Why Single-Ended?

Phase has to be inverted somewhere, either for TRUE balanced (XLR cable) or FULLY balanced design.  It involves extra circuitry that does not make sound more transparent.  For short connection, in electrically quiet environment, single ended design might be a better choice.  As for the FULLY balance design - it offers slightly* better common mode electrical noise rejection at very high frequencies and cancellation of even harmonics produced by the amps.  I don't care for both since my connections are short, while shields and twisted pair, in my XLR cable, work really well.  Cancellation of even harmonics, produced by the amp, makes it sounding colder, while odd harmonics, responsible for brightness, are left intact.  Also, FULLY balanced amp has practically two amps inside and costs much more.  I could buy much better single ended amp instead. That's why I would never buy FULLY balanced amp ("Fully" is not always "Better").

* In order to provide good common mode rejection two halves of FULLY balanced amp cannot be independent.  Negative cross-feedback has to be used to equalize gains of each half and that might be far from perfect.

Converting single ended - balanced - single ended using transformers, makes sense if you have a lot of electrical noise and connection is long. Otherwise transformers won’t add anything positive to sound. They will introduce some low frequency distortions. If your preamp and amp are both differential you still might like single ended connection more. (my power amp has only balanced input - I had no choice).  XLR cable's shield is grounded on both ends shorting that way chassis and possibly creating ground loops.

Al, since I don’t see removal of even harmonics (while leaving odd harmonics intact) as an advantage, then the only thing that speaks for Fully balanced amp is better noise rejection in comparison to amp with True balanced input. Balanced input of my amp goes to instrumentation amp (THAT1200) that has CMRR of 90dB@60Hz and 85dB@20kHz. I don’t believe that it can be achieved without cross-feedback in Fully balanced amp. 90dB would be equivalent to overall gain setting (all stages) resistors matching to 0.003%. This cannot be done (be stable) even with multiturn trimpots, etc. Perhaps Fully balanced design offers better rejection at higher frequencies, but I doubt it - frequency response of both halves would have to be identical. 

Perhaps it is achievable using differential stages, but unless gain setting resistors are not on the same substrate stable 90dB does not seem possible.  Perhaps he is using very expensive matched resistors, but price of his amps justifies the cost of components.

I can hear the difference between different XLR ICs - perhaps Ralph was talking about matched 600ohm input/output designs that I’m not familiar with. In any case, it is valid for both True balanced and Fully balanced.
Again, I would buy an amp if it sounds good and not because it is Fully balanced. If anything, I would avoid fully balanced amps as too complex (too many parts to fail).

Thanks Al,  Third harmonic, that supposed to be euphonic, will be the same in fully balanced or single ended design.  Removal of the even harmonics cannot make amplifier sound "warmer".  It is achieved by reducing higher order odd harmonics produced by the amp.  It is likely related to excellent design and not topology itself.

This statement is false. A fully differential amplifier does not have twice the parts and is not nearly two amps inside! This is a very popular myth.

 
I don't know your designs, but usually output stage, at least in SS amp has to be doubled, as well as stage driving it.  You can place differential amps in front of it, but most of the expense is already there.

If this is the case its probable that your gear does not support the balanced standard. Let me guess- you can run the signal single-ended just by disconnecting pin 3, right? 

Not sure of that.  Perhaps, since it is done without input transformer there is ground reference for instrumentation amp that is usually connected with higher value resistor (and small cap) to a chassis. Signal is still differential and I cannot understand why would it reduce effect that cable brings.  Cable capacitance, inductance and dielectric absorption are still there.  Do you think that extremely dirty copper would sound wonderful in balanced cable?

This harmonic structure thing is another myth. How it works is, if the circuit is fully differential and balanced, the primary distortion product will be the 3rd harmonic, at a diminished level 

It does not make sense.  I understand how even harmonics are eliminated but don't know of any mechanism that would remove odd harmonics in fully balanced amp.  Remember we're talking about Fully balanced amps in general - not only your designs.

One thing you are not taking into account is how distortion compounds from stage to stage. If the gain stage just does not make the distortion, it can't be compounded by the distortion of the next gain stage. We only have one stage of gain in our amps, so higher ordered harmonics really don't play a role. This allows the amps to be very relaxed

That's great, but it will be also true for well designed single ended amps.   There is no gain difference between topologies that would explain any additional distortions.

I'm not sure why Nelson Pass designs Fully balanced amps.  Perhaps it is market demand, that the tone of OP question suggests.  For the same reason designers still use unregulated linear power supplies instead of line and load regulated extremely quiet SMPS  (Jeff Rowland uses them).  I suspect that in your case objective, since there is no output transformer, might be to get more output power.  Once you have double output stage front is just a small addition.

Thank you Charles.  While not opposing any topology I think less is more.   As electrical noise is concerned, home environment is in most cases benign.  I argue to learn/understand and I have, especially from Al and Ralph.  This forum is great.

The way balanced line eliminates cable artifacts is twofold. First, ground is ignored, so the shield is not part of the sound (nor the is noise to which its exposed; in a single-ended system the shield is part of the signal path).

You cannot eliminate capacitance between wires.  Low output impedance helps to lower effect of it, but it so does with single ended design.  It will be difficult to get rid of shield to wire capacitance since many preamps have balanced output referenced to ground.  It will be pretty much any transformerless output stage including my Benchmark DAC1.

I get that it does not make sense for you. But that does not mean it does not make sense. The idea here is to remove distortion sources (now this is strictly my opinion). If you can't use feedback to get rid of distortion, how do you get rid of it? Eliminate distortion sources! A common complaint about tubes: 2nd harmonic (ask any solid state guy). OK- fully balanced differential design gets rid of the even orders. Now we are left with the odd orders. To reduce them, we set bias points in the voltage amplifier such that it cancels the odd orders. Then design the circuit to use as few stages of gain as possible (in our amps there is only one stage of gain, making them a simpler signal path than an SET). Use triodes throughout. Get rid of the output transformer (which may or may not add distortion). Take care to avoid obvious diode issues (proper metallurgy) in component selection. Stuff like that.

You forgot that we are discussing Fully balanced topology and not design of particular amp in general.  Setting bias point, as you described, can be done to single ended amp and has nothing to do with issue that we're discussing.  Please tell me how Fully balanced amp topology reduces third harmonic better than single ended amp.

Thank you Ralph for explanation about third harmonics. I understand that advantage of the fully balanced configuration is removal of second harmonic to better hear the third one (that you try to lower?). As for the balanced cable - yes it can run further because of superior noise rejection but still suffers from effects of the capacitance between wires. There is no way of getting rid of it. Capacitance to shield is different story. Reducing it would require usage of low capacitance signal transformers. Otherwise - even if you completely "float" signal ground in preamp and amp it will find return, by substantial capacitance of power transformer, to chassis (that you earth ground) hence to shield attached to it. Am I missing something?

Rane paper shows XLR cable (pattern1) with shield grounded at female end only, while common XLR cables, sold in stores, have shield grounded at both ends.