Hi and thank you all for sharing your knowledge! Im wondering if there is anyone out there that can help me with my problem. I have been working on a Y6 tiltrotor and i need a little help with the transitional tuning. It uses the firefly6 vtol parameters. Its sort of like a tricopter collided with something in the back of Burt Rutan’s barn. The pixhawk onboard is having a hard time with logs… most of the logs come off corrupt and half the time the sd card does seat properly and it wont arm. i had to bypass logging to get it to arm regularly. I mounted a gopro on the tail so its a pretty good view of whats going on.
The first part of the video is to show how it tilts on the ground when switching from qstabilize to fbwa and then qstabilize to manual. I show both because the speed is different. Then the first transition is almost perfect. all of the next transitions struggle. The very last transition in the video i went from q loiter to fbwa for a few seconds then to manual to get it to switch quickly. the machine weighs about 13 lbs and has plenty of power in all modes. It is a little nose heavy which i intend to fix once dialed in. the motors are lower quality 5208 running on 6s with 18’s on the top and 20’s on the bottom. no air speed sensor, no osd, no ground station or telemetry of any kind. It transitions from plane to hover perfectly every time. from hover to plane is very inconsistent.
The problem that i have is the tilt angle and duration seems to be a little bit different every time. I assume that this is because of airspeed. Sometimes it barely tilts. If anyone has any suggeations let me know. Im not sure if possible but i would like to completely eliminate the need to use airspeed (measured or estimated) during the transition. If i could make it tilt to about 45 degrees and stay there for a certain number of seconds while maintaining multi rotor control, then quickly rotate the other 45 and disable multi rotor control at some adjustable point. I think if i could do that i could dial in a transition that barely looses any altitude during most weather conditions.
Let me know if you have any questions or suggestions. Thanks!!
Hi, and as its your first post, welcome to the forum!
First things first you will need to sort out your logging issue. We can’t really provide support/advice without looking at logs. Sadly video on its own doesn’t tell us enough. What sd card make/model are you using? You mention that the sd card doesn’t seat properly. Why not? What is causing this?
You mention that the aircraft is a little nose heavy. What measure are you using to determine this? What I mean is where is your CG and where should it be?
Have you tried flying it in forward flight in FBWA and auto-tuning for fixed wing flight first?
Hi Matt, Thanks for your reply! Hope you are staying sane, safe and busy during this pandemic. I understand about the logging. I have a new flight controller on order and i expect to get it next week. I have tried the card that it came with and a couple of others kicking around. It just doesn’t click in nicely like it should. I understand how important the logging is when troubleshooting and i plan to get it fixed asap. I know ill need it when fine tuning
I have built a few canard airplanes over the years and i usually start with the CG ahead of where it should be. The calculator puts it about an inch ahead of the wing and i have it set about 1-2/3 inches ahead. It just feels a little nose heavy when flying but it could be not enough throw on the control surfaces. The elevators are small and I have the ailerons set to eleavons for assistance but the barely move. I have had many many failed transitions before i mounted the gopro and they always shoot the nose up and flips over so i thought a little heavy nose is ok for now to assist with the recoveries. (the software recovers amazingly by the way!!!) I just started finally getting a few successful transitions. Even had one bad crash that hit the ground and required quite a bit of rebuilding.
Im wondering if doing away with airspeed altogether for this application is possible? not sure how difficult it would be to do but i would really like to try it. I feel like with my big slow props and low stall speed i could get a reliable time based transition without the need for tilt compensation based on airspeed calculations.
If the transition could be strictly time based i think it would open some doors for other testers out there as well. Maybe an option to turn off airspeed calculations during transitions. If i could tell the machine to tilt to a certain angle (maybe 30-45) at so many degrees per second (maybe 15-20) regardless of what the airspeed is saying . And then it could stay at that angle for so many seconds (maybe 3) while maintaining multirortor control. and then complete the tilt at a different amount of degrees per second Maybe (50-100). all that while cutting multirotor support at certain tilt angle (maybe 60-80) If this is possible i would be very willing to pay money to give this a shot! I feel if i had those options i could dial in an almost always perfect transition.
