Big Tandem Helicopter with external Flybarless System – Loiter Mode

[PatrickS (Patrick)]

Hello Guys,
After a long time I need help from the community again.

Currently I am working on a pretty big Tandem helicopter for a company, which is why I can not post any photos or videos of the uav.

For reference – it is two 600 sized 2-bladed heads with two seperate powertrains (not synchronized).
AUW is changing with further development of the project but around 24-27kg.

We use an orange cube with a seperate flybarless controller (Bavarian HC3SX with tandem firmware).
We are using two GPS modules for yaw if this is an important information.

Unfortunately there is not much information about setting up a pixhawk with external FBL, but enough to get it working.

So i started by setting up the flybarless system and tweak the settings like on every „normal rc“ helicopter and it actually flys pretty good in my opinion.

Next step was integration of the pixhawk. I started in stabilize mode with zero PIDs for the pixhawk and tuned the feedforwards until the desired and actual rates of all axis where as close as I could get them. I also increased the accelaration values to get an even better matching oft he two graphs.

Stabilize mode also works really good for my feeling, apart from the yaw feeling – it is like negative 100 expo which means I do all my yaw movements within a few millimeters on the sticks. It is hard to descripe but it is just way to aggressive compared to the feeling in acro mode.

Next Step was tuning the notch filters which worked like expected.

Last step was loiter mode. With standard settings it hovered but felt like a rubber band in the air.
After inspecting the log files I tuned the PSC_VELXY_P all the way up to 2.2 and it felt a lot more locked in.
Also changed a few other parameters like LOIT_ACC_MAX from 200 to 175 to make the stops a little bit better and PSC_VELXY_D to 0,1 because while hovering there was always a little bit of back and forth moevement on the pitch axis which got better with the d gain but is still present.

Thats the point where it flew actually pretty good but if you watch the log files you can see that the PSC „DAN“ and „TAN“ graphs are far away from matching.

Until now the batteries (pretty big and heavy – 12S 30Ah) where mounted rigid to the frame and caused vibrations which is why we wanted to mount them on rubber dampers.
I chose pretty hard ones so it would not get to weak.

The next testflight with the dampers, but all same config for pixhawk was pretty much unflyable.
Could it really just be the dampening or do you have another hint?
Also which parameters need to be tuned for the new configuration to work?

I really like the dampening, all vibrations are gone, and flying in acro works still totally fine.

So my questions are:

• Why is yaw behaving so strange in any „Ardupilot Mode“? (I tried lowering the yaw slew rate for loiter mode with no effect)
• What could I try to get the acceleration graphs matching better?
• What is happening since the batteries are not mounted rigid?

I really appreciate any help from you guys.
I am working on this project every monday, tuesday and wednesday, so if you need more information just let me know! The flying field is nearby so I can collect flight data if needed and test everything you guys recommend pretty quickly.

Thanks for any help!

best regards
Patrick

P.S.: parameter files + log files from the different „set up stages“ are in this dropbox folder because I can not upload the .bin files here directly (error message every time).

ardupilot

[PatrickS (Patrick)]

Good morning,

Almost 100 views but nobody wants to talk with me?
I would really appreciate any hints from you guys, also thinking the questions are not that helicopter related.

In the meantime we switched to smaller batteries and a different mounting system, so the dampening is gone. Have not made a testflight yet to see if loiter flys like before with the batteries mounted rigid.

best regards
Patrick

[bnsgeyer (Bill Geyer)]

Hi Patrick,
Pretty interesting project! I have refrained from responding because I have a tough enough time giving advice on tuning using the built in stabilization let alone guessing what’s happening with ardupilot interacting with external FBLs. I understand why you are using FBLs but I don’t necessarily agree with it. Ardupilot is capable of controlling a tandem helicopter without FBLs. Probably not as refined as an FBL but it’s getting there. But I am glad you were able to figure it out and get it working well in stabilize.

@PatrickS

Why is yaw behaving so strange in any „Ardupilot Mode“? (I tried lowering the yaw slew rate for loiter mode with no effect)

The reason setting up with an external FBL is hard is that you have to ensure that the FBL is only acting as the rate controller. It cannot be doing any attitude feedback. You want to let ardupilot control the attitude. You should be tuning only the rate FF, like you did, and then also the angle P param. So for the yaw axis, if the FBL has some heading hold feature then I would imagine that it could be interacting poorly with ardupilot. Total guess on my part.

@PatrickS

What could I try to get the acceleration graphs matching better?

I don’t have a lot of experience tuning the position controller so I am not much help here

@PatrickS

What is happening since the batteries are not mounted rigid?

My guess here is that since the batteries are such a large mass compared to the total weight of the vehicle, any movement that is allowed could have a huge effect on the vehicle dynamics. Kind of like water or gas sloshing around in a tank. I don’t think putting dampening of the batteries is the right way to go. My guess is that your frame is not stiff enough and a flexing mode of the frame itself is resonating with the rotor and causing the vibrations. You need to move the frequency of this frame flexing mode away from the driving frequency which I think would be the rotor.

Hope this is helpful.