Servers by jDrones

Dual-motor tailsitters

(letpi) #1504

Static margin is 4.5 %
Next WE I will make more test in order to move a little the CG. I also have to add airspeed. I can see some small pitch oscillations in the log, 5 to 10° of magnitude, but I do not see them while flying. I think a big problem is the quality of flaps servo, you will see that while flying fbwa or Auto the PWM range used is very very low, less than the gear play.
I use Corona fast HV servo connected to a 6V BEC
They are very very fast but have some amount of play, not bad for a low cost metal gear servo but maybe too much for these large flaps.

I think the wing is able to carry more weight but of course the CG must remain at the same position for all 3 axis.

(Mark Whitehorn) #1505

@losawing Congratulations on the batwing, very impressive performance.

I got my little quad tailsitter back in the air with new BLHeli_32 ESCs and it flies pretty well at all pitch angles in VTOL modes. Here’s a plot of total current vs GPS speed as pitch goes from 0 (nose up) to 60 degrees:

It looks like the small wing is generating some useful lift at about 9 m/sec
This is what it looks like hovering in wind:

I think I’ll put a bigger wing on it before attempting FW flight though.

(letpi) #1506

Thanks for the update on your project. To begin with something not too fast and get experience is probably the faster solution.
Yesterday, I have been able the reproduce the bad back transition with my test wing and the quad configuration. After 4 or 5 successful test transitions I switched to manual and made some loops, then the EKF went red for few seconds, I switched back to FBWA, tried a first back transition that failed and a second one that also failed. I still don’t understand what the problem is but I am working on it.

(Mark Whitehorn) #1507

Were you testing using my new branch pr-quadTS?
I’m not seeing any problems with transitions in SITL:

But I haven’t tried using manual mode; it could be that is exposing a bug.

update: just tried several transitions from MANUAL to FBWA and QHOVER in SITL, and only had 1 problem: transition from knife-edge manual flight to qhover remained in knife-edge, but that’s an airframe limitation, not a bug.

(David Kaden) #1508

FYI: Wingtra wrote a little “Quadplane vs Tailsitter” comparison.

Also anyone tested the (NEW!!!) QAUTOTUNE feature yet?? I have no tailsitter at the moment, so please guys test this and provide tridge with feedback! :slight_smile:

There is some advice for tailsitter:

NOTE : QAUTOTUNE does not work on axes that need feed-forward. This means it doesn’t work on the pitch axis in tailsitters. If you want to QAUTOTUNE a tailsitter then make sure the pitch axis is disabled by setting Q_AUTOTUNE_AXES=5. If you do use QAUTOTUNE on the pitch axis of a tailsitter then you will end up with a very bad (completely unflyable) tune.

(Peter Hall) #1509

Thanks for the link, not sure I really agree with any of there points.

Having played with both there is a lot to be said for quad planes, defiantly not as cool as tail sitters tho.

(letpi) #1510

No I am still testing the stryker_quad branch. I made a clone of pr_quadts but the the method I used to compile with MAKE does not work anymore. The wiki say that to use MAKE is now not recommended. I also tried the cygwin method but again it was a failure. I am afraid I am very bad at all kind of computer related stuff !!
Nevertheless I have been able to reproduce the bad back transition a second time using the same process: several good transition/manual acrobatic flight/EKF red/ FBWA/ transition /crash. Hopefully the plane is very though. Every time I had a bad back transition, the EKF turned red for few seconds (I see the messages with telemetry logs). At high speed with the jetwing I had error velocity or error position message. I have the same messages when I fly manual with my test wing. I have to mention that both the test wing and the jet wing use old ublox 6M GPS and maybe they are not accurate enough. I think I will load on my test wing the last arduplane firmware in order to make the same flight test. I will see how it behave. If I get a bad transition again I will try to improve the hardware.

(Mark Whitehorn) #1511

I put pr-quadTS binaries for fmuv3 and fmuv4 here:

Let me know if you need a different build, and I’ll add it.

(James Near) #1512

Hey! I’m making quite a large tailsitter right now and have 20kgcm servos with a combined motor/prop weight of around 400g at around 5 cm above the servo gear, so I have around 2kgcm of torque during static operations. I was wondering what you guys are seeing during real flights. I assume that the torque is higher with the moments applied by the props. It seems like this servo should be good enough but will definitely test before. I just want to see how y’alls servos are holding up right now. I would love to get my hands on higher quality servos but I can’t seem to find anything other than the ones linked from aliexpress that make it easy to design a tilt mechanism for and are also 180 degrees.

(letpi) #1513

These servo should be very good for your application. According to my small experience with them, JX servo are strong and reliable.

(lorbass) #1514

Just to compare, I use for my vectored Tailsitter Skywalker 6 this Hitec Servo: hs-5245mg
Rated with 6V 5.5 cmkp stall torque, but measured max torque 3.15 cmkp at 1.1 Amp.
And 2 electronic failures fortunately on ground. (First to see in the video below)
Nown as high quality but only high price.
Motors Quanum MT4108-580 100 gr, Props 12x5.5 with 12 gr each.
More infos:

(letpi) #1515

this is done.
I am really impressed. I thought my tailsitter was well tuned but I have to admit than the autotune do a better job in 5 minutes instead of hours. Only the pitch axis has to be tuned manually.

