Dual Differential / Balanced?

Al, I agree, that was invalid point - double DF doesn’t make much difference in most cases. As I stated in my older posts speaker’s impedance is mostly resistive and it is in series with back EMF reducing max possible DF to about 1. As long as amplifier doesn’t destroy it too much it is OK. Some of Ralf’s amps had DF<1 sounding fine.

You are right that I assumed two independent amps in differential mode - that’s different from balanced. In order to make identical gain stage in each amp they have to reference each other (like in the first stage of instrumentation amp) instead of being ground referenced. In that case amps aren’t really independent as I thought they should be. Noise rejection in such design should be great but it is not worth extra expense IMHO. Single ended amps with balanced inputs might have great noise rejection as well, while better components can be utilized for the price of fully balanced design.

Charles1dad, Thank you.  I often engage in discussion with Al or Ralph to learn more.  It is difficult to understand why particular architectures are being implemented since it is often fueled by demand.  I like to understand technical merits of design but sound is the only proof.

I won’t even comment on the issue of instability because you can find it in any FAQ/primer on fully balanced design, but it appears that you question my notion that connection with XLR cables is not always the best. Experience of many people on this forum was that RCA connection sounded better than XLR.

That is true, please consider though that there is a standard for balanced operation, of which about 95% of high end audio ignores.

The result of not adhering to the standard is that you may well get inferior operation. When balanced line is done properly (and can be done with single-ended circuits, the Ampex 351 tape machine being an excellent example) the result is that the interconnect cable ceases to be a part of the system sound. A common example of an audio company not adhering to the standard is any ARC balanced preamp. The standard calls for ground to be ignored at the input and output- it is for shielding only. But the ARC preamps put signal current through the shield of the interconnect, which causes the construction of the cable to suddenly become critical. The result is that the cable might have an audible fingerprint when used with such a preamp (and I’m not picking on ARC beyond example, this is a very common situation with many balanced products in high end).

So suddenly something that is supposed to eliminate cable interaction doesn’t; IOW the balanced cable is going to editorialize just like a single-ended cable, which IMO makes me wonder ’why bother?’ As you pointed out, others do too.

IME, audiophiles obsess over interconnects and for good reason- they often sound different and often one sounds better than another in their system. I’ve experienced this a lot; I’m not one of those engineers that thinks this is all some sort of audiophile fantasy!

But the fact is, if you can get the interconnect cable to not editorialize, it will result in better sound. So far the balanced line system is that solution. That is why I mentioned the AES file 48; even though I’m not a particular fan of the AES in general (lots of the types of engineers I just mentioned tend to hang out with that otherwise august group), the definition that is file 48 is rather handy and succinct. Unfortunately unknown or misunderstood by the majority of high end audio.

So yes, there are many on this forum that think RCA connections sound better out of experience, I’m just pointing out that is happening due to the balanced system not being executed correctly (which BTW, also means that the length of the cable is irreverent). That is why there is even a debate in high end audio about the benefits of balanced!

Well Ralph your  explanation of misuse /improper application of "balanced " circuits would explain why I don't usually find them better sounding in many audio systems. Nonetheless single ended /unbalanced components can  often sound fabulous. Superior implementation of either circuit type is mandatory. 
Charles, 

Thank you Ralph, I will keep reading more.  I use fully balanced stages in low level amps without any ground reference.  It has bandwidth of few kHz only, but full scale of the signal is in single millivolts.  At the end A/D converter with differential input provides its own ground reference.  I can see a problem with fully balanced audio amp design that has no reference to ground.  Both outputs can be floating together since without output current or voltage difference feedback won't react to that.  It needs ground reference somewhere or some kind of servo on common mode. 

Shield should never be used to carry signal but it was unfortunately common practice long time ago.  Scope's coax is a good example of that.  Scope with shorted leads, touching circuit under test, shows phony signal - since ground return path (possibly thru supply) causes current flow thru the shield, that input amp (referenced to scope's BNC GND) sees as a signal (voltage drop on the shield).  Shield converts common mode to fake normal mode signal.

My small Rowland has only XLR inputs - perhaps mature decision in class D amp, so I was not able to compare it with single ended RCA cable.  I still can see substantial sound quality difference between decent XLR cable (AQ King Cobra) and very good one (AZ Absolute). 

I have to read AES48 standard, you mentioned.  There was wonderful EDN magazine issue on that many years ago. Grounding and shielding is considered by many as black magic.