[Beefy]

Quote:

Originally Posted by Koyaan I. Sqatsi /img/forum/go_quote.gif It depends on what you mean by passing "directly through" the caps.

The same thing I have been saying...... *deep breath*

Quote:

Consider the reservoir cap on the positive side of the power supply in A, the passive ground example. Load current flows from the capacitor's negative plate, through the ground node, and then to the driver's negative terminal, and then on around the loop, where there is an equal amount of load current flowing into the capacitor's negative plate.

So current is being sourced/sunk current directly from/to ground. There is no guarantee that there is current right there from the opposite rail. So localised ground impedance means that the potential of ground will change as current flows.

Quote:

Now let's consider B, the active ground example. Again, let's start with the reservoir cap on the positive side of the power supply. Load current flows from the capacitor's negative plate, through the ground node, just as it does in the passive ground example. The only difference here is that instead of flowing next into the driver's negative terminal, it flows into the positive plate of the reservoir cap on the negative side of the power supply. Of course that current can't flow into that cap's positive plate without an equal amount of current flowing out of it's negative plate. And this current flows through the negative rail, the output stage of the ground amp, finally to the load itself, and ultimately on back to the positive plate of the positive rail's reservoir cap.

So current isn't being sourced/sunk from/to ground. Whatever is required by the positive rail is always right there on the negative rail.

The current just moves from one rail to the other, without ground actually needing to contribute. Whatever goes into ground goes straight back out again, right there locally, with much lower impedance than sourcing/sinking from the whole ground plane.

The signal ground, the reference point for the whole circuit, doesn't see any sourcing or sinking of current because it is always cancelled out locally.

 

[Lifthanger]

Since the σ22's output impedance is much lower than even the best low-ESR large aluminum electrolytic capacitors, having wide bandwidth allows the σ22 to respond to fast changing current demands better than a large capacitor (or a bank of capacitors) ever would.

this is a quote from the sigma22's website.
without a powersupply the current flow is obviously correct. you are also correct about how caps work, that's for sure.

imagine 90% of the current needs would flow from the supply rails. then look at the sigma 22 schematic. simplified the electrons go: transformer -> V- -> circuit -> V+ -> transformer...

the only ground that "has" to be used is the ground of the primary transformer winding.

I'm pretty sure this is close to the truth, because you're schematic is obviously right, but ignores the current that flows into the rails.

 

[Steve Eddy]

Quote:

Originally Posted by Beefy /img/forum/go_quote.gif So current is being sourced/sunk current directly from/to ground. There is no guarantee that there is current right there from the opposite rail. So localised ground impedance means that the potential of ground will change as current flows.

Ground is neither a source nor a sink. It's not any sort of container or reservoir, or some hole that things fall into. It's simply a singular point of reference. It has no potential of its own. Any potential must be between it and some other singular point.

k

 

[Lifthanger]

Quote:

Originally Posted by Koyaan I. Sqatsi /img/forum/go_quote.gif Ground[...] It has no potential of its own. Any potential must be between it and some other singular point. k

could you explain this please? Obv. I'm no expert and this sounds new to me.

 

[Steve Eddy]

Quote:

Originally Posted by Lifthanger /img/forum/go_quote.gif Since the σ22's output impedance is much lower than even the best low-ESR large aluminum electrolytic capacitors, having wide bandwidth allows the σ22 to respond to fast changing current demands better than a large capacitor (or a bank of capacitors) ever would. this is a quote from the sigma22's website.

Output impedance has nothing to do with "speed."

The reservoir caps respond just as "fast" as the σ22.

The advantage of the low output impedance is that it keeps AC signal currents from manifesting as AC voltages across the supply where they may couple into the circuit.

Signal currents will appear as voltage across the supply by Ohm's Law. E = I x R (or I x Z if you want to get technical). If the output impedance is zero, then I x 0 = 0, and therefore E = 0. So ideally, you want the lowest output impedance possible.

Quote:

without a powersupply the current flow is obviously correct. you are also correct about how caps work, that's for sure. imagine 90% of the current needs would flow from the supply rails. then look at the sigma 22 schematic. simplified the electrons go: transformer -> V- -> circuit -> V+ -> transformer... the only ground that "has" to be used is the ground of the primary transformer winding. I'm pretty sure this is close to the truth, because you're schematic is obviously right, but ignores the current that flows into the rails.

I only ignore the current that flows into the rails from the rest of the power supply (i.e. rectifiers, power transformer, etc.). And I ignore it for two reasons.

The first is because as I've said many times now, for the majority of the time, none of that stuff's in the circuit and there is no current flowing into the rails from it. For the majority of the time, the circuit's being powered solely by the reservoir caps.

The second is because it doesn't change anything. The current still flows as it does as illustrated, with the same current flowing through the same ground node. The only difference is that the four red arrows associated with the reservoir caps will be reversed, because during the refresh cycle, they're charging rather than discharging.

k

 

[Beefy]

Quote:

Originally Posted by Koyaan I. Sqatsi /img/forum/go_quote.gif Ground is neither a source nor a sink. It's not any sort of container or reservoir, or some hole that things fall into. It's simply a singular point of reference. It has no potential of its own. Any potential must be between it and some other singular point.

But you HAVE to view it that way. It is being sourced at the caps, and dumped behind the driver, and traveling back past all the reference points in the circuit on the way through. It isn't just some instantaneous propagation of current, no matter how wonderful you think your grounding scheme is.

 

[Steve Eddy]

Quote:

Originally Posted by Lifthanger /img/forum/go_quote.gif could you explain this please? Obv. I'm no expert and this sounds new to me.

