. The aerodynamic effects seem to have a much larger influence than static noise due to props or motors. Anyway, is there a way to have less contribution of the accelerometers taken into account for the althold controller? I mean baro alt and rel_alt don’t look too bad.

9

Ok, I followed your advise and reduced vibrations. This is a graph of a 3 minute loiter without any control input. Looks good enough?

[attachment=1]loiter_vib.jpg[/attachment]

With this setup I try to illustrate my althold issue again. After loitering for 3 minutes with a very stabel altitude hold I switched to alt hold and just did a couple of moderate speed straight passes without any throttle or yaw input. Just pitch to maintain speed and a little roll to keep it in a straight line. I can see the copter loosing altitude while picking up speed and the copter climbing again after stopping. The graph clearly confirms this. As baro alt reflects the true behavior I’d say the alt hold controller has a problem. What do you think?

[attachment=0]comp_ah_vs_speed.jpg[/attachment]

10

@ StefanG
could you suggest what is the most likely cause of Z axis vibrations? my X&Y are under 1, but Z gets very often to -6 and -18.
as far as I understand Z axis is mostly responsible for alt hold issues? am I correct?

11

Triple check your cabling. In one case, I had a rather stiff cable transferring vibes to my APM from the frame.
In my other cases where I had persistent z-axis vibes which didn’t go away through prop balancing, it turned out that the arms were too soft. But that was with low-kV motors and huge (17") props. IIRC your quad has CF arms and 10x5 props sound like 800-1000kV, so I don’t think that would be a problem. I would suggest experimenting with different mounting types. Some guys seem to have good results with makeup-pads - check the General hardware/multirotors forum.

Sent from my Ainol Novo 7 Fire using Tapatalk

12

@artemen: thanks for capturing my thread

@StefanG: thanks for not considering my questions

Ideas anyone?

13

When I don’t answer, it means I don’t have an answer. It doesn’t mean I didn’t consider the question.

Sent from my Ainol Novo 7 Fire using Tapatalk

14

That wasn’t me. This was the guy’s question that had captured my thread. My post/question is the one prior to his.

Thanks,
Michael

15

Your problem is the barometer. This is an issue common to almost every multicopter, particularly ones that do not have an enclosure.

16

Thanks for your teply. My APM is in the 3DR case and pink foam is also installed. Comparing what I see in flight and what the log is telling me I’d say the baro returns correct data. The altitude change I am observing correlates quite well with the baro readings. So I think it might be a flaw in the alt hold controller maybe in conjunction with angular boost. Absurd idea?