PWM and audio noise

[sayzer]

Let this be from me:

Harmonics are generated at the latest ground pin (or say final gnd connection of power supply; the return point). Usually, this is minimized by collecting all gnd points inside the circuit at one final point and then connect this final point to the gnd of power supply through an inductor serially connected in between.

Crucial point is that, power supply gnd must first go through inductor and then be connected to your final gnd point of audio circuit .

Try different values of inductor until you get your best result.

[MikeBZH]

Hi keithv,

I just discover your post right now.

I do not see any writings about the RC integrator circuit which should take place between your CCP1 output and the audio input. This RC network eliminates most your sampling signals.

In my application I use R = 1k and C = 47nF for a sampling rate of 56 kHz and this works fine.
You should modify the values accordingly, for example C should be about 120 nF for a sampling rate of 20 kHz.

Best regards
MikeBZH

[kenif]

What everyone else says, plus:
Declare a ground star point and make all grounds radiate from that point - no daisy chains.
You should be able to trace ground noise origin with a scope.
Electrolytic caps are the bastard love child of a car battery and an RF antenna, replace with tant or ceramic if you can.
Also, make sure all caps are low ESR/ESL like TY.

[keithv]

Thanks for all the suggestions. I've been sidetracked by another project, but I will revisit this and give them a try and let you know how it works out.