MPU6050 going crazy for short period

By mlaverne , 13 March 2013
Forums
SmartMotion® MEMS Motion Sensors

MPU6050
MotionDriver v5.1
5 Hz sampling rate
100 kHz I2C

I'm working on a logging device that takes and logs samples from an MPU6050. During testing, the MPU6050 is completely stationary, no movement whatsoever.

About once every 15 minutes (but not *every* 15 minutes), the MPU6050 goes crazy for *exactly* 3 samples:

Pitch,Roll
...
23,46
23,46
170,5 *
170,5 *
170,5 *
23,46
23,46
...

My I2C analyser does not show any corrupted transmission during this time.

Anybody any idea what could possibly cause the MPU6050 to return wrong data for exactly 3 samples once every 15 minutes?

guy_mcilroy

12 years 12 months ago

We haven't experienced this issue. Is there any additional detail you can provide from your testing configuration? i.e. whether using the provided hex file, rebuilt, changes made, what host the library is operating on?

phpbb Post ID
22653

mlaverne

12 years 11 months ago

All the info you requested was already in my original post:

MPU6050
MotionDriver v5.1
5 Hz sampling rate
100 kHz I2C

No changes to MotionDriver, host is in my opinion not important, it's plain and simple I2C.

Anyway, host is a Freescale iMX25 (ARM926) running Windows CE 6.0 R3 using a BSP from GuruCE.com.

The I2C traffic capture just before, during, and just after the faulty data looks fine.

phpbb Post ID
22656

mlaverne

12 years 11 months ago

Here (https://www.dropbox.com/sh/ze1ryxxvt9i4150/cxAKrbi7AI) you can find some I2C traces that show the MPU suddenly starts returning wrong quaternion data. This happens for a couple of samples (usually spanning less then 2 seconds) before it returns to normal, correct values.

There are no I2C errors before, during or after this episode of "bad values".

The files on the dropbox link are in csv format. You can open those in Excel. There's also a text file containing an excel VBA macro that you can use to quickly calculate Pitch and Roll data from the I2C data representing the Quaternions.

This is a showstopper for us, so please respond as soon as possible. If you need more information please contact me by email directly. Response times on this forum are too long...

phpbb Post ID
22659

guy_mcilroy

12 years 11 months ago

Does this anomaly appear only in the euler angle output, or is reflected in the quaternion, or raw data as well? The Euler angle conversion function just converts from the quaternion.

phpbb Post ID
22662

mlaverne

12 years 11 months ago

SectionBest,

Did you look at the files on the dropbox link?

You can see the error is in the RAW data. The excel macro shows the calculation I use for the Roll & Pitch as well.

Please have a look at the I2C captures on the dropbox link.

phpbb Post ID
22665
phpbb Topic ID
15141