MSGEQ7 audio analyzer and servo motor noise + how to use voltage regulators


I have a project where i want to move servos to music. I am using a msgeq7 chip, and 7 servo motors.
I am working right now on making a pcb for this project, however, the servo motors generate a lot of noise so the msgeq7 chip goes crazy! So i am now working on decoupling the servos and the audio chip, and adding filtering, however im am completely new at this and would like to ask for some guidance.

I will have the audio chip as close to the audio input as possible, and will have it on the other side of the board from the servo motors.

I am powering the whole thing from a 9V power supply, which i then split to power the arduino itself, and the servo motors. However, i need to use a voltage regulator to power the servos. I am using seven sg90 servo motors, and i have been told that i should use at least two voltage regulators for them (one of them powers 4, and the other powers 3 servo motors) since going from 9v to 5v will generate a lot of heat, since the servos require a lot of power. Could anyone comment on this? Do i use the two voltage regulators and have a large cap before each, and a smaller cap after each regulator? I have found this guide: De-coupling that suggests to use a small choke. Should i use this choke? If so, should i use just one after each voltage regulator (if i even need to use several voltage regulators(?)) or should i use one before each servo motor?
So far im thinking to use a small cap (does the value of the cap matter?) next to each servo motor (as close as possible). and a choke after each of the regulators, as well as a large cap before, and a smaller after the regulator (so a standard way of connecting a regulator).
Also, if i use the choke, where exactly do i put it? Right after the small cap that goes after the voltage regulator?

And for the audio chip circuit? Do i need to add any extra filtering there? Is it ok if i power the audio chip from the arduino 5v pin?

Thanks in advance to anyone who replies.

This is one of those times when one really, really needs an oscilloscope.

Here is the source of the problem:

Rather than decide on a single solution, my gut feeling is that you will require a few tries. You already have a handle on the heat generated by the voltage regulators due to voltage-drop x the motors current. The choke will likely help, the value however is somewhat critical to minimize noise and not disrupt the power phasing to the motors (inductors shift the phase on the AC component.)

Rather than trying to answer all of the questions at once, I would prototype an entire 4 servo system and experiment a bit with different bulk capacitors and the choke. A few toroid cores from old home home CFL lamps work for winding your own choke.

If you do not have an oscilloscope perhaps you have a friend with one? This piece of equipment is very valuable for these kinds of experiments.

The analog side of the project needs special attention, too. You can Google for some good ideas, but one I can offer here is to be careful about ground-loops in the analog shielding.

Many manufacturers/vendors have white papers and published best-practices for noise elimination. For example:

This PDF gives some good concepts: There is a part #2, just change the _1 to _2 in the URL.


I see.

How about the voltage regulator part? Should i just use a few regulators or use something like this: LM317 High Current Voltage Regulator | ?

Should i just use a few regulators or

Every stepper/servo motor will cause noise. The generalized basic approach is a large 'bulk' capacitor on the raw (input) DC, a smaller 'bulk' capacitor on the regulated (output) DC, and then things like high-frequency bypass caps, chokes, and so forth.

Distributing the servo load across multiple regulators will distribute the power dissipation per linear regulator and will allow you to distribute the bulk capacitance but this is not the whole story. Generally, smaller capacitors (electrolytic) will have a smaller ESR and this helps in the overall regulation of peak currents, but it also generates higher currents in the wiring to the capacitor.
Ref: Equivalent series resistance - Wikipedia

Distributing also creates a lots of board area usage.

I think that splitting the motors across 2 regulators is reasonable due to power dissipation. I would however, power the msgeq7 from a DC to DC supply completely separate from anything else. This should give you ideal power isolation. Also, I would try and keep the analog area completely separate from the digital side of the board and with a nice ground-plane (copper pour) around this section.

I do not have any specific product recommendations, however. I have gotten some good, inexpensive DC-DC from off-shore, but I have gotten some crap, too. The value of design, prototyping, and modeling is an area that cannot be overstated when working with interference. There is simply too many "gotchas" to not try different scenarios and pick one that works within your parameters.