Really nice result, great tuning!
Thank you @marco3dr, great to hear! Tuning worth it and it is fun to fly.
Was able to build 10 in this short period. 3D printing helped to accelerate effort a lot.
Trying high lean pitch angle in hover. Set Q_ANGLE_MAX=70, Q_TAILSIT_INPUT=3 (roll by bodyframe). Unfortunately elevon oscillates at higher lean angle, needed to limit pitch angle. Thought it may be elevon movement saturation but symptom (PIQP.Limit=1) does not match oscillation in the video. Think need further tuning tweak. Wish I can fly this beautifully like losawing’s, but miles away…
Maneuvers are mostly rudder turn, even applying reverse aileron.
It is 4.2.0, log is here.
https://youtu.be/VrInmB7KWAA
To stop oscillations at high lean angle is just a question of tuning gain attenuation. The disk theory gain attenuation gives beautiful resut You may have to reduce (a lot) the disk load parameter and/or the min scale parameter.
Yes, rudder stick is the main command. The roll stick has to be applied in reverse direction (roll stick right to face left like a 3D plane when prop hanged) at low lean angle and in the same direction than the rudder stick at high lean angle (when the tailsitter begin to fly like a plane), This is very logical but need to be used with. You can imagine that when flying inverted this is even more compicate. 
Hi @losawing, thank you so much for advice! I tried in SITL+RF and working great, can’t wait to try on real one.
Thank you also for controlling advice. Yes, inverted looks a bit confusing. I can do inverted rudder turn in 3D plane somehow, but elevator stick pressure is opposite in VTOL plane so need practice at altitude for sure.
Tried with lower Q_TAILSIT_DSKLD and improved. Lowered DSKLD to 5 down from 10, then 4. Unfortunately lost attitude and crashed bad, rebuilding another one.
Separately started to try Quick VTOL Tune LUA script on config #3 which has KST servo and H743 FC. Script worked nice on SITL+RF. Verified script started ok in the air but damaged the plane during setting adjustment. Will continue. It will be tremendous help in tuning.
Config #3 high lean pitch flight with lower Q_TAILSIT_DSKLD
Log is here
Attitude is lost during elevon saturated. Plane is yawing with C4 then C3 maxed, then left motor C1 kicked in, then C2. Too low to recover.
Not sure the exact cause, at least this plane’s envelop is still tight with smaller elevon. Also I think side slipping during turn can lead to this critical situation. Need to practice well, it is getting similar to 3D flying…
hi Satoru,
So much progresses since the beginning, the Eclipson VTOL is now flying pretty well.
About the crash, 1s before the plane hit the ground you had the belly of the plane in front of you and from the log you pushed yaw stick left (rcinC4 lower PWM if I am correct) so the plane rolled left as it should but relative to you it produced a right roll.
edit: when looking at stick position (overlay top left in the video) the left stick order only occur after the crash !! is it a overlay synchronization problem or my log analysis is wrong ?
About gain attenuation, maybe you will get less control surface oscillation by reducing q_tailsit_gsc_min, only tests will tell us.
To improve control, it would be interesting to test larger prop. This is something I never tested. Relative to the E-VTOL wing span, 10 inches propellers seems correct and you motors can spin them on 3s easily. From my side I dont like slow fly prop, I use this kind CW/CCW, they are more efficient, draw less amp and are very light.
They are not as good as Graupner genuine ones but good enough for tests.
Edit 2: @Satoru_Sasaki you do not use q_tailsit_rll_mx parameter so your tailsitter is allowed to roll up to angle_max and 70° is way too much. This is very likely why you crashed your plane. I suggest 40° for q_tailsit_rll_mx , 50° is an absolute max.
1 Like
Hi @losawing , thank you so much for looking into this and very precious advices!
Yes, thanks to your help and other veterans help, it is flying much better, getting more confident to fly under windy condition. I am already printing and building, will be flying soon. It will be 11th build and always new. 
Larger prop is interesting. Yes plane should have enough gap to fit 10 inch. I may have that in my stock, or have other 10 inches a lot for sure. Did not have chance to try non Slowfly prop, should try higher pitch prop as SF prop does not work well as a plane.
Thank you for looking into the incident as well. Looking at related in detail, Roll, Yaw axis was saturated (PIQR.Limit = 1). Yes, it does not look like RC input is causing this.
I will fix q_tailsit_rll_mx, and try similar maneuver at altitude.
Had a chance to run VTOL Quicktune lua script, successfully.
It finished tuning in less than 2 minutes with good result.
Plane is config #3, Matek H743 Wing FC, fast KST DS215MG V3.0 servo. I took so long time to tune this. This script saves 10s of tuning flights, so happy. This plane was already manually tuned to certain level, so tune from default may take longer time but impressed it dealt with this fast servo well.
Video shows log messages to see the progress. Tuning progress message with parameter value, slew rate is displayed. When I tried the script on SITL with modified CAT tailsitter (I made it heavier), it lost attitude at yaw tuning, but this flight was eventless, no visible attitude change.
Log of this flight
1 Like
Hey! I’m just coming back to all this after some major life events happened this past year. How’s the plane flying? Any major updates in the last few months?