(Peter Hall) #1516

did you try a pitch autotune? would be interested to see how it did, maybe we could improve the autotune to work better in this case.

(David Kaden) #1517

Awesome! If you could upload a video of the wing while doing autotune would be cool!

Am I right with assuming that you need a kind of working PID setup before you can do autotune? I mean we can not just put it into autotune without it even flying at all, right? So we need to make some basic PID tune before. Whats the most easy way there? With the rig is quite complex.

(letpi) #1518

it is written that the pitch autotune will end in an unflyable tune so no I did not test it. I will test it with my test wing but only when other tests related to my back transition issue will be done.
I saw that your qtailsitrllmx parameter is in the last built. This is a nice feature and I will test it soon.
Also I shortly tested the weather vanning some time ago in qloiter mode and it works. But the loiter controller make a bad job to stabilize the wing as it becomes very sluggish. With 10 to 15km/h of wind it ends in a ugly dance made of uncontrolled roll and yaw large oscillations.

(letpi) #1519

you ask if anyone can test and provide tridge with the feedback, this is precisely what I did :grinning:
directly to the related blog and with a short video.
From what I have seem the autotuning is not aggressive but of course the wing must be flyable. To check if the wing is flyable I simply hold it by hand to verify is there are not too much flaps oscillations and feel how it behaves. Once this is done I go flying Q_hover and firstly tune the pitch. I begin with Q_angle_max around 20 and test how the wing recover from this angle. If you feel the wing is sluggish, increase P, if it overshoot, increase D, if it is out of trim for too long increase I. Once I am happy with 20° max angle, I increase it with 10° steps. As I said it can take some hours…

(Kikislater) #1520

lol !
They compare a vtol 14kg with wingtra 4.5kg. Wingtra at 4.5kg is heavy ! If we compare with a foxtech nimbus vtol it’s around 4.8kg, not 14 …
And comparison will be laughing when they told about ground impact. Do we have to compare ebee at 0.8kg vs wingtra at 4.5kg … No comment !

(David Kaden) #1521

14 kg may be a bit to much I think. But the point of comparison should be the Payload and Range. So a Tailsitter must be lighter than any other VTOL with same range and PL. On any other configuration you have more structure and also more components.

E.g.: You have a stock quadplane, then you will have both the motor for vertical and the ones for horizontal flight. The vertical propulsion alone will be at least the same weight as the one of the tailsitter. Then you have more ESCs, more structure, etc. All summing up to more weight, meaning that you will need tendentiously more power to hover, bigger motors, bigger batterie etc. Always a big rat tail with added weight. Pls do not underestimate that.
So actually I wouldn’t call it unrealistic to compare a 5 kg tailsitter with a 10 kg VTOL of any other configuration. But calculation of real numbers are topic for at least a masters thesis.

I am really convinced that the tailsitter configuration will have the highest payload percentage of MTOW of any other VTOL configuration. Meaning that with staying unter an certain weight cuz of regulatories you will have maximum PL capacity with this configuration.

This is why I love tailsitters so much. (not TVBS :wink: !!)

Thanks, didn’t see. Maybe he also needs a Log.

(Rolf) #1522

That’s for sure, as long as you just hover :smiley:

The simple quadplane is already more efficient in terms of range, because you can optimize engine/prop combinations separately for lift and forward flight. Only with a variable pitch propeller it would look better for the tailsitters. Just my datas from our tri-tiltrotor, vectored tailsitter and quadplane experience.


(Wasim Akram) #1523

I don’t intend to hijack or misdirect this thread, but I do think that it is unreasonable to compare a somewhat high-end 4.5kg tailsitter plane to a poorly-built 10kg VTOL plane or a relatively much cheaper Nimbus VTOL plane (which still performs fairly okay btw). Although I would slightly agree with @kikislater that Wingtra at 3.7kg (empty) is slightly heavy, it really depends on your perspective. Wingtra may be slightly heavier but it certainly looks more durable than a Nimbus VTOL plane or a typical foamie. VTOL planes can be designed and built with other criteria in mind, apart from (max) payload fraction and (max) endurance, such as wind resistance, applicability etc. Recently, we managed to build a custom composite quad-plane tilt-rotor with empty weight in the same ballpark as Wingtra but with relative higher payload capacity and endurance. I’m not saying that quad tilt-rotors are better than tail-sitters because in reality the choice of (VTOL) plane configuration can be subjective to one’s requirements or personal preferences. At this stage, I think the best thing to do is focus on developing the different plane configurations, rather than jumping on to conclusion as to which VTOL configuration is superior.
Anyway folks, I didn’t actually mean to divert this thread much, so I pass on my sincere apologies. Peace :wink: