My problem of the right motor that do not turn as the left one seems tobe solved, it depends upond the attitude of the Pixhawk, when it is levelled the motors start together.
Qhover is a big problem , motor output is completely wrong, from 1000 to 1500ms the motors do not accellerate then they suddenly go to 2000ms !
Take look at the image and compare Qhover with Qstabilize
Hi Lucamax,
better use GPS ! ALTITUDE (rather attitude) estimation wonāt work well without GPS.
How to configure a flying wing without ELEVON_OUTPUT?
Regards
Rolf
In QHOVER the Throttle Stick controls the desired altitude.
It seems. you increase the desired altitude and then suddenly the
wing want to reach this by full power.
My GPS has no problem to find enoughe Sats (15) even when vertically and in the feld.
In my last Video you can see in the middle the canceled starts with heavy Pitch vibrations.
And then how I did correct it. Mainly because the Params P or D are to high.
Currently I fight against the Pitch Vibrations during forward flight as you can see in the pics.
I make serveral test flight with different Params. Afterwards in the Log I can see which one is better.
In this time flight no 5. Pitch P is now at 0.05 and probably always to high. Who nows?
I do not yet start in QHOVER.
Some videos of the Stryker non-vectored tailsitter that I may not have posted here before:
This one shows improvements in hover agility and control after changing to digital elevon servos:
āMark
Thanks for your replies guys.
Here is my wing
I have install a Gps .
I tried to engage Qhover after I was holding the plane in Qstabilize at 1 meter from ground but as soon I did it throttle when to maximum.
I have different problems in Qstabilize:
The yaw command that move elevons in opposite direction is active only after take off.
The problem is that the elevons are not neutral when I switch to Qstabilize mode, they are almost neutral if I switch to Qhover.
So I have to take off holding the plane by hand ad adjust yaw until the plane do not rotate anymore.
The plane is not difficult to control but after a short time it just start a kind of āauto transitionā and start to fly as a plane in level mode, obviously I totally lost control and went in a tree.
Some notes:
I did a short video , sorry for bad quality , at the end you can see (hardly) the plane transition , please note the increase in RPM while I did not move throttle stick.
I changed the motors with more powerfull ones, now hovering is about 60% .
Still have the left motor that moves a little , 1 rpm , while non armed or armed.
I found out the reason for āautoā transition.
The Pixhawk corrects in the wrong direction pitch control.
I hang up with a thin Nylon rope the plane and did a test while in Qstabilize.
When the pitch angle become important moving the plane body by hand then the correction goes in the wrong direction and the motors accellerates trying to correct more and more but in the wrong way, as you can heard in the video.
Q_FRAME_TYPE = 1 , does that matters ?
This parameter is used for Multikopter and means X configuration.
But how about other params?
Q_ENABLE = 1 (else you canāt see other params for the kopter mode in the config page in MP)
AHRS_EKF_TYPE=3
EK2_ENABLE=0
EK3_ENABLE=1
Q_FRAME_CLASS=10 (means Tailsitter)
Q_ASSIST_SPEED=0
KFF_RDDRMIX=0
MIXING_GAIN =0.5 (0 without elevon activities which makes landig difficult, >0 mixed)
Q_TAILSITT_VHGAIN 0.4
and of cours the params for the tuning (PID) according to the wing
Here you can read all about params
http://ardupilot.org/plane/docs/parameters.html?highlight=complete%20parameter%20list#q-frame-class-frame-class
And especially for Tailsitters
http://ardupilot.org/plane/docs/guide-tailsitter.html
thanks lorbass,
I have update my parameters with the one you suggested.
Unfortunately I still have the same problem .
When in Qstabilize after some seconds of hover I loose control of pitch and the plane attempt to fly levelled after increasing motor speed.
I suspect because the pitch authority is poor , pitch correction via elevons is not enough .
Here is the log
https://www.dropbox.com/s/nwlzt42d7hshwxf/00000058.BIN?dl=0
I must take off holding the plane by hand because I need to adjust the yaw since it is not neutral .
How can I trim Yaw ?
Is it normal that when in Qstabilize , yaw command move the elevons only after take off ?
@tridge Re: Dual-motor tailsitters has parameter Q_A_ANGLE_BOOST been removed? I canāt find it in master.
@lucamax Donāt worry about yaw control response before takeoff; it will do the right thing once you get off the ground, even if it looks wrong to start with. The pitch problem is most likely to be due to lack of elevon authority; see the posts above by @lorbass and his decision to go with vectored thrust.
