Power Supply Resevoir Size

[AndrewT]

The decoupling must connect the output pin to both power pins.

For HF this route must be a very short route.
This requires the Zobel to be mounted on the output pin and the HF decoupling to be mounted on the power pins. Then the remote ends of all three must be connected together with the shortest possible route lengths. This works for discrete and for chipamps.

Gootee had a good analogy a long while back and he has repeated it a few times for those that were not listening.
Think of the amplifier being fed by a "battery" that consists of only the HF decoupling capacitor.

Imagine all the other supplies to be removed. Now trace out the route that the current flows from our tiny "battery" in through the power pin, out through the output pin, into the HF load (Zobel) and back to the "battery". That is the route that MUST be minimised. Repeat this virtual "battery" for the other polarity capacitor.

Then move on to consider the MF decoupling, but one can now relax the restrictions on "Route Lengths" due to tolerance of slightly higher inductances in the traces and leadouts.

Notice that Main Audio Ground has not entered into any of the MF, nor HF routes !!!!!!!
Main Audio Ground should only see Audio Frequencies, if the correct decoupling has been properly located.

[abraxalito]

Ah - I'd forgotten about the Zobel network, thanks - that makes it more interesting still but at least I now see one aspect of what I was missing before.

Seems to me that the Zobel, being as it is an RC (i.e. an AC coupled resistor) can equally well be to one or other, or even to both of the supplies rather than to ground. The 'both' case being handled by two caps (rather than the usual one) acting as rail splitters and the cold end of R going to the junction of them. ISTM this case will offer the lowest inductance seeing as the current is always sourced from one or other of the rails. Those two rail splitting caps can also do duty as HF decouplers can they not? In this case they've just created a local 'ground' (more precisely an average of the +/- rail voltages) which has no need to be returned to the star earth.

This has all been assuming the current to the speaker doesn't need to be considered. Does it though?

[AndrewT]

I thought that's what is described in post 1551 ?
The local decoupling ground right between the Power Pins and the Zobel connecting the output pin to this local decoupling ground. Isn't that what I said?

But the clever bit is the simplification offered by Gootee to "see" the caps as if they were "batteries" and these being the sole power/current supply.

The final bit is to connect the PCB Power Ground to the Main Audio Ground. Speaker current (Audio Signal) does not pass along this connection.

[abraxalito]

But in my notion, its not actually a ground because it has no electrical connection to ground. Ignore the label, just think of a Zobel network with two caps instead of one. Its actually saving one cap because in your described situation we have 3 caps - 2 decouplers (one to each rail) and 1 in the Zobel.

<edit> Agreed, Gootee's suggested visualization helps a lot.

[AndrewT]

No, you need the Main Audio Ground reference, so that signal ground and power ground and main audio ground all refer their voltages to the same "reference".

[danielwritesbac]

For example 4700u||10u, -0-, 4700u, a 50/50 capdiv also serving as power supply reservoir:
That spot marked -0- (centerpoint of the capdiv) is the only speaker return on my little single rail radios. The speaker return is the only thing connected to the centerpoint of that capdiv. Offset was not a concern because it has an output cap anyway. Works great for a couple of watts. Totally removes treble errata that I was having trouble with previously. I wonder how? Works on several radios, including Zenith Circle of Sound clock radio. It was loud enough before, but holy cow!--it doubled! And, yes, clean treble too. The outcome is the same each time I've tried it.
Questions:
Why did that modification double the speaker output power?
Why did that modification clean up to treble so much?

[DF96]

There is more than one loop in a circuit. Some are made small by adding decouplers in the right place. Others are made small by careful routing of the wires.

[OnAudio]

How practical is the following setup especially for preamp s

[DF96]

Very few people put common-mode chokes around the rectifier. What problem do you think you are solving?

[fas42]

With regard to capacitance changing with frequency, I note on a Rubycon technical document that they acknowledge a loss with increase in frequency, due a combination of the condition of the etched surface, properties of the film as a dielectric, properties of the electrolyte, and the nature of the construction of the capacitor: they show a 10uF, 50V unit being down by about 15% at 20kHz. But only by that much ...

At the higher frequencies the ESR and construction becomes everything for getting effective decoupling, the actual true value of capacitance becomes almost completely meaningless ...

It's easy to know where one end of the decoupling caps should go -- to the power pins -- but where to attach the other end? My take is that it should be the ground point that the feedback network sees, which should be electrically equivalent at all frequencies to the ground of the audio input.

Frank