Hi good to see you here again! The E VTOL is flying great.
Have not posted this here yet. This build #11 with 2M PixRacer FC. It is stable after limiting q_tailsit_rll_mx to 40. This video contains 2 hops. 1st with maneuvers in QHOVER, then transitions in next hop.
Log of this flight is here
Fun to fly, but FC lacks memory to run LUA. Should focus on H7 FCs.
My mistake I printed this wing with gray color LW PLA, which collects sunlight heat. Need extreme caution in this season. (10 minutes direct sunlight, wing surface deforms…) I rebuilt 2M FC one as no dihedral layout and tuning as build #12. 10 inch prop next.
3 Likes
hi Satoru,
Now you have got a very stable flight, are you able to indentify why ?
Thank you!
Your advice helped a lot to limit q_tailsit_rll_mx to 40, it can fly at high lean angle at q_angle_max = 8000 without issue. Need caution to control not to sideslip.
I set same parameter to other config build (Matek H743 Wing, fast KST-DS215MG servo one) and it also flew stable as you see the video.
For smoother transition, I set lim_pitch_max to 60, pull up manually a lot before transition to hover. To transition to FBWA, lean pitch down a lot before kicking transition. This plane is stable at high angle of attack and it helps a lot.
Unfortunately elevon horn was broken at the end and ended up with a dramatic crash… Building another now. Q_A_RAT_PIT_P is still high and can see elevon vibration, needed to adjust after quick tune.
Here is log of this flight
2 Likes
Even if it crashed I would say that the tuning is very good, congratulations.
You could have definitely saved it by switching to QStabilize and cranking the throttle to maximum.
If it loses control in QHover with that max angle at 80 degrees it is normal that it may be in a difficult condition to recover being bound to a maintenance of height.
The derivative on the pitch is also too high, those vibrations on the elevons do not have to be there.
1 Like
Thank you so much for encouraging comment! It is getting better!
Re-built another one, now 13th build. After checked it flies ok, I changed with larger prop. Good old GWS SF type plastic 10x4.5 (I have lots in stock.) from 8 inch. Looks like E VTOL has room to mount even 11 inch.
I was worried about charasteristic change. Instantly noticed elevons are working better. It started oscilate badly and needed to lower Q_A_RAT_PIT_P. Still need to adjust other parameters. Hovering throttle position also changes until it is tamed by Q_M_HOVER_LEARN.
Quick video of 10 inch short flight while adjusting params. Almost no wind condition and fun to fly time. Log is here.
1 Like
10 inch prop config is good, tuned further and elevon vibration is less now.
Short video of flying around on an autumn sunny no wind day,
Log is here.
2 Likes
Hi all!
I am reviving this topic since quite recently I started using the eclipson VTOL plane. I have performed some flights and have tried to tune it in QHOVER and QSTABILIZE mode.
My setup is:
- Matek H743 Wing - V3.
- Feetech STS3032 servos.
- I use the LW-PLA version.
- 8x6 props.
- Flying indoors for now so no GPS, no magnetometer. Just IMU and barometer measurements.
Following the Wiki docs about tuning a tailsitter, I started by configuring the Notch filter. Then I calculated the FeedForward gains for the Pitch and Yaw rate controllers as well as the Integral term of the PIDs, using this formula:
- Q_A_RAT_PIT_FF = 0.5
- Q_A_RAT_PIT_I = 0.5
- Q_A_RAT_YAW_FF = 0.18
- Q_A_RAT_YAW_I = 0.18
Then I followed the steps for configuring the P,D terms of the PIDs and ended up with these gains:
In this flight log I perform sharp pitch movements and yaw as well with the latest tune configuration. The plane is quite steady I think, however there are some points that confuse me and I would really appreciate your input.
- First of all I d like to make sure what signals should I check when tuning the rate controllers. I compared the RATE.PDes (desired pitch rate) and RATE.P (actual pitch rate); similarly for yaw.
- Given the above, I noticed a rather strange situation. Decreasing the P term, led to more overshoot. This is rather not logical. How is this explained? Could it be that the FF term is very large?
- Q_A_RAT_PIT_D term was tuned to 0.1, according to the docs procedure. But I noticed that zeroing it did not change much. Should I choose 0 then?
- Changing Q_A_RAT_YAW_D from 0 to max (0.02) did not have any noticeable effect on the output. Again, should I choose the smallest, ie 0?
- Apart from the above it would really help if anyone had anything else to recommend in general for the tuning according to the flight log I provided above.
Hi @ntouev , I believe you requested modified files at YourTube. I placed them temprarily on GoogleDrive. (You need to request access when clicking this URL). This zip includes modified wing1 (in 3mf, I set 2 layer wall for strength to prevent vibration), and modified fus1,2 to move battery to the bottom. https://drive.google.com/file/d/1bsdGEmFuJb4xMVgDHQ7KttVZOPlhmdYz/view?usp=sharing
1 Like
Hi Satoru, yes that was me. This is great! Thanks a lot! If I understand correctly this is the left wing. I assume you apply some mirroring for the right, correct?
Great! Yes, please mirror it to print the other side.