[CLOSED] Problem with IMU?

[Call_me_a_Cab]

Hi guys,

I haven't really played with my CF2 as much as I would like and that's because I suspect there is a slight problem with the acel/gyro on board.

Basically when I fly (or pick up and put down) the CF the roll axis takes a little while to settle even though the flie is still. I previously posted about this and Marcus advised it was most likely because the props were unbalanced, I now have balanced them as good as I can (spent an hour) and the problem is still there.
It doesn't sound like much of an issue but sometimes when flying I seem to loose roll in one direction and I'm constantly Left and right with the trim.

Here is a video of the flight data and the plotter with me picking the flie up, giving it a quick shake and then setting it down. At first I thought it may be the PID loop recovering from "error" but then I figured that the PID(s) must only output on target data and not actual?

http://youtu.be/Va5RZ-2vxeU

The Video is not the best instance of this happening but you can get the idea, Especially right at the start of the vid.

Could anyone give any insight as to whether this is "normal behaviour" or I may be seeing a fault with the HW?

Thanks in advance,

Cab

[chad]

Here's a video I posted of how my artificial horizon reacts to the motion of the airframe. Maybe it'll help you to compare your Crazyflie's behavior?

[Call_me_a_Cab]

Thanks Chad,

Its hard to see but it looks like your telemetry settles immediately after motion stops. That makes me worry even further...

I think tonight I will remove the motors and mounts and check it again just in case any flex in the mounts is causing the telemetry issues and then I'm making a meal of it in mind regarding the trim.

[tobias]

This behaviour is normal for the imu. When moving it fast enough the sensors (mainly gyro) can saturate, resulting in wrong measurements. The sensor fusion filter will then over time correct this behavior using the accelerometer data. This can take a while to settle, depending on the configuration parameters.

However, a bad gyro could also show up as an effect like that but looking at the video I don't think that is the case.

If you fly in a room without any draft does the Crazyflie ruffly stay in the same place with only small stick movements needed to keep it there?

[Call_me_a_Cab]

Tobias,

Thanks, yes it does stay roughly still but only with outrageous trim settings for roll (Sometimes up to 14) then once I have it staying still, after flying around a bit and trying to stay still again, the trim needs to be changed.

I'm going to check it's not a controller zeroing issue by flying with my brothers iphone tonight and then I'll report back.

Thanks,

Cab

[Call_me_a_Cab]

Today I was unable to get flying with an Android tablet I borrowed as the trim range from -0.5 to 0.5 was insufficient to gain any control.

Can anyone suggest any next steps please?

[tobias]

This sounds a lot like a vibration issue from the propellers. Have you balanced the propellers? Here's a guide.

[Call_me_a_Cab]

Yes perfectly balanced...

[tobias]

Hmm, this is very strange. Is it always the same direction you need to trim it or does the Crazyflie become "nervous" and take off in different directions? Does it get worse with time?

I have attached a firmware that runs the gyro and accelerometer manufacturing test. This test should tell us if there is something strange with the sensors. When you startup with this firmware make sure it is stable and level for the tests to work correctly. (I have not been able to try it my self since I don't have any Crazyflie 2.0 here to test with so I hope it works)

When you have started the Crazyflie 2.0 with the test firmware the console output in the cfclient should tell if it is OK or not.

[Krytical]

Hi

similar issue here, at first the CrazyFlie was stable, then after a few flights, it started flying slightly backwards every time.

Then randomly while flying it flipped and hit the ground, and stopped passing the self test.
I contacted support, they suggested I reflow the chip, which I had my company do for me (I happen to work for a MFG company with reflow and whatnot)

So the reflow was successful, and the CrazyFlie was able to pass the self test and fly again one time.

As soon as the CrazyFlie went upside one time, the CrazyFlie will not fly anymore. I made a video of how it responds now: http://youtu.be/zthq0N1BLrg


Now I have flashed the test firmware and the CrazyFlie will not pass the self test.
here is my console output through 5 restarts:

SYS: Build 6+:29624e7f6e3b (2014.12.0) MODIFIED
SYS: I am 0x313238323133471529003E and I have 1024KB of flash!
IMU: MPU9250 I2C connection [OK].
IMU: AK8963 I2C connection [OK].
IMU: LPS25H I2C connection [OK].
EXPBRD: Found 0 memories.
EEPROM: I2C connection [OK].
MPU6500: Self test acc X [FAILsured: -2147483647.2147483647
OW: Found 0.
SYS: ----------------------------
SYS: Crazyflie is up and running!
SYS: Build 6+:29624e7f6e3b (2014.12.0) MODIFIED
SYS: I am 0x313238323133471529003E and I have 1024KB of flash!
IMU: MPU9250 I2C connection [OK].
IMU: AK8963 I2C connection [OK].
IMU: LPS25H I2C connection [OK].
EXPBRD: Found 0 memories.
EEPROM: I2C connection [OK].
MPU6500: Self test acc X [FAIL]. low: -14.0, high: 14.0, meaSYS: ----------------------------
SYS: Crazyflie is up and running!
SYS: Build 6+:29624e7f6e3b (2014.12.0) MODIFIED
SYS: I am 0x313238323133471529003E and I have 1024KB of flash!
IMU: MPU9250 I2C connection [OK].
IMU: AK8963 I2C connection [OK].
IMU: LPS25H I2C connection [OK].
EXPBRD: Found 0 memories.
EEPROM: I2C connection [OK].
MPU6500: Self test acc X [FAIL]. low: -14.0, high: 14.0, measured: -2147483647.2147483647
OW: Found 0.
SYS: ----------------------------
SYS: Crazyflie is up and running!
SYS: Build 6+:29624e7f6e3b (2014.12.0) MODIFIED
SYS: I am 0x313238323133471529003E and I have 1024KB of flash!
IMU: MPU9250 I2C connection [OK].
IMU: AK8963 I2C connection [OK].
IMU: LPS25H I2C connection [OK].
EXPBRD: Found 0 memories.
EEPROM: I2C connection [OK].
MPU6500: Self test acc X [FAIL]. low: -14.0, high: 14.0, measured: -2147483647.2147483647
OW: Found 0.
SYS: ----------------------------
SYS: Crazyflie is up and running!
SYS: Build 6+:29624e7f6e3b (2014.12.0) MODIFIED
SYS: I am 0x313238323133471529003E and I have 1024KB of flash!
IMU: MPU9250 I2C connection [OK].
IMU: AK8963 I2C connection [OK].
IMU: LPS25H I2C connection [OK].
EXPBRD: Found 0 memories.
EEPROM: I2C connection [OK].
MPU6500: Self test acc X [FAIL]. low: -14.0, high: 14.0, measured: -2147483647.2147483647
OW: Found 0.




Edit
------------------------
Just curious, if the chip is bad, and I was to purchase another MPU6500 from here: http://www.invensense.com/mems/gyro/mpu6500.html
Solder the new one on, and re-flash the CrazyFlie, I'm guessing that would solve the issue? I just really don't want to send this thing back to Seeed in Singapore, and wait for the US postal system again...
last time I had to literally chase the mailman down to avoid having to wait an extra 3 days for my Drone that I had already waited over 2 weeks for...
and that's IF they would even still support it after I chose to have my company reflow instead of send it back to seeed... I could understand the question of heat damage/improper reflow now.
and I just want to fly the awesome drone, the few flights I had have me hooked!