Output capacitor: subjective and objective views?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Administrator
Joined 2007
Paid Member
This is getting very confusing,

I beg to disagree that the power supply caps are NOT in series with the loudspeaker and don't impart their limitations on the overall sound.

That's a bit of a vague statement John if you don't mind me saying so... my definition of "in series with the speaker" is more precise... and I think this is where all the confusion is arising... and it is that the audio signal we hear, the current that moves the speaker is not the same as the current/s that flow in the PSU caps.
If that were the case, disregarding PSU ripple, what would happen with two channels running off one PSU. If the speakers currents really did pass directly through the caps, and that is what we actually listened too, then the two channels would interact.
Obviously that doesn't happen.

The reason for this confusion is the definition of GROUND. You see, ground in an amp or preamp, is really NOT the Earth, but something completely artificial and just called 'ground'. No one can insist that 'ground' is always perfect, in fact, at low and high frequencies, it is usually kind of lousy.

Ground confuses everyone seemingly... it is an arbitrary reference point that we decide upon.
It is the point (for this discussion) that all signals are referred to... and we have to be careful that any points referred here (to ground) really are at the same potential. A cm or so of print or wire that we are calling ground, with say the speaker return current flowing through it and the effect of that ground is compromised severely.

Now, for some reason, people here are talking about using output caps in series with a loudspeaker and using a single power supply to power the amp like the Dyna ST120 and the Quad. Many British designers stayed with it, long past its time, but the rest of us saw the ADVANTAGE of taking the output cap and putting it in the power supply, making a +/- supply. Then the 2 caps are in PARALLEL and do double duty, as the low Z series pass for the loudspeaker, and filtering the power supply ripple. This did not come, overnight, and many of us had to think about it for awhile, but we rarely looked back once we switched over to dual supplies. In either case the power supply caps are still significant as the return current from the loudspeaker flows through them.

That's an interesting take on it... "taking the series output cap (which IS in series with the load) and putting it in the PSU making a -/+ split supply".

When you arrange the PSU thus, the output signal (which is now DC coupled) does not flow "through" the PSU caps in the same way that it did through the single DC blocking cap.

In series with the load, and the amp could never go down to DC... the LF performance was restricted by the output cap.
Move the cap to form a split supply, and now the amp can goes down to DC, the LF performance is perfect down to DC.
(Question ? If the caps are "in series" with the load how can that be ?)

Why so... because the output of the amp should be a copy of the input signal, and these signals are with respect to one point only. Our ground. What leaves that ground on it's way back into the PSU doesn't matter... it could well have the return currents for the other channel, or a dozen other channels... but they don't interact and alter the output current of the channel we are looking at.

Is it the term "in series with" that's confusing here ?
Is my definition of "in series" different to everyone elses :)

So what do the PSU caps do... they do reduce the impedance of the supply.
If the amp was totally immune to PSU variations they would be even less important. All amps to some extent see disturbance on the PSU couple back via whatever means, perhaps through a poorly implemented current source, a poorly decoupled reference... many reasons.
So reducing the impedance of the supply is important in that respect, but that is not the same as saying the load current is in series with the PSU caps.
 
I must correct a misassumption that I made that the Quad 405 used an output cap. I must of been thinking of an earlier circuit from the 1960's. However, why people wish to 'down-peddle' the importance of quality caps, both in the power supply or in series with a loudspeaker, is beyond my understanding.
 
Is it the term "in series with" that's confusing here ?
Is my definition of "in series" different to everyone elses

Yes, since the beginning of this discussion, it is the term "in series" that is confusing.
In my diagram, it is clear that the caps, used as generators, are paralleled to the load.
The load seen by these generators being: speaker in series with the output transistors.
This is why DC can pass.
And, opposite to output caps, they are in the feedback loop.
 
Last edited:
Administrator
Joined 2007
Paid Member
Yes, since the beginning of this discussion, it is the term "in series" that is confusing.
In my diagram, it is clear that the caps, used as generators, are paralleled to the load.

