Servers by jDrones

Tiltrotor support for plane


(tridge) #283

It’s about 10 degrees on my convergence.

yes

I don’t know if this will work at all, but you could try setting the SERVOn_MIN/MAX values to prevent the motors moving all the way forward, and then set Q_TILT_MAX=90 and set Q_ASSIST_SPEED=100.
I haven’t tried it though.


(tridge) #284

really nice! I hadn’t heard of DashWare till now. Nice system!


(tridge) #285

A quick note on how to configure tilt-quadplanes with vectored yaw.
The two key parameters are:

  • Q_TILT_TYPE=2

  • Q_TILT_YAW_ANGLE=10
    The Q_TILT_TYPE=2 means that you have a tilt rotor that uses motor tilt for yaw (ie. vectored yaw).
    The Q_TILT_YAW_ANGLE is the angle in degrees that the tilt motors are tilted back when at their full up value (as set with SERVOn_MIN or SERVO_n_MAX if the servo is reversed).
    For my convergence I set Q_TILT_YAW_ANGLE=10 as the motors tilt back about 10 degrees when fully up. This parameter allows the code to work out what PWM value to use for the motors to be straight up, and also to work out what range of movement to use in controlling yaw when hovering.
    The only other unique setup for this type of vehicle is you need to tell ArduPilot which servo output controls tilt of the left motor, and which controls tilt of the right motor. The SERVOn_FUNCTION for left motor tilt is 75 and for right motor tilt it is 76.
    So I have:

  • SERVO3_FUNCTION 75

  • SERVO3_MAX 2050

  • SERVO3_MIN 1000

  • SERVO3_REVERSED 0

and

  • SERVO4_FUNCTION 76
  • SERVO4_MAX 2000
  • SERVO4_MIN 950
  • SERVO4_REVERSED 1

that sets up the aircraft so that servo output 3 is the left motor tilt, and servo output 4 is the right motor tilt. The MIN/MAX values and have been adjusted to point the motors straight forward in manual, and so they are centred on being straight up in QSTABILIZE mode. Notice that for the convergence the right motor tilt is reversed (which is why SERVO4_REVERSED is 1).


(Rolf) #286

@tridge

Marvellous ! SERVO3_TRIM and SERVO4_TRIM value 1500 may remain ?


(tridge) #287

yes, when used for tilt of this type of vehicle the motor tilts are treated as range values, not angle values, so the trim is unused.


(tridge) #288

Another note on the EFlite Convergence. The wires going into the ESCs can be very confusing. It uses reasonably thick 3 wire cables, which correspond to power, ground and signal (signal as PWM). The colours are very confusing though, and there is a hidden connector inside the wing cavity which on my plane swapped colours. So white went in on the fuselage side, but it emerged as black at the ESC. It is really baffling when setting it up until you pull out the wire and see there is the hidden connector inside the wing cavity which swaps the colours.
Make sure you trace the connections carefully with a multimeter.
When we discovered this we wondered if the people in the factory were just trying to confuse DIY people :slight_smile:


(Rolf) #289

Hi Tridge,

Could vector control of the front motors be used not only for yaw but also for forward flight ? This could reduce battery consumption:

It costs a lot of power to fight nose down against the wind due to downforce on the wings. Here is an example, as at 15 kph headwind the climbing ability decreased to zero and current inreased significantly. I then let the VTOL drift backwards with the wind and was thus able to win a safe height for forwardtransition.

Especially at the end of the flight with an almost empty battery vector control could save time (and batterycapacity) while exact reaching the desired landingpoint in windy conditions. We ask from expirience at the end of today’s flight (Backtransition at 09:05)

Regards Rolf


(tridge) #290

yes, that’s a good idea. It could be linked to Q_VFWD_GAIN. The question is really what angle limit to apply to retain stability. Maybe just Q_TILT_MAX ?
I can try it on the convergence and see how it goes.
Thanks for the suggestion!


(tridge) #291

It turned out to be one line to implement it:


I’ll try it on the convergence on the weekend. Works fine in the simulator.


(Iam) #292

What modes will this work in? In Quadplane we only get to enjoy the benefits of Q_VFWD_GAIN while in QLOITER mode. I have a Github request for that to be expanded to QHOVER.
With this tilt rotor, will the motors tilt forward in QHOVER, or is it limited to QLOITER also?


(Iam) #293

I’m not clear what should be set so that the aircraft can move in reverse. So on your Convergence, does it hover level at 0 degree motor tilt and fly backward at -10 degrees?
Since you aren’t using negative values when the motors are tilted back, what value would you use if one a particular build the motors couldn’t tilt backwards at all, but rather were always 5 degrees forward?


(tridge) #294

It works in the same modes that Q_VFWD_GAIN currently works in, which means it works in QLOITER, QRTL, QLAND, and AUTO, but not in QHOVER.
The algorithm used for Q_VFWD_GAIN is tied into having a velocity controller running. We could use a different algorithm and not use Q_VFWD_GAIN at all, but instead some other parameter for QHOVER. Exactly how do you think it should work? How much should it tilt the motors forward for a given stick input?


