Dual-motor tailsitters

MissionPlanner offers a 3.8.1beta1. Is this the version to use Pauls system ?

Rolf

no, I havenā€™t merged Pauls changes yet. You can see the changes in 3.8.1beta1 here:

And this is the first price for a really good realisation of the VTOL Tailsitter.

No risk, no Tuning stress, no support needed.
and cheaper. Just Fun.

it does look like a lot of flying fun. I wonder what micro-controller it has in it? With the work by Sid to support ChibIOS we may be able to port ArduPilot to lower-end micros in aircraft like this.
For me a lot of the fun is in the coding, not just the flying :slight_smile:

Yes, thats our impression.
But it would make more coding-fun if it works in reality like shown and not
in a simulator only.
There is a great gap between, filled up with hundreds of Params with the result of 118MB for 10 Min. of
a lot of unknown Log Datas.
Non coding specialists canā€™t interprete this to reach the target like Horizon e.g. to make a Product.
May be, thats the reason of the escaping or missing user-followers.
By the way, the same as wiht PX4 and QGroundControl.
Kind regards
User Otto

[quote=ā€œlorbass, post:613, topic:15302ā€]
No risk, no Tuning stress, no support needed.
and cheaper. Just Fun.
[/quote
I am developing my system with the hope of a useful tool not just fun and that requires lots of risk,stress,and work and that,s what makes it fun for me.And we are doing stuff that has not been done before so that makes it interesting .

Iā€™m not fan of buy and fly.
My understandig was to test the Master Firmware and to give feedback such as positiv results or issues in order
to help to make it perfect.
This costs material, time and nervs. And a long waiting time for the ordered material. And Iā€™m aware of the risk.
But verry early the feedback was ignored and question not answered and so I stopped this. To get the wing flying I made 23 Testflights with trial and error to find the appropriate Params. For me the 118 MB of Log Datas for 10 min.didnā€™t help to find the issue. Finally found the right Params with systematically tests in a Rig for each axis individualy.
All I can say, they react verry, verry critically. A little bit higher or lower and the Wing behave crasy.
Without support from Professionals its difficult.
And now frustrating, a cheap Product can all this I worked for weeks:frowning:
By the way, did you recognize the smooth Transition?
Here just a part of it: (Pitch and Roll)
Tuning Pitch https://drive.google.com/file/d/0Bxldq38bfbqHYlJEbHBab2EzaW8/view?usp=sharing
Tuning Roll https://drive.google.com/file/d/0Bxldq38bfbqHaDhYallzeURlbU0/view?usp=sharing
Will see, if a least FBWA works as it shouldā€¦

So was it fun?..and your data is helpful
I agree, too complicated, lots of unknown params.
I will keep reading and trying , And find some very useful params
so itā€™s fine that I have to learn Ardupilotā€¦ and figure out how to read the logs.
the professionals are very busy with other great projects so itā€™s ok
Big THANKS to themā€¦btw, they donā€™t have all the answers, thatā€™s what we do
also horizonā€™s stuff is limited in what it can doā€¦tailsitterā€¦ old techā€¦meh

Dear Lorbass, be a bit more confident in yourself and just fly your perfect hovering tailsitter instead of complaining and blaming the devs in advance for your own lack of expierience in building and flying unmanned planes.
Our Tailsitter (and tiltrotor and two conventional planes) are flying with arduplane without issues. Be sure, FBWA works ā€œat least ā€¦ as it should.ā€

Yesterday (gusty wind 20 kmh) we had a citical situation while backward transition:

We donā€™t realy know what happens.

Logfile: https://www.magentacloud.de/share/ualbgv5hup

Rolg

@Rolf Iā€™m away for a couple of days, but I just wanted to say that the log shows its a state estimation issue. The two EKF instances disagree with each other on attitude, which is a red flag, and the position estimates went crazy
more when I get home ā€¦

Hi, does anyone know if aileron and yaw/ motor mix , available in TVBS and what is the parameter? looking to mix a fair amount of yaw( motor,) with aileron on the ft versa, wonā€™t turn so good with just ailerons and probably not so good in auto, no vertical surfaces
thanks in advance
Rick

Iā€™m guessing you mean when in forward flight?
The parameter you need RUDD_DT_GAIN. That is a number from 0 to 100 for ā€œrudder differential thrust gainā€. It is controlled by rudder output, allowing you to use differential thrust for rudder to turn faster.
If you want the ā€œrudderā€ to be automatic, then set KDD_RDDRMIX, which is the ā€œfeed forward from aileron to rudderā€. So whenever it uses aileron, it gives rudder equal to aileron times KDD_RDDRMIX.
Iā€™d start off by using the rudder stick on the transmitter and see how the plane reacts while in FBWA mode, then raise RUDD_DT_GAIN until you can do flat turns with just rudder input.

Hi, Tridge, yes in forward flight, It flyā€™s fine but need lots of rudder to coordinate the turn. have been playing with both rudd_dt_gain and kdd_rddrmix but was not clear about it, was hoping for a simple yaw to aileron mix param.but this is helpfulā€¦thanks ā€¦also is it possible to get some kind of index of the parameters for the specific type of model. Hard to know which yaw param work with TVBS for instance. it has a weird flat turn with no vertical appendages keeps going in the original direction after the flat turn for a bit

Welcome to the club !

It had happened the same very often with my setup while in Qstabilize my plane start to fly leveled , in a kind of auto transition.

With only little wind the problem did not occur.

We also decreased Q_A_ACCEL_P_MAX (from 110000) to 40000.
The transitions are smoother now.
Thx to lorbass for this hint.

Regards Rolf

@tridge

The first FBWA Flight ended with happy feelings.:sweat_smile:
As requested, the link to the video and the log.
https://youtu.be/DHYB4iTebJA
https://drive.google.com/file/d/0Bxldq38bfbqHdmlKd1d5bktES1k/view?usp=sharing

Thank you for analyzing the ā€œFlameoutā€ during the backtransition.

The Param Q_A_ACCEL_P_MAX ist not yet modified from default 110 000 to the recommended 40 000

Regards
Otto

1 Like

@lorbass

Congratulations, Otto.
Welcome to the club of the successful.

Regards Walter

Thank you for the patience :wink:
I think to continue
Otto

Congratulations on your first transitions! I know itā€™s been very hard to get good tuning, and Iā€™m sorry its been frustrating for you at times.

Iā€™ve had a look, and I think based on this log Iā€™d like to make some changes to the way we handle back transitions.
First though the first transition to FBWA. As youā€™ve already suggested, it went over backwards as it was leaning back when the transition happened. The reason it was leaning back is there is no I term set on the pitch controller (ie. Q_A_RAT_PIT_I is zero). Without that I term it is unable to learn the right trim value for holding itself upright. Iā€™d suggest you try Q_A_RAT_PIT_I=0.1 to start with, but you may need to adjust.
For the back transition (from FBWA to QHOVER) the code just goes sharply from fixed wing flight to hover. So the hover controller is started with a pitch error of 90 degrees. Weā€™ve seen that some planes handle that fine, but yours really didnā€™t handle it at all. Tuning may help, but I think it will be better to introduce a transition where it continues to use the fixed wing controller until it manages to bring the nose up a bit.
Iā€™m thinking weā€™ve had a back transition where the fixed wing controller would put in maximum up pitch rate until either the pitch reaches 60 degrees or one second has elapsed. Then it would hand over to the bi-copter controller. That means the bi-copter code would be starting with a much smaller pitch error, which should make it much happier.
Iā€™ll implement this and then if you have time to test it would be appreciated. Iā€™ll do initial tests in the simulator.
Cheers, Tridge