Servers by jDrones

Dual-motor tailsitters

(Pavel) #1193


I’m testing my tili motor tailsitter model now.
Have successfully tested it against QStabilize, QHover, QLoiter and QLand modes.
However, every time I try to make a transition to a horizontal flight, in my case FBWA, or perform the mission like VTOL_TAKEOFF the plane almost immediately loses control.
Just few times I was able to save it from crashing by sending QLand high enough above the ground.

last crash: after 1:20

logs are here: (sorry .bin file was lost)

My configuration:

  • ArkBird VTOL mo from banggood
  • 3S 2500mAh battery
  • Turnigy PARK 480 1320kv motors
  • Pixhawk 2 flight controller
  • SW: ArduPlane V3.8.5. (7dc5e040)

Pls. help me to analyze this behavior.

(letpi) #1194

It seems to me the wing want to fly trailing edge first.
Where is the CG ?

(Pavel) #1195


the center was about 1-2cm to the front from manufacturer recommendation.

(lorbass) #1196

It is verry unstable in QStbilize oder QHover.
I needed more than 100 test flights to finde the correct parameters
Your wing is similar to mine.
So compare my parameters with yours to finde whats wrong.
Vor wechsel auf 3 8 3 beta3.param (16.1 KB)
And this was the flight with this params.


(letpi) #1197

So OK for the CG. I suggest to switch to manual and make a hand launch to test.
Because your wing fly in VTOL mode I dont think something is reversed. But just in case, hold the aircraft in your hands and switch to FBWA mode, then pitch and tilt the plane to confirm that the control surfaces move the correct way to return it to level flight. Sorry for this suggestion if you already made this test.
I checked your fixed wing PID. They seems to be almost default but strong oscillations we can see on your log and on the video may suggest they are too high. You can test rll2srv_p and ptch2srv_p= 0.3 instead of 0.6.
You have set Q_tailsit_vf = 0.4. This is a high value and I suggest to set it to 0. Vectored thrust is a nice feature but it is not needed for horizontal flight, you will raise it again in a second step.

(Mark Whitehorn) #1198

Yaw control is better now for the little “Dart” quad-tailsitter. Larger elevons with more throw and disabled the code which abandoned yaw control when errors are large. Note that there were no large yaw errors induced during this test though.

(letpi) #1199

I like the idea of this small and fast quad-dart. I hope to see transition soon and speed tests.

(Claus) #1200

Hi there I am new in this forum and I have bulit a Tailsitter Caipirinha2 with non vectored dual motor setup.
I am playing around with arduplane,copter,and traditional heli. the last 6 years.
I read nearly all of this discussion here an tested a lot (crashed about 12 times) -
May elevons are about 1/3 of the wing as recommended. Now I exteded the servo throw.
The Plane flies acceptable in the ‘wing’ modus FBWA - but in qstabelize and qhoverI have the following problem:
The hovering is quite stable until a light wind blows or I lean more than 30 degrees to the back or the front.
If this starts …the motors are accelerating- and the plane starts to climb and wants to fly !!! I once hat that issue - switched to fbwa and then flew a round …but It should hover until I flick the switch to fbwa not before.
Now I examined the logfile and the throw of the elevons when starting to lean in either direction are max 45 degrees - although the physically move could be 60 degrees (for example in acro or man mode).
My question now is: which parameter determins the max elevon lean angle ? I went through the Q_ and Q-A parameters many times but could not find a solution.
Maybe one could help me :grinning:

to My person : I am from Germany, just turned 50, and build about 8 drones, 2 helicopters, 2 planes and now the tailsitter :wink: came from the very beginning of apm 2.5 now to pixhawk mini micro and pixracer(like the wlan configuration via ipad )

Thanx for your help in advance ,


(Pavel) #1201


thank you for your fast answer.
After the test on last Friday we have to change the motors, props and the battery type, thus to correct the CG the battery was moved back.

About surface test:
We did that many times, it looks like on this video. However, the surface move is opposite to what is mentioned here in Servo Setup & Reversal section

Yesterday morning we have successful loiter test:
loiter test

Same afternoon we tried Auto mode mission and crashed. The mission was very simple:
VTOL_TAKEOFF (height 30m) – Waipoint (about 100m away) – VTOL_LAND
Auto mode test

(lorbass) #1202

The recommendation is: Change to vectored as I did it with the same frame.
Already after the third crash.

And it works perfect:

The age is not a negativ criteria, I’m 77.

(Claus) #1203

Thanks for your answer ! From reading this big document I already felt that non vectored tailsitter are not so windresistant that a vectored one. It it secessary to have a fully 180 degree angle or may it be enough to have 120° ? I am not sure I could realize a 180° total tilt angle…
Greez ;

(lorbass) #1204

If you look some postes above, JUSTASON seems to work with less.

I use this servos:

If you buy the programmer they can be changed to 180°
The same supplier can also it program for you with 180° angle if asked.

(letpi) #1205

There are 2 things
The first one is to get the control surface moving in the right direction when you move tour TX sticks. I can see in your video this is not the case for elevator.
The second is to verify in FBWA that the control surfaces move the correct way to return to level flight For this control, hold the aircraft in your hands, pitch and roll the plane.
If your plane is able to hover I think the second item is OK. So you may just reverse elevator with your TX and make both test again.
I think the first video crash is an other problem.

(letpi) #1206

±60° is plenty
+90 allow to land on the belly

(Claus) #1207

Many thanks for your input ! before I convert my VTOL to a vectored one I will give it a last try …Therefore I would like to extend the ailerons throw …The q_Pitch_P value is not the right parameter …Does anyone have an idea which parameter to increase to get the foll throw of the servo ? I saw on the log that ist barely hits the ±40 degree value - although 60 are available …

(letpi) #1208

q_a_rat_ptch_p is the parameter which converts pitch speed error to servo throw.
I think this is the parameter you are looking for.
There is also mixing_gain that you can raise up to 1.5 but the elevon mixer may saturate.
mixing_offset is a parameter to configure the aileron response vs the elevator response. If I understand correctly, a negative value will give more throw to elevator at the expenses of aileron. I never tried it.

(Mark Whitehorn) #1209

Hi Claus, have you checked whether parameters SERVOn_MIN/MAX are limiting your elevon throw?

(letpi) #1210

I have read your message until the end lately and have some questions / suggestions.
How are you able to fly your wing in q_loiter mode with the TX elevator stick reversed ? Could it be the normal way for you ?
Your wing seem to be very powerful and responsive but fly pretty well in q_loiter.
You are probably the first one in the world to try an auto take off with a belly sitter. I dont think there is a bug in the code but this is a possibility. But, from what I see with your auto take off attempt, there is not enough dampening, I would try to increase q_a_rat_ptch_d, q_a_rat_rll_d and q_a_rat_yaw_d to 0.01 or 0.02. Another possibility is the integrator is saturated and cause the overshoot, try to halve q_a_rat_ptch_I and q_a_rat_yaw_I.

(Claus) #1211

Hi Marc, yes the limits are set and work when I am in manual mode :+1::sunglasses:

(Claus) #1212

Hi Losawing, your suggestion makes sense, will try it out this afternoon, if the weather stays dry ! thanks for your help!