I have a project where I need to drive a 24V 30A DC motor following a minute long sequence triggered a kind of remote control. I have 4-5 different sequences, for example 100% forward for 1s, 40% forward for 20s, 100% backward for 500ms, stop.

I've used bought a 720W DC power supply, an Arduino Leonardo, a BTS7960 based motor driver, a TLP281-based optocoupler board and a Wii Nunchuk. I use the Arduino PWM output to control the motor driver, and the Nunchuk on the I²C bus to select a sequence (with the analog stick) and trigger it (with a button). I have a big cap (4700µF) between the PSU and the motor driver, nothing between the motor driver and the motor. I also have relatively strong 2.2kOhm pullups on the I²C.

I've had several iterations before that, with a 12V power supply and a Monster Moto Shield, which worked for a few minutes in the middle of winter when the motor was cold, but the load is too heavy and I need the full 24V. However since I switched to 24V, I have disconnections on the I²C bus when the PWM goes beyond 15-20% duty cycle, and I need to power cycle everything to reset the problem. I borrowed a scope to have a look and when the motor is running there is a lot of noise on the I²C line.

So first, is that normal or did I do something wrong? Is there something I can do to reduce emitted noise from the power side? Is there a more robust alternative to I²C I could use?

 

Are you able ... can you ...Post a block diagram or sketch showing the component power, ground, and major interconnections?

 

when the motor is running there is a lot of noise on the I²C line.

Try beefing up noise supression at the motor itself and at its connecting cables.

 

Are you able ... can you ...Post a block diagram or sketch showing the component power, ground, and major interconnections?

Yes, of course. Can you recommend an easy to use online circuit drawing tool? Otherwise I'll do it in Paint, but it's gonna be ugly.

 

Well, here it is in Paint.net:

 

...just some things to check out:

A common ground point sometimes eliminates problems. Voltage differentials can develop between ground terminals.

So, identify the DC ground on each component and run a 'dedicated ground wire' to a central node ... maybe something conductive.

If that doesn't fix it, you might have to get a scope and trace the signals from point to point ... looking for something that doesn't look right.

Are you sure that you have the Nun Chuck controller wired correctly? ... had to wire up a track ball to a joy stick controller, on one occasion, and it turned out that the joy stick was controlled by changes in current, not voltage. The joystick had potentiometers, but they were wired as rheostats ... and I didn't catch that.

 

One other suggestion ... The 5v regulator chip normally has 1 uF non-polar capacitors placed across the input and output ... power to ground. Doing this suppresses high frequency noise. So add the caps and see if it makes a difference. Also, add a 100 uF electrolytic across the 5v regulator output ... gives it more flexibility
Found this ... for 7805 regulator:
This regulator does not require capacitors for stability, but we recommend at least 10uF electrolytic capacitors on both input and outp

Edit: try this first, since it could be the main problem.