Extreme drift resulting from black magic

[grifo7]

Hello everyone,

I'm having a problem with both my CF2s and I was hoping you could help me.

The problem is extreme drift. I was testing my programs with ROS, making it fly autonomously and out of a sudden it started drifting a lot.
At first I thought it could be my software, but when I tried to fly it manually and I got the same issue. So, I figured it could be the motors or something like that, and I replaced every part (all 4 motors, plastic motor holders and propellers). Still the same.

At this point I thought the problem could be on the main board. So, I open my completely new CF2 and I mounted it with completely new parts, which I had stored in case the first one broke. The first couple of tests with the new CF2 worked well, but then... the same type of drift again!

As a desperate measure I even flashed the last 3 firmwares releases (2017.04, 2017.05 and 2017.06) in both CF2s, to see if there was any difference. Nop, still the same drift.

I tested it with a simple program that simply gives thrust to it, and manually with both the Android app and the BitCraze client with the CrazyradioPA and controller. Again, the same. Tried also to give it pitch and roll trim values (even extreme ones like 5 and 10 degrees) which hold it in place for around 3/4 seconds but will drift eventually in the same direction.

When connected to the BitCraze client, it reports around +/-0,3 in Pitch and Roll which I think is normal, so I'm really without a clue of what to do next

The drift is ALWAYS in the NW direction in both CF2s. Every once in a while (1 test out of 50, really rarely) the Crazyflie works well.

I know I didn't provide much data but I don't even know what to provide. Maybe someone already had a similar problem and could help.

Sorry for the long post,
Thank you in advance,
Grifo

[tobias]

It could be a sensor issue but we have also seen extreme drift because of heavy vibrations. Since you seam to have tested this already I'm more leaning toward a sensor problem. A last test for this I suggest trying the propeller test FW we made. You can find it here and please post the cfclient console result here.

[sarthakjain]

Hello everyone,
I am facing a very similar issue as mentioned above. But there are a few differences!

I am using a Feedback from Optitrack motion capture system to fly my Crazyflie. When I am giving a set point to my crazyflie, it goes to that point initially, hovers for around 5-15 seconds and then suddenly it starts to drift in a particular direction (Always the same) and then crashes. Note: I am only facing this issue with one of my Crazyflies. Others are working fine!
I have no clue why this is happening? I have tried all the things as Grifo did, but in vain.

@grifo7 Were you able to solve the issue?? By the time, I'll try and perform the test and post the results here.

Thanks in advance!
Sarthak

[sarthakjain]

Hey,
I am back with the test results!

I performed the test for a few more times, but those didn't show up any propeller test results for M2. In this case as well, M3 motor propeller test didn't show up!!!
What could be a possible reason? What do these numbers mean?? Are my motors and propellers working fine.

Thanks in advance!
Sarthak

[tobias]

To me it sounds like a sensor issue, probably communication problem with the sensor chip MPU9250. Has the CF2 ever failed the startup test? If so it needs to be reflowed or the board replaced. See this post of possible solutions.

Also looking at the propeller tests it looks like M1 has a lot of vibrations. Would be worth checking out. M3 doesn't show up because it does not fail the test. Good luck!

[Grifo]

Hi!

@sarthakjain, unfortunately I still didn't had the time to play with this. So I'm still with the same problem.

Once I have the time, I'l perform the mentioned tests.

Thanks for the help,
Grifo