Self Test X [FAIL]

[xpeiro]

Hi,

My crazyflie 2.0 when turned on no longer passes self tests. The m1 light flashes red 5 times and repeats. The output of the console tab is the following:

Code: Select all

h!
CFGBLK: v1, verification [OK]
IMU: MPU9250 I2C connection [OK].
IMU: AK8963 I2C connection [OK].
IMU: LPS25H I2C connection [OK].
DECK_DRIVERS: Found 4 drivers
DECK_INFO: Found 0 deck memory.
DECK_CORE: 0 deck enumerated
EEPROM: I2C connection [OK].
I2CDEV: Error callback nr: 256
AK8963: Self test X [FAIL]. low: -200, high: 200, measured: 16253

AK8963: Self test X [FAIL]. low: -200, high: 200, measured: 16253
So I'm assuming some acceleration sensor or something of that nature is not responding correctly. It's being turned on without any movement on a flat surface, as always.

Any help would be appreciated.

[tobias]

It looks like the magnetometer is damaged. It could be caused by a crack in soldering or something with the magnetometer itself. You can force a system start (if you have a crazyradio) by connecting to it with the cfclient. Then open the parameters tab and locate "system->selftestPassed". Set this to "1" and the system will start. As the magnetometer isn't used you should now be able to test fly the Crazyflie. See if it behaves normally and check the log for any "I2CDEV: Error callback nr:". If it works normally the magnetometer is damaged and you have the option to disable and run without it. If you get I2C errors there is probably a crack in the soldering and the sensor MPU9250 needs to be reflowed.

[xpeiro]

Just wanted to post an update for future reference:

I tried the tests tobias suggested and got the I2CDEV: Error callback errors. After looking up what reflowing meant, I decided to try it myself, because what the heck, right? if it doesn't fly I don't mind if I roast it.

I bought a cheap heat gun (http://www.amazon.es/gp/product/B000B9R ... ge_o00_s00), removed struts and motors, covered the board with aluminum foil, cut out a small hole where the mpu9250 was and put it under the gun.

I tried first at 300ºC for a while (a minute or so), without seeing much of a change. Then I cranked it up to the second level 550ºC and after about 30 seconds I stopped. I lifted up the aluminum foil and the plastic pin holder closest to the mpu9250 came off, stuck to the aluminum foil. I thought that was it, I had burnt the whole thing.

To my surprise, once i put everything back together, it worked! no more self test fail, and it flew fine . Battery was low, so I haven't tested it much yet, but it flew for at least 30 or 40 seconds. Previously, it would at most lift a couple of inches and then flip out of control.

So I guess this is a partial success story. It flies, but I melted off one of the plastic pin holders so I guess I'll have a hard time using my led-ring. But hey, I guess the flying is the important part, right ? If I had let it cool for a while before pulling off the foil, it would have probably been better off.

PS: This is my second Crazyflie. The first one flew off over a couple of rooftops never to be found again (I almost fell off my roof trying to find it )

[xpeiro]

Another update:

Well, that didn't last long. It flew fine for 3 or 4 battery charges, now I'm back to where I started, sort of.

The strange thing is I no longer get any self test fails, ever. It always passes all tests. Everything seems to be working fine (moving it around in my hand, the cfclient graphic shows correct orientation and ASL), but when it lifts even a centimeter, the ASL goes berserk and it's impossible to control.

Does anyone know if this is the same issue I had before? Again I've had no self test fails at all, which was the main symptom previously. Could it be that the heat screwed up the barometer?

Any help would be much appreciated. As I said, this is my second crazyflie after the first one "freezed" at almost full throttle in mid air and went away over the rooftops like a british nanny. It would be a shame to lose a second, too.