In the context here, "potential" ultimately means "potential difference" which ultimately means voltage. The key word is "difference," as you have to have two of something for there to be any sort of difference.

My point was that any "potential" as it relates to ground is meaningless without it being relative to some other point.

k

 

[Lifthanger]

I'm sorry, but at this point I'm also out of here. Had a nice laugh though with the ground potential stuff. sorry for the douchy question though.

I'd like to see one more drawing though:

the B circuit and what would happen at the point the caps are only 90% charged. at this point the voltage at the cap will also be only be 90% of the supply voltage.
now where does the current go?
edit: ok fair enough.

 

[Steve Eddy]

Quote:

Originally Posted by Beefy /img/forum/go_quote.gif But you HAVE to view it that way. It is being sourced at the caps, and dumped behind the driver, and traveling back past all the reference points in the circuit on the way through. It isn't just some instantaneous propagation of current, no matter how wonderful you think your grounding scheme is.

The propagation of current is instantaneous. At least for our purposes here. I mean, the current is propagating at nearly the speed of light. Are you saying that's not instantaneous enough?

k

 

[Beefy]

Quote:

Originally Posted by Koyaan I. Sqatsi /img/forum/go_quote.gif The propagation of current is instantaneous. At least for our purposes here. I mean, the current is propagating at nearly the speed of light. Are you saying that's not instantaneous enough?

To keep the ground absolutely stable? No, from the evidence AMB has presented, I don't think it is.

The return current flows directly through the ground which has its own impedances despite how perfect you think you can make it, past key points that should have a stable reference. They can be isolated from it by using the active ground that diverts it to the opposing rail instead.

 

[Steve Eddy]

Quote:

Originally Posted by Beefy /img/forum/go_quote.gif To keep the ground absolutely stable? No, from the evidence AMB has presented, I don't think it is.

The "stability" of the ground has absolutely nothing to do with the propagation velocity of the current.

Quote:

The return current flows directly through the ground which has its own impedances despite how perfect you think you can make it, past key points that should have a stable reference.

But a singular node has no particular impedance. And the key word here is "reference." If the amp is referencing a singular point, then that point is inherently "stable."

This is the principle behind what's known as "star grounding."

Quote:

They can be isolated from it by using the active ground that diverts it to the opposing rail instead.

But this simply isn't true.

Again, the load current doesn't just go to the opposing rail and then magically disappear like a fart in the wind. As I've illustrated, it goes through the ground node just as it does in the passive ground scheme.

k

 

[jcx]

it may help to draw more detailed expansions of the "gnd node" - 3-ch amps use the active gnd for output only - the signal gnd may be the "passive" center of a dual supply or a lesser current capability supply splitter circuit

the "gnd contamination" comments justifying active/3-ch gnd amps don't really convince me when we add a couple of real world conditions

a good gnd plane can have ~1 mOhm point-to-point R drops between RCA signal gnd entry, feedback R gnd and TRS plug gnd - a really good star gnd implementation will make these connections so that little Vdrop from gnd current returning to the supply occurs across even this small R

the TRS contact R on the common "gnd" can exceed 10 mOhms - a much bigger source of "gnd contamination" crosstalk - not cured by any active gnd scheme discussed here

there seems little justification for worrying about audible effects linear crosstalk of <60dB with music - certainly not with the common open construction of many higher end headphones

nonlinear current is drawn from the PS by Class AB stages - when the return loops through gnd aren't co-located there can be some nonlinear Vdrop in the gnd plane impedance - proper star gnd layout can keep most of this out of the signal and feedback reference gnd path

Class A operation of the output stage removes the possible nonlinear return current - the only "defect" is a very small linear crosstalk term - easily lower than the common inductive coupling of the headphone cable

For the B22 operating Class A the only ”gnd contamination” would be the linear crosstalk type and I believe proper star gnd implementations can keep the linear crosstalk term an order of magnitude below TRS jack/plug interface gnd contact resistance caused linear crosstalk

Until you start wiring SE amps with non-TRS or 4 terminal headphone connections the whole argument for “3-ch” amps with Class A outputs is pointless IMHO

 

[Beefy]

I find your views interesting jcx...... all the work many designers put into lowering distortion by, say, 0.0001%, and you are saying that up to -20dB of stereo crosstalk is not significant?

Usual criticism of this result, fine. Can't somebody with a 3-channel B22 and RMAA functionality please do a comparative test?

 

[jcx]

no, I'm claiming a 20-40 dB improvement on -60 dB linear crosstalk is not audibly meaningful - open cans have higher natural air path crosstalk - coiled headphone cable has higher crosstalk - no live event recording can have less than that coupling between microphones

only totally synthetic/processed studio tracks can have huge R/L channel separation # - even there "stereo image" is painted on by the mastering engineer - nearly universally by pan-potting - mixing the individual instrument stubs in linear ratios to R/L channels

high end audiophile phono carts typically give up at 30-35 dB channel separation specs - and we All Know Vinyl is Superior no?

 

[Beefy]

Quote:

Originally Posted by jcx /img/forum/go_quote.gif no, I'm claiming a 20 dB improvement on -60 dB linear crosstalk is not audibly meaingful - open cans have higher natrual air path cosstalk

Sure, but still...... up to 20dB improvement. Something seriously funky has to be going on to get that sort of change. And I don't think it is just because the B18 and MMM aren't 100% exactly comparable but for the ground channel, or that AMB somehow completely borked up the ground on the B18.

So what is really going on that the ground channel is potentially improving so much?

Quote:

and we All Know Vinyl is Superior no?

Heh, if you say so

I'm a FLAC over USB guy myself. About as far removed form vinyl as you can get.