So far the only other thing that i want to do as far as programming goes is try to get the front props reversed on one side to reduce torque. If i change Q_frame_type from 11 to 10 will this accomplish what im looking for while leaving everything else the same? The ff6 configuration spins all uppers one way and all lowers the other way. I think that the torque should be (somewhat) balanced due to the coaxial props but i eventually want to experiment with differential thrust for yaw control and i think it would just work better.
Thanks again and i look forward to continued testing!!!
With regards to the motor mixing you are looking for. I don’t fully understand what layout you are looking for. Are you looking for something like this?
With regards to the transitions. Can you please post your latest .param file.
Have you flown the vehicle in FBWA without doing transitions (i.e. hand launching or taking off from a temporary landing gear) and done a fixed wing autotune?
It is my opinion that the way in which the transition is currently handled should provide enough flexibility to achieve a good transition. The transition is predominantly time based already. The airspeed is only used as a check to begin the timer phase of the transition. i.e. once the vehicle achieves ARSPD_FBW_MIN the timer begins.
I am not aware of any airspeed based compensation on the motor tilt angle. Happy to be proven wrong though can you (or anybody else for that matter) link me to where in the code this compensation is done if I am incorrect.
In short, I think it is important to get logging sorted and the vehicle tuned for flight in FBWA first. Then I believe we can get your transitions tuned to essentially achieve what you have described (a time based transition whereby you provide a max tilt rate)
Yes something like that would be great for motor layout. if i change q_frame_type from 11 to 10 will i loose some other functionality pertaining to the ff6 style of vehicle? that was my concern. So if i make that change and change the props to match it should fly pretty much the same?
The latest param file is posted above in the top post above. before trying any transitions I did fly in manual and FBWA a few times with the tilt locked in place. Except for feeling a touch nose heavy (very minimal) It flies perfectly. It climbs great and cruises level at 60% ish throttle. I did not do fixed wing autotune.
Here is a video of a good test flight. https://www.youtube.com/watch?v=_RwcaEmqXyc
The second transition half way through the video is 100% perfect. I wish it would do that every time.
I have zero knowledge of how programming works other than changing parameters. I just assumed that airspeed is being looked at in live time and tilt adjustments were being made to overrule the timing. My biggest issue is that the tilt does not act as i thought the params should make it act but it could be that im doing something else wrong. the tilt seems to have a mind of its own sometimes.
I understand the logging and i agree and plan to get it fixed asap. the machine does fly great as an airplane and as a y6. maybe not the most energy efficient but very very stable. the recovery from a bad transition is outstanding.
If you look at the video above at between 1:40 and 1:41 it tilts to about 35 degrees. which setting decides that angle? Q_tilt_max seems to be something else? Also what parameter decides how long to stay at that angle before making the final quick tilt? Thanks again for your help Matt!
Yes, changing the frame type will just change the motor mixer in this case. Be aware that changing from Y6F to Y6B will also change your motor ordering. So as well as changing the props around you will also need to change the SERVO_FUNCTION parameters to be the new motor numbering. See here for the motor numbering.
You appreciate though that moving from Q_FRAME_TYPE = 11 to 10 won’t “get the front props reversed on one side to reduce torque” it will swap the motor spin directions on top and bottom. So all motors on the top will spin CW and all motors on bottom will spin CCW. This doesn’t sound like what you wanted. Am I miss-understanding something here?
What other functionality are you referring to here?
The good news is that as your aircraft is flying well in forward flight modes and VTOL modes, I am confident that getting a good transition is just a matter of tuning.
Hopefully I will get some time this evening to have a look over your .param file to see if I can spot anything obvious in there. Otherwise, it may just be a case of waiting until we get logs.
