Crazyflie wants to flip from take-off, unstable in hover

[awk]

My crazyflie flips from takeoff, if I suspend it from a bit of string and apply thrust whilst it generally stays level it will veer off in one direction or another. I've double checked both the prop order and orientation and the wiring (see photos in album below). There doesn't appear to be any hair or similar stuck anywhere and the motors spin freely. I've also adjusted the radio channel to Channel 80/2Mbit which doesn't seem to improve anything.

Any other suggestions ?

(Photos of build and takeoff : https://www.dropbox.com/sc/owdnskn611dv2mx/YigDRayzOB)

[marcus]

Hi,

The propellers looks good, but it looks like you may have mixed M3/4 while soldering. Could you post an image of the back side of the Crazyflie so we can see which motors are soldered where?

[awk]

I added a photo of the connections on the reverse :

https://www.dropbox.com/sc/jgitoea5dmms9ia/my0gmItXiv

I think the connections for M3/M4 are correct - The only thing that strikes me as slightly 'odd' is that from a wiring length/layout perspective M4's solder pads are closer to M3's mount - so M4's wire runs over the top of M3's

On the other hand having the pads in M1, M2, M4, M3 order is likely to confuse too So I get it

[marcus]

Everything seems ok from what we can see. When you run the client, does the sensor output/horizon look ok?

It could also be the ground affect you are seeing, since it's not really taking off. What happens if you push the thrust a bit so it gets into the air? Does it flip over? And does it always flip over the same way?

Also check so that the reading from the input device seems ok. Does it make sens for roll/pitch/yaw/thrust?

[awk]

Thanks for the quick responses!

The client shows a stable artificial horizon when the quadcopter is on the floor etc. There's a bit of change in the Yaw value - but it's less than 1 degree per second (perhaps 0.3-0.5 degrees per second).

If I give it some extra thrust to try and overcome ground effects it just seems to flip faster 8-( I did try suspending it from a bit of string to avoid the ground effects, it doesn't 'flip', but it will 'pull' in a direction. I need to repeat that experiment to see if it's always in the same direction.

I'm away from that computer so I can't check the input device readings - is there a simple way to do that in the client ? I don't recall seeing those readings in the UI?

(I'll update things later when I'm back at home with access to the computer and crazyflie)

[marcus]

You can see if output from the device in the UI without it being sent to the Crazyflie. The target values should change if you move the joysticks on the input device.

[awk]

Indeed the 'at rest' values were off a little. I was able to trim them out using the UI to get roll & pitch to zero - it's still difficult to fly, it's very prone to drifting quickly - but it's not 'flipping' anymore.

One problem is keeping the 'front' consistent. From the perspective of roll & pitch is M1 the 'front' ? Perhaps trimming to zero in the UI controls isn't enough and I need a little more?

[tobias]

You could try playing with the flight parameters a bit to make it easier to fly. E.g. like in this post.

[awk]

I tried the alternative parameters suggested, and also tried taking off from the corner of a table (which appeared to reduce the ground effects some). However I'm still struggling greatly to maintain a reasonable hover. My point of comparison is controlling other RC electric helicopters with counter-rotating blades and no tail rotor. The crazyflie will move rapidly (usually to the left) almost immediately after takeoff and trying to correct for this can't be done fast enough (IMHO).

The other RC helicopters take off and hover much much more easily than the crazy file. Do I have unrealistic expectations of how easy it is to fly the crazyflie ? Or is something truly wrong with mine?

[tobias]

No it should not be like that so something is definitely wrong with yours. It should be very stable and take off straight up. Have you identified towards what motor it tend to drift? I'm thinking it might be a bad motor.