Could you tell me the assignment Servox to the Mots (left,right,top,bottom) it jumps up and tilts in QStablize or QHover uncontrolled.To short to learn in the Log. Can’t believe thats only the CG.
I found another “new” Param value in the Param Diff: AHRS_EKF_TYPE 10.0000
@losawing Thanks, I meant frame_type. I corrected it above.
@lorbass I’ll export and upload my version of the model to the git repo. Radio channels 7 and 8 had no inputs assigned, and I set them to inputs 7 and 8.
Servos 1,2 are left,right elevon
Servos 5,6,7,8 are right, left, bottom, top motors
From caipiQ_diff.parm, the servo functions are (1st column)
SERVO1_FUNCTION 77.0000 33.0000
SERVO2_FUNCTION 78.0000 34.0000
SERVO3_FUNCTION 0.0000 35.0000
SERVO4_FUNCTION 0.0000 36.0000
SERVO5_FUNCTION 34.0000 77.0000
SERVO6_FUNCTION 33.0000 78.0000
SERVO7_FUNCTION 35.0000 0.0000
SERVO8_FUNCTION 36.0000 0.0000
AHRS_EKF_TYPE 10.0000 is for SITL, it uses the RealFlight attitude estimate and has the best stability. You must have low-latency communication between ArduPilot and RF8 and high frame rates to use types 2 or 3.
Thanks Mark,
I see a possible reason of my crashes. In the Wiki with the Diagram above (Post 247) Motor #1 is designed left, but seen from top is is on the right wing. And in the Full Parameter List Servox_Function #33 correspond to Mot #1. So I did it, but in your setup #33 is left. Because your definition flies, the Wiki image must be wrong. ?
Will check it with your RF8 Model.
Servo functions 5,6 are ThrottleRightServo, ThrottleLeftServo
and they are assigned as 34,33 which is right=motor2, left=motor1
I thought this was correct in the wiki, but apparently not.
It looks like bottom/top = motor3,4 is correct on the diagram.
Edit: IIRC the wiki diagram is correct, and verified using actual copter tailsitters by @losawing @hwurzburg and myself. There seems to be something swapped in the RealFlight models instead. To verify the problem is on the RealFlight side, I tried setting up your new caipirinhaTailsitter as a quad plus frame in ArduCopter, and it also needed motors 1 and 2 swapped to be stable.
I swaped Motor 1/2 and it is stable in QSTabilize and QHover. The lift off is critical and it needs a Jumpstart to get control.
However I had to set Elevon left to Reverse and Elevon right not. They move so correctly checked in Mode Manual. Thats oposite to your Params. I use Q_TAILSIT_INPUT = 0.
But perhaps due to this, in FBWA it starts to rotate in the Roll Axis and crashes.
Did you test FBW?
In order to test this, I set one motor(Servo) after other in RF8/Edit/Electronic to the knobCH6 of the RF8 Controller. So it’s possible to observe which Prop is rotating slowly and on which direction. Then this Servo (e.b. ThrottleRightServo) has to set again to the corresponding Receiver Channel.
And I did not find wheres the swap.
I did not found it so far.
Sorry, I forgot to push my changes. My version and parameters are there now:
It flies OK in FBWA, and elevons look right in manual mode.
qhover demo:
Hi Mark, Thanks for your assistance.
I’m verry interested to find the issue in my system.
In the meantime I can liftoff verry smooth like a llarge rocket. The reason for the need of a jumpstart was a bad AHRS_TRIM_Y. The horizon in the MP Hud was alsways too low. Accel Calibration didn’t help.
Mots left/right are swaped, but also Elevons Reverse.
I had to increase Q_A_RAT_RLL_P and D. Now Roll is OK but with about 45 Pitch it starts to Roll or Yaw.
But at first I will download and share your success. 
Hi Mark, good news in the new year.
Your Model and Params works perfectly.
I only had to increase Q_A_RAT_RLL_P and D, as well as PTCH2SRV_P. (Framerate 350)
It is even possible to fly aggressively (for my knowledge) in FBWA with perfect Transitions.
In FBWA exactly as I did it with Caipirinha as vectored Tailsitter.
Unfortunately I can’t make a video, with my Video App Wonderhare Filmore because the Frame Rate drops to 60 FPS and then the attitude is bad.
