Would this work (class d power supply)

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.

I am currently making a 2.1 system that is self-powered. I have built a previous system with class AB amplifiers, which have awesome power supply ripple rejection, however class D is another story. Right now I have a 200VA transformer full-bridge rectified to about 10uF of capacitance and I'm still getting about 60Hz hum at the speaker output. I know this is because of the switching properties of class d.

My question would be, if I take the v+ (the amp is single-supply tpa3118d2) going to the amp-chip and run about 4-7 lm317's in parallel to the chip from my ps and drop a little over the minimum voltage for the lm317 would my noise problem be solved a little?

Would also adding a bank of caps on the other side of the lm317 help too?

I would suggest a MUCH higher capacitance.

What is the average current drawn from the power supply?

Since i = C dv/dt for an ideal capacitor and you can say that dt is approximately 1 / 2•fmains, you can pick what dv you want to tolerate and solve for an approximation of C:

C >= i / (2 dv fmains)

For example, if you pull 6 amps and you want 0.5 volt of 120 Hz voltage dip ("ripple") after the capacitance, then

C >= 6 / (2•0.5•60)

C >= 0.1 Farad

C >= 100000 uF

You don't yet know what dv value would solve your problem but I would start with 10000 uF (or whatever you have at hand) and go up or down from there, to see what works.

The regulators would not work with a too-small capacitance, because your ripple voltage amplitude would be so large that the input voltage for the regulator(s) would drop below the desired output voltage 120 times a second. Regulator input voltage must always stay above the desired output voltage PLUS the dropout voltage spec from the regulator datasheet, or the output gets very ugly.
Last edited:
Here I am months and months later, having figured out all the problems on my PCB.

Firstly, I mis-typed my question at the beginning, I had 10mF of capacitance on the rails, which I later increased to ~25mF.

Turns out the hum was a bad value I had chosen for one of my lm317 active filter power supply caps.

I do thank you for your replies though.
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.