Did you also raise the output rate for the elevon servos? In the 3.8.0 release you can set that with the SERVO_RATE parameter. Paul found it made a significant difference for his vectored tailsitter.
@tridge No, that Stryker flight was on 14 May, and the log shows no SERVO_RATE parameter. Iāll plan to use a rate of 70Hz for the Convergence, since that is what the stock controller was using.
Need some help here.
I really do not understand, each flight the same, the plane after a while loose the pitch attitude , motors increase rpm and the plane tends to fly horizontally with a kind of auto transition and finally a crash.
Here is a short flight with the event
https://www.dropbox.com/s/n4lx0xwktw2e07i/00000063.BIN?dl=0
Here my params
pitch problem1.param (16.1 KB)
After these changes in parameters and some refinements in plane elevons trims I was able to fly nicely without wind in Qstabilize , Qhover and did a perfect transition to FBWA.
Here the changes:
Q_A_ANGLE_BOOST - from 1 to 0
Q_TAILSIT_ANGLE - from 30 to 45
Q_ANGLE_MAX - from 3000 to 2000
The purpose of these changes was
The plane flies well in FBWA mode without any Pid tuning even if there was a strong wind but the transition to FBWA back to Qhover failed and the plane in Qhover was out of control .
I do not know if the above changed parameters have a negative inpact on FBWA to Qhover transition.
With strong wind it appears to me that it is very difficult to fly a non vectored dual motor tailsitter .
My plane tends to tilt on pitch axis when in Qstabilize and again go in auto transition .
If the wind hit the plane perpendicularly while in Qstabilize the generated force is by far much higher than the one generated by props airflow on elevons trying to maintain pitch attitude so I guess only vectored motors can fight with that force.
Congratulation, I was not succesful so far.
You have more courage than I.
Iāts a shame that āprofessionalā do not want to help (or are not able) as we can read back.
A development is succesful only when confirmed in practice.
And practical tests also need time AND money.
Thats the reason why I stopped posting issues.
I wish you a lot of fun
Otto
That was the reason why I changed from Tailsitter to vectored Tailsitter.
Or look what large Elevons they have here:
great, congratulations!
the small Q_ANGLE_MAX may mean the plane is unable to hold position well in wind. How much wind it is able to handle will be very airframe dependent.
yes, that is indeed a problem. Even with vectoring it is challenging with our current control scheme. Iām hopeful that the new TVBS control system that Paul Riseborough is working on will handle wind a lot better.
You are right lorbass but this brings an interesting question , since the Wingtra developer has an important knowledge on dual-motor tailsitter design , why didnāt he choose the vectored solution ?
Your plane seems to me to fly pretty well so you should definitely go for the transition .
Thanks for the explanation, I will increase again that parameter and see there will be an improvement with wind.
About the failure in transition from FBWA to Qhover , I suspect that my throttle level was too low when I did the switch as you can see in this short video.
I flew behind a line of trees so until 7-8 meters the wind was moderate while higher , where I did the FBWA to Qhover transition , the wind was much stronger.
there is a trade-off between simplicity and control. The vectoring adds complexity, but also adds a lot of control, and allows for smaller elevons. Plus it solves the āfalls over on landingā problem 
Hi, I donāt think they thought of it , neither did NASAā¦ see 1st comment on my video
Iām not āfanā of Wingtra System, I mentionned it only to show how they use verry large elevons.
Therefore I changed to vectored. The landing is now solved, even hover in āNWCā (no wind conditions)
Also climbing, sinking and hovering sidewards is not a problem.
But moving forward/backward produce a slow āpumpingā
I know, itās not made to move in hover, but itās needed to stay on place against wind for landing or last approache for it.
I agree, its a complex control task. When moving the tilt this produce a counter tilt to the wing even with stopped motors. The elevons compensate a little bit. This effect is more effectiv with a small wing with the weight (Akku) in the center and with rotating props (gyro-effect). And this counter tilt of the wing ist just the opposite direction of the correction by the FC.
And finally with increasing speed the frame ist tiltet more than the FC want to do it and so on.
Tuning is difficult, because in order to eliminate shaking of the Frame, even with stopped motors, the PIDās have to be reduced.
And this setup is not the right tuning for the flight.
This is the finding of 30 flights with different PIDās and verry constant āpumpingā while 1h 45 min in the air with
absolutly no wind.
Kind regards
Otto
