PixHawk Poor Altitude Hold in Wind

This morning I flew two identical missions with my PixHawk- based quad. The first was in near-still conditions, the second in about a 4m/s breeze. On the first mission the quad maintained its required altitude at 35m very well. In the second mission I noticed surging and dipping in altitude of up to 3m. The quad always recovered its altitude but this ‘dolphin’ flight is inefficient and also destabilizing.

Also, yesterday I flew the same quad in about 6m/s wind and the effect was far worse. In fact it ‘surged’ to 80m at one point, despite AH being set at 40m manually and for the mission. It seems as if the surging worsens exponentially with wind velocity.

Both times I didn’t have a log, but I will repeat the missions and supply logs if necessary. I just wanted to find out if this is an known issue. My details:

  • AC 3.1.5
    -Approx 750mm X-Quad with 3508 380kV motors, 15 inch props, 30A HobbyWing HV ESCs
    -Flying HV on 5S 3000Mah Lipo
    -AUW around 2kg, which suggests the quad is over powered to some extent, but I want endurance and not weight-carrying ability.
    -The quad has a ‘dome’ cover over the PixHawk but the effect was the same with or without the dome
    -I haven’t used Autotune because I’m using low kV motors and the Wiki warns against this. The quad flies well enough though, it descends well through its own wash, which is always a good indicator. Its not as stable as a small-prop model but that is expected given the drag area of the larger props.

Any feedback appreciated

I now have some logs and a screencap to post.

The TLOG shows a surge beginning at about 49%, where the quad rises out to 60m from a mission-set altitude of 35m. It hits a 10m/s climb rate at one point! Then it quickly sags out again, hitting a -5m/s sink rate over the space of a few seconds, to overshoot its mission altitude to about 30m. These incidents are not just stressful, they are wasteful of power.

Analysing CTUN on the log it appears as if there is a long lag between ThrOut and BarAlt, implying slow ESC response. These are stock-standard HobbyWing Platinum 30A ESCs, I haven’t flashed them because I had strange results when I flashed some HW FLyFun 30A ESCs with BL-Heli.

I’ve also attached a screencap of the incident. It appears as if the long lag between ThrOut and the actual change in BarAlt is the cause of this ‘overshooting’. I will put my PixHawk onto another frame with different ESCs and see if it still happens.

BTW this is a geniune 3DR PixHawk. Order #: R341543424

Change the subject to “bad vibrations give poor alt.hold” - then fix it :slight_smile:

Thanks Andre K for taking the time to look at the data and reply. When I built this quad I did all the initial tests and vibration was one of them - albeit in still air. At that point the vibration levels were well within acceptable.

As you can see, for most of the flight the vibration levels are within normal and the excess vibration is a spike rather than a constant feature. This, and the fact that it flies very well in calmer conditions implies it isn’t a normal vibration but something that occurs when triggered by a combination of events. This also implies that I may not be able to solve it by using different mounting foam?

Do you have any other suggestions? I’ve mounted my PixHawk on a different frame and tomorrow I’ll do another test flight.

If you also viewed IMU.AccZ - you’d see crazy high vibrations around line 8-10k , but also now and then.
Z-axis accelerometer gets confused, and what you see is a classic :slight_smile:

Try for youself to plot the data, thrust me, that’s you problem.