Bosch IMU drift

[max246]

In the bosch datasheet is said that the calibration do count for hard and soft. My calibration goes to 1 after 5 minutes of driving, so it is not the car as I calibrate the sensor with the car below.

I think there is something else going on...

[adafruit_support_bill]

Does it regain calibration after stopping? There is a post on the Parallax forum to that effect:

http://forums.parallax.com/discussion/1 ... -for-quads

Even this limited sample should be good for orienting cameras(or Kinects, which is what I am doing next) and for land or sea bound robots. One thing that interests me is that the unit tends to lose calibration during certain kinds of motion.
What I find amazing is that it regains calibration and after returning to a stationary position, no drift seems to have occurred.
I don't know how they do it... and I could be wrong... but it looks really good so far.

This thread over at OpenROV seems to point the finger at it being an electrical problem: https://forum.openrov.com/t/how-do-you- ... imu/4006/8

I coach a few robotics teams and they use these IMUs for robot navigation. They have found that it is necessary to locate the IMU as far away as possible from any electrical motors, power distribution or any wiring carrying substantial current.

[max246]

I am putting the sensor on the roof as it the far point of any electrical motors or wires, I am having an IMU with Arduino nano to send over via i2c to my PC.

There are not electrical problems, I just have 4 cables and thats it. Is your board not grounded properly ?

The calibration returns to normal if I remove the box and rotate to get more magnetic info, but this is a bit pointless as the sensor should just stay calibrated all time after done once.

[adafruit_support_bill]

There are not electrical problems, I just have 4 cables and thats it. Is your board not grounded properly ?

The board has a solid ground plane. How well is it grounded to the rest of your electronics? How long is the cable? Is it shielded? How clean is the supply voltage? Automotive electrical systems are notoriously noisy and hostile environments for electronics of every kind.

[max246]

It is about 30 cm cables in a sealed box in plastic, not shielded. I am was running the setup via USB of my laptop without charger but I did have a Raspberry Pi powering the whole system and still had the same issue.

Do you think shielded cable will improve something? They are secured so not sure it is a problem and I dont have spikes on my system.

[adafruit_support_bill]

Shielded and/or twisted pair wiring can help minimize noise pickup in the wiring. It is more critical with longer runs - especially if the cable is routed near potential noise sources. But 30cm within your enclosure is probably not an issue.

If you are powering via the Pi or your laptop, the supply voltage should be fairly clean.

Automotive applications can be tough for electronics. But it seems like you are doing the right things. If you can't get it to function correctly in your project, please contact [email protected] with a link to this thread and we can arrange a refund.

[PaulRowntree]

Because of the problems otehrs have had with the BNO055, I took mine out for a 30 minute drive this morning. It was connected to a Propeller controller, bit-banged I2C, DIY clock stretching. 1 sample per second, on front seat of car, in cardboard box to reduce motion, connected to a Surface that powered the boards by USB.

1) the Magnetic Calibration 'quality' did dip down to 2, then 1 when driving, but there was no sign of corrupt data.
2) speeds went up to 80 kph, no sign that faster caused more problems
3) most often the only way to get the Mag Cal 'quality' to get back to 3 was to rotate the board a bit
4) the system status showed the most variation, often down to 0, no observed consequences.
5) Acc and Gyro quality values were rock steady at 3

Next test will be with a GPS attached, higher data rate, and the ability to re-install the calibration factors on the fly to see if it changes things.

So to sum up, behaviour was odd, not what I expected, but not a crash-and-burn.
Cheers!

[max246]

Paul, give a try to do a g turn few times or very high acceleration on the side. This made the sensor to offset and drift the heading.

Are you using in 9dof mode? I think all other values a correct but for the heading is a bit tricky.

I have sent an email to Bosch in case they know about this issue. I don't need a refund, I would like to know more about this issue and if it can addressed in somehow.

[adafruit_support_bill]

Please let us know what Bosch has to say on the subject.

[max246]

Never got an answer via email, do you have a more direct contact to ask them?

[adafruit_support_bill]

Unfortunately, no. We don't purchase direct from Bosch, we buy through distributors.