But, I’m not satisfied, I want to know why my setupt works so badly. 
Therefore I will compare character by character of both Setups to find the reason or your magic. 
Anyway, thank you it makes fun and motivate to order the material for the real Wing.
Just a question: How is your setup in RF8 to fly behind the Wing?
Congratulations for a great start to the new year!
I hope it’s not too difficult to find the differences between your setup and mine.
The RealFlight “chase” view is available in any 3D (non “PhotoField”) airport environment:
In a 3D environment, you can also choose Nose camera and add additional cameras to any model like the ones I added to the Convergence to show motor tilt.
Have you tried the built-in Windows 10 screen recorder? That’s what I’m using.
It fly also with a low Frame Rate while recording a Video from the Screen.
The target is to know where I made the wrong consideration to learn for future Models.
For me its like a orientation Run (Sport) in my Scool Time. My group where the first, but crossed the finish line from the wrong direction. 
One point is I used other ServoOut. This to be compatible with my FC Pixracer with 6 Channels on my Caipirinha Wing. But I adapted also the Servo Functions to this Channels.
I will report when I found the fault.
Thanks for the explication, I searched in the wrong Airfield.
RF8 runs actually on two old Win7 PC’s. Will install it later in a faster Win10 PC and test.
Are your Framerates better than in my Vid above?
The last capture I posted was 1104x768 (not full screen) and the physics frame rate varied between 110 and 130 fps with a 3D airfield and “chase” camera. The frame rates would probably be higher if using a photofield and a fixed camera. But the builit-in (xbox app graphics window only) capture seems to have no impact at all on frame rate.
Without video capture my Framerate is at 340, with video capture around 60.
In the meantime I compared both RF8 Versions at first for the Motors only.
To simplify the docu, I showed the Setup for Motor 1.
In a short sentence:
One version flies but do not correspond to the Wiki, the other version correspond but do not fly. 
To justify my statement I tried to document my findings as good as I can.
Please have an eye on it. The last pic was to proof which Mots turns with which Channel without SITL
@lorbass Thanks for documenting that so carefully, it agrees with what I observed here:
and points to a possible bug in the flightaxis SITL code. I’m looking at that now.
So you suspect the swap in Flitaxis. This means, in a real FC it works correct. Would be interesting to know, which SERVOx_FUNCTION (33 or 34) @losawing has used for the right motor in his Hacker Wing.
(I avoid mentioning the numbers in order not to start discussions where is #1)
See this message and the reply:
I believe @hwurzburg has also verified that it is correct for a real FC.
@lorbass, Motors layout in the tailsitter wiki are exact. They are in accordance with copter motor layout assuming the arrow point the plane belly (so tailsitter drawing is 180° rotated along the fuselage axis when compared to copter). Nevertheless motor rotation are not the same. From the beginning I considered copter drawings and when I want to add torque yaw control I also consider motors rotation according to copter.
My jet wing (quad plus) is still ready to fly and I will check motor layout this evening.
Thanks for checking. In the meantime I got “my” Version of Caipirinha Tailsitter flying with the Setting 1-4 for Motors and 5,6 for Elevons with the corresponding Servo Functions like Mark.
Mot 1, right = 34
Mot 2, left = 33
Mot 3, bottom = 35
Mot 4, top = 36
Elevon left 77 Reversed
Elevon right 78 non Reversed.
Does your RF8/9 work now?
This is the configuration of my jet wing quad +
servo2 = 77
servo3 = 78
servo5 = 34 connected to motor left wing
servo6 = 33 connected to motor right wing
servo7 = 35 connected to motor bottom
servo8 = 36 connected to motor top
Concerning RF9 I have a problem with the identification number, I am complaining to the seller. The SCP driver indicated by Mark seems to work as my dualshok is connected by USB and found by windows.
This means, the Wiki is correct, thanks.
The connection of my Taranis X-Lite to the PC as Joystick works also. But in RF8/Controller/Edit, Channel 1 to 4 moves only. The controller has also an important red button to reset or restart the connection to Flitaxis. Therefore I repaired the broken gimbal part of the controller even verry difficult. Would be interesting to know how it works wit your system. If yes it’s the cheaper solution.
