Brutal unsteadyness

Post here to get support
Post Reply
Mitteau
Expert
Posts: 248
Joined: Fri Jan 29, 2016 6:00 am

Brutal unsteadyness

Post by Mitteau »

Hi

My new cf2 has suffered numerous shocks due to hard landing on a wooden parquet. Now I fear bad electrical contacts. This is the case for the M4 Led, communication led.
But worse is the behaviour of the machine. I have brutal unsteadiness as can be seen in the following video:
https://lud9732.phpnet.org/REPOS/SAM_3257.mp4
Two other symptoms are the the instability of the flight data on the client which lasts dozens of seconds:
https://lud9732.phpnet.org/REPOS/SAM_3256.mp4
Also an error "Fatal python error" due to persistent threads:
SAM_3229_r.JPG
I wrote down the console message for two different runs:
§§§§

Run 1

SYS: ----------------------------
SYS: Crazyflie 2.0 is up and running!
SYS: Build 5:dc3de5771fa2 (2017.06-5) CLEAN
SYS: I am 0x3131393135355116350037 and I have 1024KB of flash!
CFGBLK: v1, verification [OK]
DECK_DRIVERS: Found 10 drivers
DECK_INFO: Found 1 deck memory.
DECK_CORE: 1 deck enumerated
MPU9250 I2C connection [OK].
AK8963 I2C connection [OK].
LPS25H I2C connection [OK].
ESTIMATOR: Using estimator 1
EEPROM: I2C connection [OK].
MPU6500: Self test gyro X [FAIL]. low: -14.0, high: 14.0, measured: 38.70
AK8963: Self test [OK].
DECK_CORE: Deck 0 test [OK].
SYS: Free heap: 18544 bytes

Run 2

SYS: ----------------------------
SYS: Crazyflie 2.0 is up and running!
SYS: Build 5:dc3de5771fa2 (2017.06-5) CLEAN
SYS: I am 0x3131393135355116350037 and I have 1024KB of flash!
CFGBLK: v1, verification [OK]
DECK_DRIVERS: Found 10 drivers
DECK_INFO: Found 1 deck memory.
DECK_CORE: 1 deck enumerated
MPU9250 I2C connection [OK].
AK8963 I2C connection [OK].
LPS25H I2C connection [OK].
ESTIMATOR: Using estimator 1
EEPROM: I2C connection [OK].
AK8963: Self test [OK].
DECK_CORE: Deck 0 test [OK].
SYS: Free heap: 18544 bytes

§§§

X-gyro seems weak. In both cases, calibration process, as seen by the slow red Led M1, also lasted dozens of seconds. The deck is the z-ranger sensor.

Cfclient is the last commit 47868d, lib is e24044. Firmware is the 2017.06-5.
System is xubuntu 17.04. Joypad is xbox360 (but I think this has no influence). I had correct flights in the past with my PC/system.

Motors are new ones and there is no hairs in the axis.

My hypotheses are:
1/ damaged sensor (40%
2/ electrical failure (40%)
3/ too many packets between cf and client (20%)

Did you meet such behaviour in your cf tests?

I met the same behaviour: my unsolved post viewtopic.php?f=5&t=2036#p10334. This is ecxactly the same thing.
tobias
Bitcraze
Posts: 2339
Joined: Mon Jan 28, 2013 7:17 pm
Location: Sweden

Re: Brutal unsteadyness

Post by tobias »

I think you analysis is very resonable. What about you send the board to us and we can repair it for you. You also had another board we can have a look at right? Send us an email to contact (at) bitcraze.io and we will help you out.
Mitteau
Expert
Posts: 248
Joined: Fri Jan 29, 2016 6:00 am

Re: Brutal unsteadyness

Post by Mitteau »

Hi, thanks a lot, I am writing you an email.
Jean-Claude
Mitteau
Expert
Posts: 248
Joined: Fri Jan 29, 2016 6:00 am

Re: Brutal unsteadiness

Post by Mitteau »

Hi
To add to my analysis of the behaviour of my cf2, I wrote down the graph of Gravity vector during an unsteady flight :
SAM_3259_r.JPG
After the lost of balance, the system begins a steady large oscillation...
At the exit of the client, there are Python error messages:
SAM_3229_r.JPG
Jean-Claude
Mitteau
Expert
Posts: 248
Joined: Fri Jan 29, 2016 6:00 am

Re: Brutal unsteadiness

Post by Mitteau »

Hi
My feeling is that my problem is not in the calibration process but in sensor fusion.
As you can see in:
SAM_3265_r.JPG
giving the gravity of sensorfusion6. I just shake briefly cf2 between 340 and 350 seconds. Then fusion enter in a hard oscillation process lasting dozens.
I had the same oscillatory behaviour after a unsteady flight.
Perhaps a wrong parameter value in my parameter TOC?
Jean-Claude
tobias
Bitcraze
Posts: 2339
Joined: Mon Jan 28, 2013 7:17 pm
Location: Sweden

Re: Brutal unsteadyness

Post by tobias »

A very strange behavior and my best guess it is because the senor is damaged. Did you get our email?
Mitteau
Expert
Posts: 248
Joined: Fri Jan 29, 2016 6:00 am

Re: Brutal unsteadyness

Post by Mitteau »

Hi
Yes, and I answered using "contact" page. Did you receive it ?
Modif:
I received the last one mail, but it was in my junk mail.
I will send you the 3 boards with small notes. Is it correct if I send only the boards without motors and their mounts?
Can I buy a board alone to go on flying?
Jean-Claude
Mitteau
Expert
Posts: 248
Joined: Fri Jan 29, 2016 6:00 am

Re: Brutal unsteadiness

Post by Mitteau »

Hi
More information:
After a shock with my finger, IMU sensors are immediately stable, but sometimes with non zero values although cf2 is perfectly fixed and horizontal.
If shaking 2 or 3 times, the gravity vector enters quasi steady oscillation with damping half-period around 1 minute. After 2.5 minutes there are still visible oscillations.
Jean-Claude
Post Reply