Hi Matt, Sorry for the delay work was really busy but off now for a while. Went to the post office yesterday and no new flight controller… I am anxious to get testing. Once the transitions are good and im confident there will be no more crash i want to finish and paint.
Regarding the motors now that i looked closer i see what you mean about the motor numbering. I did not notice that it all changes like that. The upper props are different than the lowers so if i have the props will try that unless there is a simple way to just reverse the rotation of the 2 front motors on one side by changing some other parameters? The other functionality i am referring to is maybe it would change not just motors setup but other things in the code that pertain specifically to the ff6 but but sounds like nothing else changes. I have no idea about how the coding works behind the scenes. I am going to leave that part alone until i sort out the transitions first.
Yes the aircraft flies really good and transitions are so close… Hey i am totally willing to hire you and pay you a to have a look at the parameter file and see what you can see? there might be something obvious to you. Please send me your paypal and ill send you some cash?. Can we bypass whatever wont let it tilt on the ground exactly like it would in the air? If it is mostly based on timing than it should be easy enough to set and test from the ground. The ice has melted at my place so now a quick test flight has turned into a big event not too many fields around and the aircraft does not fit in a regular car. Let me know. Thanks!
I have looked over your param file and noticed a couple of things. If you are keen to do some flight tests then please understand that I am ‘taking a shot in the dark’ here and cannot be confident in the numbers. But this is at least something you can try. From the param file it looks like your aircraft wings are likely still stalled when the transition timer starts. The transition timer is set to 0.1 sec so the tilt rotors tilt forward at your Q_TILT_RATE_DN speed of 30 deg/s for 0.1s, then just move straight to full forward whilst the aircraft is still stalled. This appears to agree with what I have seen in your videos as the rotors tilt forward rapidly and the nose drops. Therefore my advice is the following:
You currently have ARSPD_FBW_MIN = 5. This seems low to me as your aircraft looks like it has a rather high wing loading and therefore must have a reasonably high stall speed (just guessing here though). This value needs to be higher than the aircrafts stall speed. Maybe try this at somewhere between 10-12 m/s. This is the only airspeed check in the transition.
Up until the aircraft reaches that speed the aircraft is still flying as a multirotor using the copter controller. Your maximum motor tilt angle which is used by the copter controller seems high at 70 deg. I appreciate that you have said that your aircraft is very powerful in vtol modes but I still think that this should be brought back to around 50. Set Q_TILT_MAX = 50.
Once the aircraft reaches ARSPD_FBW_MIN the transition timer begins. Currently you have this set to 0.1 s which is making your tilt motors point level very rapidly, providing no additional vertical component of thrust to help your aircraft through the transition. Set Q_TANSITION_MS to 4000.
Thanks for your suggestions Matt!, Im going to do a test flight today with Q_TILT_MAX = 50 and Q_TANSITION_MS to 4000. I had Q_TANSITION_MS low in an attempt to get the transition to finish without having high airspeed. The higher i set ARSPD_FBW_MIN the more failed transitions i get… I have done much testing with this. I dont mind losing a little altitude on transition as this canard airframe stalls nose down every time. Its the inconsistency of tilt rotation degrees and tilt rotation speed that is my problem. Sometimes the tilt even begins un-transitioning on its own.
is it possible to add a parameter that would disregard airspeed ARSPD_FBW_MIN during the transition? Flick the switch and the tilt rotates based exactly on seconds and degrees. With no interruptions from airspeed checking?
is it an hour, couple hours, or weeks of programming work? im willing and i want to contribute! thanks
Big thanks to Matt at KH Unmanned! 10 out of 10 for the job setting up a custom transition and custom motor mixer to my specs. here is a link to a transition that looses almost zero altitude
I still have some tuning to do but i wanted to get a quick video out so Matt can see his work in action! more to come. thanks
can you (or anybody else for that matter) link me to where in the code this compensation is done if I am incorrect.