Qstabilized mode not working properly

Hello,

I am currently working on a project. The VTOL has 4 tilt rotors, front two rotors are facing upwards while rear two rotors are facing downwards.

Question

  1. One motor out of 4 always runs slower at one throttle level and it is random (HOW)
  2. How QSTABLIZED works and why it is giving wrong input in level flight?

I am using 4.3.7 arduplane, frame Quad H with Pixhwak 6C.

The motor numbering and rotation direction are exactly same as recommanded Quad H configuration by Arudupilot. The gyro and compass and RC were calibrated properly and checked the movement on HUD.

3 FL (CCW) 1 FR (CW)
2 RL (CW) 4 RR (CCW)

Q_M_PWM_TYPE is Dshot600 for ESC telemetry (the values of rpm are not reliable though).

While performing bench test (load cell testing) with all four T-motors, they were running at approx same rpm with same PWM values.

Initial flight test yesterday, for tuning as described on Quadplane document. we started VTOL QSTABLIZED mode and were checking the RPM for pilot command and it was working perfectly.
(i.e., While giving roll right command, the RPM of rotor 2 and 3 were increasing and 1 and 4 were decreasing.)

But when there was no pilot command, all stick in middle (R, P, Y) and rolling quadplane right manually to right the QSTABILIZED was not working properly. Ideally, it should increase the RPM of 1 and 4 and decrease the RPM of 2 and 3, but during testing it was working exactly opposite, the RPM 1 and 4 were decreasing and 2 and 3 were increasing, making it roll more on right. Same was noticed during pitch down and pitch up.

Thanks,
Dhruvil

Did you do it on the bench?
If yes the test is meaningless, you are testing a closed loop system in an open loop way. It will never behave the way you expect it.

1 Like

yesterday, it was not on the bench, the all motors were connected to the quadplane and the plane was hanged to the crane via rope.

Not flying, so no closed loop.

Make sure the value for AHRS_ORIENTATION is correct.

1 Like

Ok, so once it stared hovering properly, lets say at 5 m altitude and initial manual runing of roll and pitch is finished, the qstabilization kicks in properly.

The AHRS_ORIENTATION is set to none. and pixhawk and gps pointing in flight direction. plus the movement on HUD as correct accorging to the plane movement (when done manually).
The yaw is also showing correct, checked with mobile compass.

But generally (not in my case at the moment) when we have fixed wing plane, and performing ground test before flying, the control surface moves and tries to stablized the plane when we give rotate the plane. shouldn’t it be the same case with qstablized also with motor rpm?

No, it must not. Lots of people misunderstand the problem and think like you. But it will not work like that. The math just works differently than you expect in this case.

1 Like

OK, I will try to hover again on monday and will let you know.

Not sure what you mean about tethered hover tests…
This is qstabilize on a tether: tethered qstabilize


Something similar as shown in figure but with a bigger model (11-12 kg).
Configuration is same, 4 tilt rotors, front two facing up and rear two facing down.

@kd0aij @amilcarlucas
So, we started new test yesterday, and something very similar as shown in figure but without hanging it to the rope from the beginning. The drone was on ground we started trying to make it hover, at +35 (-100 to 100) throttle it lifted up a little and then started turning in clockwise direction and then we disarmed immediately.
We started test again but at this point even with +85 throttle, the drone was not lifting up.

The motor and propeller combination produce enough thrust to lift the 8 kg drone and during second try batteries were still healthy (24.8 V 6S).

As mentioned before one rotor always runs slower than other at one throttle level.

For the starting point for PID tuning, PID paramters of one drone with 10 kg weight are used in the last 3 testing. Before that, the default values were used and still had same problem.

Here is the latest parameter list.

parameter_10-10.param (20.1 KB)

Can you post the dataflash log from that test?
Have you verified that the motor rotation directions are correct? If they are not, then yaw control won’t work right.

1 Like