Power Supply Resevoir Size

[gootee]

Yeah, fast or slow couldn't refer to the speed of the electricity itself, but, rather, probably to the speed with which the signal amplitude can change, at the output. And that can vary fairly widely. Or it might refer to the frequency response's upper limit, which would be related to the former.

[sextaafondo]

Originally Posted by gootee

Yeah, fast or slow couldn't refer to the speed of the electricity itself, but, rather, probably to the speed with which the signal amplitude can change, at the output. And that can vary fairly widely. Or it might refer to the frequency response's upper limit, which would be related to the former.

Tom, based on this fact, and the one that many people recommend adding large value (1uF-1000uF) bypass capacitors right at the power pins of the amplifier boards or if possible at the very side of the output devices is that I can't clear out some big doubts.
Some people say that if you install a fare amount of bypass capacitance as close as possible to the output devices, transients will be better handled as the requiered power for the transient will be delivered FIRST by the bypass capacitors until the large filter ones start to deliver their own charge. This is mostly the reason why bypass capacitors are adviced to be installed.
Well, I must admit this sounds WAY nonsense to me . Even without having made any specific test to support my thinking.
Some people also say -AFAIU- that they hook up the bypass capacitors through low value resistors (0.15 ohm or so). I can't get why someone would want to manually increase the ESR value

My understanding is that being both the large filter capacitors and the bypass ones at the power pins interconnected through heavy gauge wires, this provide a very low resistive path between them.
Then, how in this world would the bypass capacitors give out their charge FIRST being tied up to such big tanks like the filter capacitors are??
Voltage charge works in this case, and in many others as well, as liquid fluids.

I would agree with this to be true if there was a high resistive path between both -main bank and bypass-, in which case, once the bypass capacitors had delivered their charge they wouldn't be able to recharge any faster than that resistive path allows.

I would also agree the theory might be true if there was a fair amount of inductance in the line between the main filter bank and the bypass bank, and that to be true only at a certain frequency range close to resonant, which might also be outside the audible range of the PA frequency response.

Having said all of this, I'm close to being certain that high value bypass capacitance is, in any case, not fulfilling the planned task . Of course I might be missing something important and maybe all I said is wrong so please, anybody and everybody, feel free to point out any error or missing fact.

One big question would be, if PSU is a decent one - even not an outstanding one, just a good average PS with sufficient power for the task, wouldn't the audio quality be just dependent on the PA design? and of course of the passive crossovers and loudspeakers?
It's been recorded pages back on this thread that great sounding amplifiers with LOTS of deep bass have what looks like powerfull (BIG transformer) but of mediocre design PSU.

I repeat, I'm no EE.

[fas42]

A simple way of looking at it, is that the design of the PA is usually based on the concept that the voltage rails are "perfect". Spice simulations, back of envelope guesstimates, start with ideal DC voltages being "just there". Well, when you do real simulations, and real measurements, they tell you something totally different, the power supplies are varying inputs to the PA circuit in their own right.

Once you dispense with the concept that the power supply is one black box, and the PA circuit is another, that just happen to occupy a particular casing, then everything changes. The amplifier is one circuit, that happens for convenience of thinking to be separated into 2 sub-parts. The electrons don't know which sub-part they're in, they just behave as the electrical forces influencing them dictate they should ...

Frank

[sofaspud]

...many people recommend adding large value (1uF-1000uF) bypass capacitors right at the power pins of the amplifier boards or if possible at the very side of the output devices...

Tom has differentiated (correctly IMO) bypass and decouple within this thread. A 1μF bypass cap should shunt transients around the device, while a 1000μF decoupling cap should provide energy storage right where it's needed, to prevent affecting the entire DC supply.

Some people also say -AFAIU- that they hook up the bypass capacitors through low value resistors (0.15 ohm or so). I can't get why someone would want to manually increase the ESR value...<snip>...I would also agree the theory might be true if there was a fair amount of inductance in the line between the main filter bank and the bypass bank, and that to be true only at a certain frequency range close to resonant, which might also be outside the audible range of the PA frequency response.

Inductance not just in the line between the main filter bank and the bypass bank, but within the banks, too. Which puts an inductance in parallel with a capacitance. The "ESR boost" can squash the resonance, at least at critical frequencies.

[AndrewT]

Originally Posted by sextaafondo

.........................................I would agree with this to be true if there was a high resistive path between both -main bank and bypass-,.................................................
I would also agree the theory might be true if there was a fair amount of inductance in the line between the main filter bank and the bypass bank, ........................

you said "high resistance", where you should have used "high impedance".

Then everything falls into place.

[gootee]

Sex,

Maybe it will become clear if you first imagine a chipamp power pin with NO local capacitance:

For a given conductor size and shape, inductance increases with length. So the farther away the PSU is, the longer it would take for a needed fast-changing current amplitude to arrive (and the larger the rail-voltage dip would be). So without ANY decoupling or bypass caps at the chipamp, the speed of the transient response, and the upper limit of the overall frequency response, will depend on the distance to the PSU.

Also, you will just have to accept the fact that the response speed of a local cap will not be lowered by having a large capacitance at the PSU.

Also, keep in mind that we need the frequency response and the speed of the transient response to be WAY higher than the audible range, because the amplifier will always be trying to cancel high-frequency harmonics.

It should be clear, now, that SOME amount of decoupling capacitance is beneficial. So now it's just a matter of figuring out how much, based on the needed or desired upper end value of the frequency response.

The benefits of adding even more C than is actually required, right at the amp, will usually quickly diminish, since the rails' and capacitors' inductance will cause less of a problem for lower frequencies. But more C close to the amp would never hurt anything.

[gootee]

Sorry, I understated the benefits of using as much decoupling capacitance as possible, right at each output device, and actually omitted their main purpose:

While the high-frequency response can impose requirements for a minimum value and a maximum connection length from the point of load, for the decoupling capacitance, and while avoiding high-frequency and transient response distortions might not provide a compelling reason to increase the decoupling capacitance value by too much, past a certain point, I forgot to mention, in my last post, that "decoupling" alone is a very good reason to increase the decoupling capacitance. The main benefit of more local decoupling capacitance is lower-amplitude disturbances of the voltages rails, everywhere else in the system.

I think that in most cases, for audio power amplifiers at least, there will need to be three sets of capacitors near each output device:
1. Bypass - a small cap right across power/gnd pins, to prevent instability
2. Decoupling for transient and HF responses - inductance/connection-length is a concern; might often be under 500 uF
3. Decoupling in general - board-level or sub-system-level or device level; typically has at least as much capacitance as is needed in order to meet ripple-voltage spec for worst-case sustained low-frequency (or DC) output signal.

[sextaafondo]

Thanks sofaspud and AndrewT for the clarifications
Tom, special thanks to you for taking your time to elaborate an explanation about decoupling capacitance I feel this topic makes far more sense to me now, and feel I learned something important.

I'm at a fast pace with a PA project using this boards:

LME49830 K1530 J201 300W Mono Power Amplfier Board | eBay

I also already have both channels massive toroids and many other parts. When ready to start I'll open a thread in the correct forum section.

Still need a nice enclosure; and price rises exponentially with beauty like in many areas of life..

[gootee]

You are very welcome. Thanks.

While not many details of the traces can be seen, that board's layout doesn't look too bad!

[OnAudio]

I have to say again thank you to all contributors to this thread. A great resource that explores issues missed out in texts

kind regards,

Harrison.

Aside: Nico did you strike oil or gas ? You went silent about your PS