Unstable QuadPlane

My Skywalker X8 quadplane has become quite unstable and I could sure use some help figuring out how to fix it. I was running 3.8b4 and upgraded to the 3.8 release but both had issues today. Logs of the flight are at https://www.dropbox.com/sh/7kjhq49gl16jlbg/AACN2W7RGRxOEKZGcOWIUqrEa?dl=0

On VTOL take off the plane goes to 30 meters, pauses for a second, then the rear motor comes on, then it flips (360 degree roll) after which the rear motor increases and once forward motion starts it transitions to forward flight. It does this in both Q_Loiter takeoff and manually switching to FBWA or in an Auto flight with VTOL as the first command. I think it actually flipped twice on one takeoff.

The plane doesn’t track at all during a grid mission generated by Mission Planner. It misses waypoints and turns well past them. When it finally turns it turns the opposite direction expected towards the outside of the grid rather than the inside. I’ve used the same type of plane with the same type of missions before without any problem.

During RTL, which I set to use hybrid mode, the plane doesn’t descend from the last waypoint on its way towards the landing point. Then the vertical motors come on and the plane rolls wildly until it gets to around 30 meters high where it becomes stable and does a perfect landing.

If you are interested in the build or want to see a video of it flying properly I blogged about this plane not long ago https://discuss.ardupilot.org/t/a-tale-of-two-quadplanes

Hi Darrell,
You’ve got the same problem you had on the previous plane, where the battery voltage suddenly drops. When it flips over, the voltage suddenly drops from 3.8 volts/cell to 3.1 volts/cell. At one point it drops to 2.5 volts/cell.
The motors will be shutting down at that voltage. I don’t know what causes this in your plane, but it definitely can’t fly with such low voltages.
Cheers, Tridge

I also note that POWR.Vcc is fluctuating from 4.5V to 5.1V. Whatever is powering your flight controller is not happy.

I think that’s probably it. I’m beginning to think that running the vertical motors off the same power supply as the forward motor simply isn’t workable. I’m going to separate the two and do some more tests.

Any idea why the plane won’t track a grid in Auto mission and rolls so badly during a RTL vertical section?

It’s powered by a new dedicated BEC on the rail and a new Power Module on the front. Not sure what else could be done to improve it. The BEC is powered by it’s own battery. But the front PM is powered by the main batteries so maybe it just can’t handle the sag. So separating the batteries might help it as well.

There seems to be a severe asymmetry in roll in fixed wing mode. When rolling from right to left it tracks reasonably (not really fast, but good enough). When rolling from left to right it tracks extremely slowly, causing slow turns.
I notice that SERVO2_TRIM is 1379, which probably came from SERVO_AUTO_TRIM learning. That implies there is a strong asymmetry in the airframe. Basically it is left-leaning. It won’t vote for Malcolm Turnbull :slight_smile:

yes, I think that is the most likely cause

I have another set of wings that have better elevons with more throw. I think that they may help. It’s still a bit odd that the plane turns the wrong way when doing a grid survey. I’ll do some more tests next week and report back.

Using another set of wings that I had with more rigid elevons and a bit more throw on them seemed to solve my tracking issues. I’ve also rewired the plane so that the vertical motors and the forward motor run off different batteries. While I like the idea of having only one set of wiring it does seem that even with high C batteries that there are voltage issues during transition from vertical to forward flight caused by voltage sag when all the motors are on. This way the two can’t effect each other and I can also run the vertical motors off 6S rather than 4S batteries which will give the vertical lift a bit more authority/thrust. So I’m considering this solved for now and I’m off to do some duration flight testing.