Servers by jDrones

Yaw on large aircraft steals lift

I’m trying to hover a large quadplane for the first time in QSTABILIZE (wingspan ~ 2.7m, about 18 kg MTOW) and every time the autopilot tries to yaw a significant amount, it loses a lot of lift, many times to the point where it doesn’t produce enough lift to stay in the air / take off again even at 100% throttle.

A couple of videos of the behavior (QSTABILIZE mode):

  1. https://photos.app.goo.gl/LEdpwssudmawG9DB2 - aircraft takes off, yaws, loses lift, forced to land
  2. https://photos.app.goo.gl/pO83bSPtmdXuAvDA2 - aircraft cannot take off, too much attempted yaw on ground

Notable events in the logs:

  1. 'EKF Yaw Resets": every single takeoff is followed by an EKF Yaw reset message. I’m using three compasses (1 on the Here GPS (primary), 2 Pixhawk 2.1 internal). I’ve recalibrated in the field twice without any change to these messages.
  2. Q_A_RAT_YAW_I and Q_A_RAT_YAW_D was set to 0 throughout to isolate the P gain.
  3. Q_A_RAT_YAW_P has been exercised through the entire range of 0 to 0.55, from 0 - 0.2 there’s insufficient control of the aircraft, at 0.3 there’s decent control of the aircraft but once in
    a while it yaws to the point of losing most lift, and as I go towards 0.5 almost any yaw input causes the aircraft to lose most of its lift.
  4. Q_A_ACCEL_Y_MAX was reduced from 27000 to 15000 to 10000, mainly to reduce sharp drops in altitude due to large changes in yaw output.

I just need the aircraft to reach a stable loiter and control its heading in wind before I can begin testing transitions, and I don’t have much experience with control theory and PID tuning. I would really appreciate any help with this :slight_smile: For example, is there any way I can reduce how much total yaw output the autopilot is capable of commanding? Or is that even the right way to proceed?

Logs (all similar, just different YAW_P / ACCEL_Y_MAX):

  1. https://drive.google.com/open?id=1veMu4OIWYxUfcCsSoQZ6ZjpXkKEVIL9d
  2. https://drive.google.com/open?id=1ZZy_PRWy3WW2-V51MFZhbjRL3Ij2YXbd
  3. https://drive.google.com/open?id=12h6K8e4oryrVsswRuUhQAf2nu6KS1aBl

Many thanks in advance.

So it’s basically very underpowered ?
And the solution you are looking for is …?

That’s a possibility.

One of CanberraUAV’s quadporters had fairly similar sorts of issues.
Ensuring all of the vertical lift motors were actually vertical through
the judicious use of washers underneath the motors mounts helped
enormously.

Thanks @peterbarker, will look into that. Maybe that’s what’s robbing our effective thrust :slight_smile:

@Andre-K the MTOW is 17 - 18 kg, but the quad motors can produce 24 kg of thrust total (individually measured). So I see a couple of possibilities:

  1. It’s still too underpowered.
    Then, I’d like to see what I can do to only cause small yaw outputs and just accept that the aircraft will be sluggish. Specifically, to reduce the maximum overall yaw output by limiting the maximum yaw rate such that even if the maximum yaw rate is commanded, the aircraft has sufficient lift to fly. I’m aware of how to reduce the maximum yaw rate acceleration, but not the maximum yaw rate.

  2. It’s not underpowered: something else is the problem
    The EKF yaw reset messages constantly appear in this large quadplane but do not appear frequently at all on my test quadplane. This feels indicative of another problem in the system.

Looks like the parameter you’re looking for is Q_YAW_RATE_MAX. For auto missions, you will also want to set ATC_SLEW_YAW. You may also want to experiment with Q_M_YAW_HEADROOM.

That said, if it is truly a lack of power that is causing this, then yaw won’t be your only problem. Limited throttle room means your maneuverability and wind resistance will be likewise limited.

Great, thanks! I can definitely try those parameters.

Definitely agree on having future problems if it’s really a lack of thrust. I hope that isn’t the case…

Cool. Graphing (e.g.) RCOU.C1 through RCOU.C8 will tell you quickly
enough. May also tell you if you are running out of power.

If you send through a log file which shows this I can add a test to
dronekit-la and/or LogAnalyzer.py

Looking at your log, just like your description of the event, its inderpowered.
you can reduce or zero out Q_M_YAW_HEADROOM to get a little bit more thrust… or you can reduce local gravity :slight_smile:

…or just generate more thrust.

Okay - will also try setting Q_M_YAW_HEADROOM to 0!

Maybe one day I can find one of those local gravity manipulating machines on HobbyKing :smile:

Hi Sriram,

Flight behaviour looks exactly like one of first quad builds before I understood much about lift to weight ratio etc.

Gut feeling tells me you’ll need more lift to get a reliable behaviour - even if you manage to tweak some parameters and squeeze out 10% more lift out of your current config - it does not look like it could be really stable (think strong winds) like that…

What you maybe could do is convert it to a X8 setup, i.e. add 4 motors/props/ESCs and couple of cables - would (nearly) double your lift without need to change everything (and allows you to reuse your existing drive parts).

I personally would use a calculator like ecalc though to ensure any change to your config would yield enough lift before buying more - doubling existing motors/props might not be sufficient…

Just a thought…

Christian

P.S. A nice plane btw, like the design!

Thanks for the advice @CLabeck, I’ll bring it up to the mechanical team. Did you end up using your initial quad at all?

It is underpowered for sure. If you have an MTOW of 18kg and a total thrust of 24kg, that would be a thrust to weight ratio of 1.33 : 1. All motors need headroom to stabilize the vehicle, so you can’t use 100% power. Yaw is already weak on copters, but it is much worse on quadplanes with vertical stabiliizers and other surfaces causing drag. Yaw control is done through torque, so to yaw in one direction two motors rotate faster and two slower, diagonally opposite. With a thrust to weight ratio of 1.33 : 1, there is not much torque difference to be generated and even small yaw corrections will eat up the small thrust excess there is. I would suggest a ratio of 1.7:1 at least.

Hi Sebastian, for the future, how can I derive a good rule of thumb for a quadplane’s thrust to weight ratio to give myself enough yaw headroom? Did 1.7:1 work for you personally on a large quadplane?

Thanks for the tip!

I do not own a quadplane of that size. The principles stay the same, no matter what the size is. Just the consequences (the size of the hole in the ground i.e.) are different. I use ecalc.ch to calculate my setups. It does not have a section for quadplanes, but you can combine the plane and multicopter calculations.
My MiniTalon quadplane shoukd have a thrust to weight of 2.1:1.

Try changing the gravity center to the lift center of the four propellers, you should have a better result. Always check these RCOU.C in log, make sure all the four motors are equal loaded.

Servers by jDrones