Self-balancing robot version 3 update!

It works! Now it balances on hard or soft floors for more than 30 minutes!

What changed?

Firstly, 2xAA rechargeable Eneloop batteries are insufficient to power those FS90R motors. The specification of the motor allows up to 6V. Two cells provide only around 2.7V total. I switch to 4xAA Eneloop batteries, providing around 5.4V. The motors were much much faster, more than twice as fast. The immediate consequence of this is that the PID parameters need to be re-tuned and scaled down to 10-20% of the values for 2xAA batteries.

Secondly, I turned up the builtin digital low pass filter in the MPU-9250 aggressively. Now it is set to 5 for both accelerometer and gyroscope. That translates to a sensor delay of 17ms and the 3dB bandwidth of about 10Hz. That is probably averaging around 17 samples at 1kHz sample rate. And that did the trick!

Here's a graph of motor power, gyroscope angular velocity and tilt angle.

The filtered angular velocity seldom exceeds 0.3 degrees per second. The filtered tilt angle seldom exceeds 1 degree. The motor power is mostly less than 10% of maximum power.

I've updated the InvenSense MPU-9255/MPU-9250/MPU-6500 module. You can now specify the low pass filter level for the accelerometer and gyroscope in the call to InvMPU.reset_mpu(). Copy this new version into your custom.ts.

The only changes from the version 1 code are the PID tuning parameters and the call InvMPU.reset_mpu(5, 5).

See the previous post on the parts for building version 3.