(tridge) #295

I haven’t tried flying in reverse. I try to avoid flying quadplanes backwards as the elevons end up reversed, so it can develop some bad roll/pitch instability as the airflow over them reverses.
If you did have both motors back 10 degrees then the plane would yaw quite a lot. That doesn’t happen at the moment as it sets the motors like this:
https://github.com/ArduPilot/ardupilot/blob/master/ArduPlane/tiltrotor.cpp#L369
so it sets one to “straight_up + yaw_offset” and the other to “straight_up - yaw_offset”. The average is straight up.
Cheers, Tridge


(Iam) #296

Tell me if my logic is flawed. QHOVER is equivalent to Copter’s ALT HOLD mode, and in ALT HOLD mode forward velocity is just a byproduct of the pitch angle that is controlled the same as in STABILIZE. So the goal would be to replace Copter’s pitch control with motor tilt. The priorities are: 1) maintain altitude 2) maintain level (hover) pitch, 3) increase motor tilt proportionally to pitch stick until either 1) or 2) can no longer be achieved. This will define the max forward velocity in QHOVER.
For a conventional Quadplane, we use the horizontal motor throttle rather than motor tilt, so we are not limited because the motor isn’t also trying to handle altitude and attitude duties. In this case, I would just like to have the option to lock the pitch at level and have the horizontal motor throttle mapped to forward movement of the pitch stick.

I feel that QSTABILIZE should be the only mode where a winged aircraft is using down pitch to accelerate forward.
Thanks for hearing me out.


(PG) #297

I’ve been working on a pretty gnarly scratchbuild I plan on using. Still not ready for testing yet, but I’m very close to completing the hardware side of things, there’s a lot of 3d printing which slows the process. Hopefully its ready around the same time you confirm v1 for this differential thrust vectoring to control Yaw in hover is working smoothly. I’m hoping for a complete transition from Tri-copter to Fixed wing twin plane with functioning ailerons and possibly a functioning V-tail. this would entail full 90° rotation of the front rotors and locking into that position, and possibly having the tail rotor stop completely. Any suggestions on what setup would work best with the software side of things?- V-tail or flying wing.?
Thanks


(tridge) #298

I think this does make sense as an option, but I’m not yet convinced it should be the default. I’d like to try it out on my convergence and see what it’s like.
Cheers, Tridge


(tridge) #299

both should work with the current code. We’re still really learning what the best configuration is. The setup @Rolf has built works well, but I think there are a lot of other options to explore. I don’t think we’ve used it enough yet to be sure of the best approach, so use your own intuition and see how it goes!


(Walter) #300

@Pete … V-tail or flying wing.?
@trige … but I think there are a lot of other options to explore. I don’t think we’ve used it enough yet to be sure of the best approach, so use your own intuition and see how it goes!

The first question that @Rolf and I asked ourselves was "what do we want?"
The following specification for Mozart has emerged:

  1. a very good flying airplane with
    • Clean aerodynamics
    • Noncritical profile
    • Good slow-down characteristics, conventional landing possible in an emergency.
    • Even high speed can be reached (return “gliding” with a few power)
  2. Long flight time (use as UAV possible)
  3. Tricopter (Only one engine / ESC / propeller as a useless ballast in plane flight)
  4. Sufficient space for the equipment and LiPo (up to 10.000 mAh)
  5. Low noise
  6. visually appealing
    This has resulted in:
    A) The profile ClarkY (for everything to fly well :slight_smile:
    B) The double fuselage was chosen because of the higher stability of the Depron construction.
    C) The conventional double rudder, because easy to build. The choice was also a V-tail as with this: (One servo less)

    As Tridge says, there are many ways. This is an example of a lot of possibilities.
    The fundamental discrepancy of the suitability of the motors / propellers for hovering and forward flight
    would only be possible by using adjustable propellers. The optimum pitch is about 1-2" different.

Pete, good luck with your own specification.

Walter


(Iam) #301

Thank you Tridge. I hope it proves successful on the Convergence. I suspect it will be very smooth and intuitive. I do think it would be just as useful on a conventional Quadplane. I’d be happy to test it for you if don’t have something smaller than your Porter.


(Jacobo) #302

Hi, yesterday I test flew a tiltrotor I built for a university group project. Right after takeoff the tilt mechanism failed which caused the motors to rotate freely, going past 90 degrees and crashing into the ground. We used tape to fix the motor frame in place to show the covering capabilities, however as soon as the throttle was raised in QStabilize mode the controls of the transmitter had no effect on the UAV. The UAV shot up into the air, with no response to pitch, roll or yaw inputs and reducing the throttle from the transmitter had no effect on the vertical velocity; It kept going up.
The UAV was drifting away from the flight area and there was a panic to bring it down, so I tried changing through the flight modes and plane manual cut the throttle and it fell to the ground.

I’ve tried to load the log from this flight in mission planner but it is unable to display any data.
I assume the first crash must have damaged the Pixhawk but maybe someone can load the log and find out more?
First flight log:
https://drive.google.com/open?id=0Bw9jwkPqSLnsS1c5RnpMMV8yYms

Second flight log (no control):
https://drive.google.com/open?id=0Bw9jwkPqSLnsejRMZ2FTUi00ZFE

Thanks,
Jacob