Audio power decoupling

[trevwhite]

I am working on a Class D audio design. I am looking at a TI evaluation board schematic and it has a decoupling capacitor and a series resistor and capacitor coupling connected to it. I wondered why it has both?

I attached and image of the circuitry.

[diyaudio]

The circuit you see is tuned, based on the calculated PCB track length + Drain/Source Mosfet Leg (which is an inductor) not shown, this is basically how you damp oscillations its a very important concept in class-d amplifiers/power electronics.


Try this... take the same circuit and place an inductor in series with it, use the value 1nH per 1mm trace length, and test the results in your simulator (now,  change the value of L provide random lengths). you will be very surprised! (see how the resonance shifts)

This is why sticking random values like 10n, 100n ect.. is a bad idea, most of the time it just excites the oscillations at the resonating point, resulting in more oscillations!

I'm lazy and would have given you a ltspice simulation but you get the idea. 

[MagicSmoker]

This is called "shunt damping" of supply line oscillation. Ideally, the resistor needs to be equal to the characteristic impedance of the resonant LC network, while the capacitor is just there to block DC through the resistor (which is to say, damping could be achieved by simply wiring a resistor across the supply rails). Usually the DC blocking capacitor is 4x to 10x larger than the rest of the capacitance across the supply rails so as to not become part of the problem itself. I don't agree with @diyaudio that the damping resistor value is terribly critical - almost any resistance is better than none at all - but it is fairly obvious from the schematic the OP attached that not much thought was given to the damping network besides there needing to be one...

Also note that much the same effect can be achieved simply by making some of the decoupling capacitors small electrolytics: their typically high ESR can act as a damping resistance all by itself. Conversely, the ultra low ESR of MLCCs tends to exacerbate supply line oscillations, and as MLCCs take over more and more jobs formerly only possible with elkos I expect we'll see more issues like this.

[diyaudio]

I don't agree with @diyaudio that the damping resistor value is terribly critical.

 where did you see I make reference to this resistor argument? Too much Magic Smoke ...

[MagicSmoker]

I don't agree with @diyaudio that the damping resistor value is terribly critical.

 where did you see I make reference to this resistor argument? Too much Magic Smoke ...

You're right in that you did not make reference to the resistor, which is a shame because that is the relevant component with respect to damping... So, you really didn't do yourself any favors with your snarky retort.

[diyaudio]

I don't agree with @diyaudio that the damping resistor value is terribly critical.

 where did you see I make reference to this resistor argument? Too much Magic Smoke ...

You're right in that you did not make reference to the resistor, which is a shame because that is the relevant component with respect to damping... So, you really didn't do yourself any favors with your snarky retort.

Trolling Rubbish as usual.

[trevwhite]

I appreciate the time and effort to respond guys, thank you.

I am thinking that they threw a few things at this circuit because in another part of the schematic there is a 100uF electrolytic, 10nF ceramic and 100nF ceramic across the same rails. I have not got a simulation but what you have given me in insight into what to look for if I start to see oscillations on my circuitry.

Hopefully, if I prototype this circuit the 100uF electrolytic will already be taking care of things but if not I shall be able to investigate things further.

[calexanian]

That is the standard Zobel Network used in most audio circuits. Its been a fixture since the intro of the complementary solid state amplifier circuit was introduced.

http://www.wavecor.com/html/zobel_networks.html

[MagicSmoker]

@trevwhite - a 100uF elko somewhere on the board should provide sufficient damping. If you want to check for sure just probe the power supply rails to the Class D controller IC with your scope set to AC coupling. Which reminds me... Class D has no intrinsic ability to reject noise on the power supply rails*, so a clean and well-regulated supply is a lot more important than for, say, a Class AB (or, especially, a Class A) amplifier.

@calexanian - the component values are, indeed, similar to what is commonly used for a Zobel Network, but in this case the RC is across the 12V supply rail and ground.


* - a feedback loop around the Class D modulator - which is not always present or used, btw - will improve the PSRR immensely, of course, just as it would with any other kind of amplifier.

[calexanian]

Oh I did not notice. Pesky predilection to see a familiar pattern and switch to auto pilot brain.