Kevin's Bi-Copter Thread

[Kevin_Erik]

The ducts fit / work awesome, though it does add allot of weight to the copter.

[jstremmler]

Did the iNav Blackbox analysis and think you should create a notch filter for pitch oszillations around 40Hz (from 30-50 Hz). And then I would reduce P for pitch significantly in the next trial… You have almost a harmonic pitch oszillation which causes the crash. As I guessed already, servos are creating the problem…
Btw today my external BEC arrived so that I can complete my bicopter in the next 2 hours. Will see how it goes…

Best wishes

Jürgen

[Kevin_Erik]

Started The process of removing the ducts. This should save approximately 118 g. after which I will default to copter back to David’s norms and add the notch filter you suggested.

Shit thing is that the Ducts looked and functioned perfectly. Even tried a few test flights inside, without fear of the props going ginsu on my sofa.

[jstremmler]

I totally agree, that your ducts are looking good, but they are missing the “Coanda Effect”, which is important for VTOL UAVs. Maybe youare going to redesign them a little…

Juergen

[Kevin_Erik]

The current Duct design primarily acts to smooth airflow, reduce prop tip vertices and act as prop guards. It does all three really well despite weighting 60 g. It’s possible that a FDM print using 50% infill could get this down to 40 g or less.

Keep in mind that a Liped-Duct provides for the Coanda-Effect and thus has a “self righting” ability. This may have negative effects on the Bicopter’s flight characteristics… As such, i opt’d to have it removed.

[Kevin_Erik]

Revised weight without ducts: 454.3 g

[Kevin_Erik]

@jstremmler If you look at David’s suggested PID’s he has the Yaw LPF set to 50 Hz, which i believe is being used to counter the same jitter shown in my video. I have set the Notch filter to the parameters you’ve suggested while leaving out the Yaw LPF. (Should be more efficient)

Will go for a flight today and attempt to make an FPV video, assuming it flies.
My current Diff is attached…

Attachments:

1. BiCopter-Diff-1-Oct-2018.txt

[Kevin_Erik]

Was able to complete the first flight!
Unfortunately, my DVR did not record the flight and worse yet the camera took a hard hit.
Most likely the camera is toast.

That being said, I did recover the BF Log for your viewing pleasure.
https://drive.google.com/file/d/1IgNtywRkWuXk8SRFoiBEN5uSzkUoU3Ug/view?usp=sharing
Also see the attached Plasma-tree of the same file.

My first impression is that the Bicopter is so responsive that it on the edge of being unstable the entire time. One moment flying nice and pretty and the next recovering from the slightest gust of wind. Was even ballsy enough to try a few Rolls! Loops… Ain’t no way in hell this is happening without some added stability.

Looking at the Log file, the Notch filter appears to be doing it’s job nicely and I’ll keep it for now. (30 – 40 Hz) The Terms, specifically the Roll’s I Term seams to be really high (As shown on Plasma-Tree) and thus is Oscillating.

Attachments:

1. 

   Bicopter-Flight-1-Oct-2018Bicopter-Flight-1-Oct-2018_0_response.jpg
   

2. 

   Bicopter-Flight-1-Oct-2018Bicopter-Flight-1-Oct-2018_0_noise.png
   

[Kevin_Erik]

Will try the following filter changes for tomorrows flight.

Attachments:

1. 

   BF-PIDs.jpg
   

2. 

   BF-Filters.jpg
   

[Kevin_Erik]

Just tested the camera, while it powers up, there is no RF output.

Unless there is some form of protective Camera mount for the copter, I’m not sure if its worth mounting another one considering the risk.

[jstremmler]

Great! you are making progress! Plasma-Tree analysis showed already that P- Roll has to be lowered (you did it already!, if I look at your revised PID values) Notch filter is working great.
I myself have extreme problems setting up my servos. Yaw is working fine but pitch is a desaster…. I am not going to do my maiden flight until having solved my servo problems…

[Kevin_Erik]

@jstremmler

IMO, Pitch and Yaw PID values “should” be somewhat similar to each other. This is considering that they’re both controlling the same Servos within the similar physical limitations. Main difference being that with Pitch, the Servos work together, while Yaw they opposite of each other. That being said, the Pitch and Yaw “Rise-Time” will be pretty slow in comparison to a Quad due to the limitations of the Servo and the distance the Load is from the Servo’s rotational axis. Shortening that distance should improve both the Plasma-Tree Rise-Time and the available Servo Torque.

Keep in mind that Servos are rated with the functional Load attached 1″ (25,4 mm) away from the axis of rotation. For every doubling of that distance, the Torque drops by half and I assume the response is more or less relative to the mass of the attached load resisting said Torque. Thus spinning lighter Motors / Props closer to the Servo, should reduce Gyroscopic loading and increase performance significantly. It’s why the old HQ props David has been using work so well. They’re really light and are not so aggressively pitched to present a significant load.

Note: 7.4 Volt Servo could also have a significant advantage with it’s higher Speed and Torque.

While I see that there are some gains to be made with BF, those gains may only work for my build. All others that follow may not have the same experiences. Besides, once dRonin is proven, there’s really no point in anyone pursuing a BF build that isn’t really optimized for Servos anyway. (Oh and yeah, dRonin Awesome-Tune) Beyond all of that the question is now, should I hold off on dRonin and stick with BF to see where the Rabbit hole goes?

[Kevin_Erik]

Did a quick test flight before the bad weather came in.

Roll looks significantly better and Yaw does a bit as well. Pitch however appears to be much worst so rolling that change back.

Attachments:

1. 

   LOG00001Bicopter-Flight-2-Oct-2018_0_response.png
   

2. 

   LOG00001Bicopter-Flight-2-Oct-2018_0_noise.png
   

[jstremmler]

Well my servos are behaving strange.

Yaw works perfectly but for pitch up they move to opposite direction and for pitch down also to the oppsite but same!! direction.

I have set:
Mixer BI
resource SERVO 1 A03 (former motor 3) and
resource SERVO 2 A02 (former motor 4) and

# servo mix
smix 0 2 2 100 0 0 100 0
smix 1 2 1 -100 0 0 100 0
smix 2 3 2 100 0 0 100 0
smix 3 3 1 100 0 0 100 0

Any idea what is going on?

Juergen

[Kevin_Erik]

You using BF?

I posted my CLI-Diff a while back. Have a quick look at that for comparison.

Also are you using an F3 or F4 processor?
I know that the F3’s have the code trimmed and thus the user has to manually enter the settings, while an F4 user such as myself can use the Bicopter defaults supplied within BF mixer type.

Oh shit…. Check your RC control inputs on the Firmware! You may need to manually reverse the Pitch RC input. After that test the RC control inputs to make sure they match. After that, manually move the Bicopter around in your hands to see if the servos motors fight against the changes in orientation. When you tilt forward, the servos should both tilt back. (and so on and so on)

[Author]

Posts