• This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn more.

HELP: APM 2.5 board plays the spinning top!

#1
Hi guys,

I'm building an H-quad 1mx1m (a bit over 3'x3'), and I have the power system finally figured out thanks to guys in the forum here. I'm using an APM 2.5 board from 3D robotics with a separate GPS (but that works, I get a satellite lock in about 20 seconds)

But, when we're trying to fly, here's what happens:



I found this link, http://diydrones.com/profiles/blogs/resolved-arducopter-2-5-3-yaw-issues and tried the magnetometer calibration, but only with a live calibration. Didn't work.

I updated the firmware to 3.0.0 and tried again with similar results. I'm at a loss to explain why I can't get any kind of yaw input at all! Does anyone have any idea what might be causing the problem?

Thanks
Goeland
 
#3
ur frame is way to flexible when the unit detects yaw its trying to spin up the motors ant the frame is twists preventing it from correcting
 
#4
So, there is some yaw input when on the ground, the props spin in the correct direction. But in flight it's just spinning itself. And yes, I am using counter-rotating props. The APM/power setup worked fine on the DJI F-350 frame. I just went to 4S instead of 3S.

Jamiedco, if the twisting was the problem it wouldn't be able to stabilize horizontally at all, would it? It'd just be wobbling back and forth. The "wobbles" here are actually my control inputs to get it forward or back, to avoid hitting objects as it spins.
 

colorex

Rotor Riot!
Mentor
#5
Your frame is twisted. However rudder input should either slow it down or speed it up.

Maybe check your receiver outputs?.
 
#6
You can see in the video how the arms are twisting considerably in relation to the center of the body where the controller is located. For all intents and purposes, the controller is level doing its job. You need to a stronger frame.

It looks like the props you're spinning are too small for as big of a frame you have; meaning there is no reason for the frame to be as big as it is. Make the frame smaller to make it more rigid; arm length 1.25 times longer than your prop length is a good starting point (e.g. 8" prop means 10" boom). Might even consider picking up a any copter frame build kit; easy way to get strength.

From the glance it looks like your prop and motor direction should be OK, just sort out the flexing. Oh, and when taking video with your phone, turn it to landscape orientation before pressing record; it produces a video which fills the player's frame.
 
#7
So there's a purpose to the 1x1m frame, and eventually the weight will drop significantly. I'll rebuild the frame using spruce instead, it's cheap enough. But I don't think the flexing is the ONLY problem. I was not getting any yaw changes when giving yaw input and that's what bothers me.

Can anyone tell me if this assumption is correct? To yaw clockwise, the clockwise props spin up and the counter-clockwise props spin down, and vice-versa for counter-clockwise yaw motion?
 
#11
Yes, it's at the top of the list - spruce frame is still separate pieces at the moment. But knowing which way to make the props spin will also help. When I was looking for that image yesterday I was getting the error "project has reached it's quota", whatever that means.
 
#13
Ok, so... Built the same H-style frame with pine bars instead of fiber-glass covered balsa, but the "fuselage" (the central bar) is still twisting along the length, leading to the same rotation issue.

Does anyone have a suggestion as to how to strengthen it to avoid that type of problem? Diameter is 140cm, and the pine struts are 18x18mm cross-section. The central support is done with 3 instead of 2 bars, spaced 10cm apart (middle one in the center), but instead of plywood we used balsa sheets. How thick of a piece of plywood should I use?
 

xuzme720

Dedicated foam bender
Mentor
#14
Just box the middle section with the same stock on the perimeter as you have used on the booms. sheet top and bottom with a light ply 1/16 or 1/8 should be plenty.
 
#15
Alright that's been done. And it works better, but still has yaw control issues. Telemetry feed to the PC shows that the compass values are correctly detected. So it seems as though the APM is not using the compass properly to determine heading.

Anyone have a clue where to look for a solution?
 
#17
Thanks. Maybe I should downgrade to an older version of AC... Like 2.9.1b or even 2.8.x or something, where it will work without being too futzy. And upgrade again when it's been thoroughly tested.
 

Tritium

Amateur Extra Class K5TWM
#18
Thanks. Maybe I should downgrade to an older version of AC... Like 2.9.1b or even 2.8.x or something, where it will work without being too futzy. And upgrade again when it's been thoroughly tested.
Yea I never fly the latest firmware. I prefer that it be debugged first so I generally wait for 3 months at least. The latest is VERY sensitive to vibration and electrical noise causing compass issues and also the "calibration dance" is required during initial setup to calibrate the sensors.

Thurmond
 
#19
Yea I never fly the latest firmware. I prefer that it be debugged first so I generally wait for 3 months at least. The latest is VERY sensitive to vibration and electrical noise causing compass issues and also the "calibration dance" is required during initial setup to calibrate the sensors.

Thurmond
Yes, and sadly it took a while for me to find the youtube tutorial about that calibration dance and the compassmot setup (running motors up with props pulling the copter down to the ground instead of up) to compute motor interference of the magnetic field.

Anyways, that's been done and it seems that 3.0.1 should work... Sort of. But now I've got a problem and I think it might be caused by the PIDs.

Keep in mind that in both videos below I am ONLY in stable mode, I never switched out of it. (Or edited the video clips for that matter, aside from youtube's video stabilization feature, figured you wouldn't want to spill your lunch watching them.)


Notice that in the video below I go up and then come back down - my throttle stick has NOT moved from nearly full position (75 or 80% maybe). I don't understand it, clearly the motors have enough punch to lift the quad, but the APM brings it back down? Weird!


I suppose I could revert back to 2.9.1b or even 2.8 - the list of improvements doesn't seem that necessary to our project (we're a team building this and I'm the pilot). Problem is that I don't know where I can find the 2.9.1b or 2.8.x hex files (I downloaded the 2.8.x source and have it compiling on arduino, thinking I might upload that directly on the board if all else fails, but I'd rather have access to the "official" versions.)

Any clue as to what I should do for PID settings? Up the P's, lower the I's, or what? I'm hopelessly confused about this particular situation. It seems to be very very slow to implement/correct attitude changes. Of course this is on the default settings for the APM, which uses a much smaller frame.

Thanks!