:) I disagree, because if they were "in parallel" then the load would also (by definition of being in parallel) also be passing all the ripple current and other varying currents flowing in those PSU caps.


The load seen by these generators being: speaker in series with the output transistors.

Yes :) which is different to saying the speaker is in parallel with the caps. The return end of the speaker is a reference point. The other end of the speaker (that goes to the amp output) is controlled by the amp under the action of feedback.
The voltage at this point is a copy of the input signal.
By definition therefore the current in the speaker (assuming a resistive load) is also a perfect image of the input signal.
The current in the PSU caps is not.

This is why DC can pass.
And, opposite to output caps, they are in the feedback loop.

Yes... DC can pass as long as the output of the amp is DC coupled and the PSU maintains sufficient voltage... whether it uses caps, batteries, it doesn't matter. They are not in the feedback loop.

:)
 
Administrator
Joined 2007
Paid Member
Hi John,

Mooly, I stand by my opinion. Just measure the current through the caps, rather than the voltage across them.

Which won't be the same as that flowing in the load alone... it is not in series... :) John, why am I saying that to you ? you know what goes on better than most.

I'm not saying the PSU is irrelevant in practice, of course it is, but many here seem to think that the caps really are just "in series" with the speaker, which they are not.

Another proof... the mains is constantly varying, sometimes by several volts as large appliances switch on and pull the voltage down. That sudden rise or fall causes a similar "large" current to flow in the PSU caps as the voltage suddenly increases on the transformer secondary. That constantly varying current in the PSU caps is not reflected by any such increase or decrease in the load current.
 
They are not in the feedback loop.
They are in the meaning the current flowing from them into the load is controlled by the feedback.
In the loop cap + load + transistor, the control by feedback of one among the three devices imposes the current into the others to be the same.
The return end of the speaker is a reference point.
This ,combined to feedback makes ripple rejection

By definition therefore the current in the speaker (assuming a resistive load) is also a perfect image of the input signal.
The current in the PSU caps is not.
The part of tha
 
Disabled Account
Joined 2006
Well guys, we certainly do have a good range of responses.

The opinion seems to be split bewteen:
*You can hear an output capacitor, but only if your system is resolving enough
*You can not hear it if it is big enough and of 'good quality'
*It must be placed in a global feedback loop in order to not hear it

Underlying my original question was the notion of designing a small and simple amp without a global feedback loop, and I think that using an output capacitor would help in these ways...
*Easy design of Single Ended stages
*Everything conveniently 'ground referenced'
*Relatively simple single polarity power supply
*No need for a DC servo
*No need for DC offset protection circuit

In post #44 tiefbassuebertr kindly mentioned some other points.

Do you think there are any other potential advantages?
 
Administrator
Joined 2007
Paid Member
There's absolutely nothing wrong with single ended at all... it has many advantages which you list.

What amp and speakers are you using at the moment ? If it's DC coupled have you tried just adding a non polarised cap in series with the speakers to see if you hear any difference.
 
I draw one half only for simplicity.

The generator is a voltage source .

Seen from the receiver:

The receiver is in parallel with the generator.
Into the loop: generator +transistor + load flows one and only current.

The feedback makes this current such as the voltage across the load is a copy of the input signal. This eliminating the ripple.

The cap isn't in the feedback loop per se but the current flowing through it , which is the same as the current flowing through the load (at least seen from the receiver), is under te NFB control.


There is no problem for generating DC into the receiver, Its amount depending of the transistor conduction.

This is how i see it

My post 87 is by mistake
 

Attachments

  • Gene.JPG
    Gene.JPG
    22.3 KB · Views: 229
Last edited:
I disagree, because if they were "in parallel" then the load would also (by definitio

Indeed, Mooly, I agree. Would not the upper cap be in series during part of the signal swing, then alternately the lower cap, each "taking turns"?

