Accelerometer error

[agi gun]

I was looking for a datasheet that showed the mg/sqrt(Hz) error of the accelerometer in the NavX MXP, but couldn't find one. Does it have one?

[admin]

> I was looking for a datasheet that showed the mg/sqrt(Hz) error of the accelerometer in the NavX MXP, but couldn't find one. Does it have one?

Yes the data sheet for navX-MXP is on the product web site here.  It uses the MPU-9250, the datasheet for that is here.  The spec you're looking for is entitled "Noise Power Spectral Density".

Here's the link to the data sheet for navX2-Micro.  It uses the LSM6DSM IMU, the datasheet for that is here.

And then there's navX2-MXP, which will be available in October.  It uses the ISM330DHCX IMU, the datasheet for that is here.  The spec you're looking for is entitled "Acceleration noise density in high-performance mode".

We will be publishing a summary of all comparative specs for IMUs in October.  Without a doubt, the ISM330DHCX IMU has the best accelerometer noise specs, we've been working on this for 8 years now and it's the one we've chosen for navX2-MXP, with the goal of being the most accurate IMU in educational robotics.

[agi gun]

Thank you! Sorry if this is slightly off-topic but would you happen to have any data regarding fusing data from multiple IMU's? We were actually looking into the ISM330DHCX, and we were wondering if it would be worth it to buy 4-5 of these and either averaging their outputs or running it through something like a kalman filter. 

[admin]

> Sorry if this is slightly off-topic but would you happen to have any data regarding fusing data from multiple IMU's?

I don't have any direct experience with fusing multiple IMU datastreams together, although I am confident the theory is sound.  I'm confident you'll find something on this if you search on the internet.

However I must tell you that I believe the engineering to pull off the high-quality fusion you're discussing will be quite challenging.  What type of circuit design are you planning in order to fuse the data from multiple ISM330DHCX's together?   You'll need something like a very fast microcontroller, interrupt lines from each IMU chip, and firmware that "lines up in time" the data from each sensor before averaging it and then sending it on to the Kalman Filter that calculates the orientation.  And you'll want to run at a fast update rate, because accuracy goes up the faster you can process the data (e.g., less time per sample).

Also, be sure to study up on axial alignment and the calibration necessary to correct it.  If you are assuming that laying multiple of the same IMU on a circuit board will automatically line them up perfectly with each other, you might be surprised to know that the alignment tolerances (in degrees of rotation) when circuit boards are fabricated can allow for some rotation.  There is also axial misalignment that can occur within the chip itself.

[agi gun]

Thank You! We were planning on just buying some pre-installed on breakout boards from adafruit, and then run all timing and fusing code on an arduino, which we think is fast enough. We were planning to have a calibration sequence that corrects for any irregularities between the boards. This is for FRC so we don't know if it would be worth it to make our own pcb, if you had any information on manufacturing our own it would be greatly appreciated, along with any other critiques to this idea. Thank you!