Bigun--how does the phase of the signal through the output cap relate to the phase (either across or "through") the PS caps? I am trying to understand this, and I have always thought that the PSU caps at the very least bypass the signal or a portion thereof to ground.

Just empirically, subjectively speaking, I have been astonished at the improvements an upgrade from a Mallory 4700 uF output cap to 5 paralleled 1000 uF Nichicon KZ made, and have no reason to doubt that improvements might be had if the right caps were inserted into the PSU.
That's just it--I believe there is an optimum part for every position in a circuit, but good luck finding that perfect one.

Terry
 
Instead of questionning if reservoir caps are or are not in series with the output stage and the load, can't we consider the voltage at the power supply caps as coming from a DC voltage source, but with some imperfection as impedance and ripple ?

In an AC coupled output amplifier, including the output cap in the NFB loop, we can get rid of its defaults. An example is this amplifier which has one of the lowest harmonic distorsion ever seen (and very good sound) :

MJR7-Mk3 Mosfet Audio Power Amplifier
 
Instead of questionning if reservoir caps are or are not in series with the output stage and the load, can't we consider the voltage at the power supply caps as coming from a DC voltage source, but with some imperfection as impedance and ripple ?

In an AC coupled output amplifier, including the output cap in the NFB loop, we can get rid of its defaults. An example is this amplifier which has one of the lowest harmonic distorsion ever seen (and very good sound) :

MJR7-Mk3 Mosfet Audio Power Amplifier

What is important is not where the cap is but where is the feedback sensing.
The best place is across the speaker itself.
 
Administrator
Joined 2007
Paid Member
Indeed, Mooly, I agree. Would not the upper cap be in series during part of the signal swing, then alternately the lower cap, each "taking turns"?

The current through the "upper and lower" caps in the PSU varies as the output current swings positive and negative, but neither is a copy of the current in the load... they are the sum of all the ripple currents caused by all the loading on that rail.

Imagine connecting a resistor via transistor used as a switch across one of the rails. We apply say 1khz squarewave to the transistor, so the resistor "pulses" the supply. If we make that resistor low enough we can draw say 1 amp in perfect pulses from the supply. The cap on in the PSU will have a large ripple component corresponding to this 1khz.

It's not audible though, as the speaker isn't in series with those PSU caps, however if the amp has a poor PSRR then it might become audible, but that's because the 1khz ripple is entering the amp via another mechanism.
 
Mooly,
For shure, the total current flowing through the caps is far from being a copy of the signal. Far from a sinus. What you can see if you cut a cap leg to look at the current.
And you can even add your square wave to it, i know that.

I just say that a part of that current flows through the load also. This part is the amplified signal.
This current have to come from the caps as long as there is nothing else overthere which is able to provide current. If it comes from the reservoir caps, it has to flow through them.
Or are we playing whith words?
 
Administrator
Joined 2007
Paid Member
Hi bobo,
Not playing with words :)
What I am trying to get across is that the signal across the speaker is determined by the action of feedback, and the points of reference of that feedback.
What happens to, and what gets added to that current from then on as it returns to the PSU doesn't matter.

If you change the caps in the PSU from the best caps in the world to just ordinary ones, that change does not alter the speaker output. That's what I am trying to get across... everytime someone on here says the caps are "in series" with the load I am trying to show that that isn't really the way it works when the amp is DC coupled on split supplies.

I can't explain it any differently.

:)
 
What I am trying to get across is that the signal across the speaker is determined by the action of feedback, and the points of reference of that feedback. What happens to, and what gets added to that current from then on as it returns to the PSU doesn't matter.

We are together.

If you change the caps in the PSU from the best caps in the world to just ordinary ones, that change does not alter the speaker output. That's what I am trying to get across...


I do not know.
Though, bigger caps give powerfull basses and small bypass better highs...
I prefer long life, High ripple,low ESR/ESL good caps in my PSUs
Just in case.

And i must confess: I love big caps, they look nice in an amp.

Or, do you mean that if a signal can cross an ocean of dirty currents without damage, never mind the cap quality?
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.