I don’t see benefits of that. Removal of even harmonics will make amp sound "colder" while doubled gain and power stages cost money. You can have much better unbalanced amp for the same price. |
I can't see how fully balanced system can help with noise reduction. Common mode noise introduced at the input will go thru multiple stages of pre and power amp in both chains of balanced amp. Sum of gains in each chain can be different especially for hight frequencies resulting in amplified noise across speaker terminals. The more stages in fully balanced configuration the bigger max. unbalance. It is better, IMHO, to reject common mode noise as soon as possible and use unbalanced configuration. Fully balanced configuration will reduce a bit harmonic distortion by removing even harmonics. It is because even harmonics are always positive, even if phase is inverted, resulting in the same signal on both outputs, hence zero difference. Unfortunately unpleasant odd harmonics, will stay. |
A single-ended amplifier with a balanced input and/or output is still a single-ended amplifier and does not offer all the advantages of a differential amplifier, which provides common mode rejection.
This is not true, IMHO. Single ended amplifier with balanced inputs can provide huge common mode rejection, better in my opinion than fully balanced amp. Fully balanced amp cancels (at the speakers) common mode input noise, only if gains of both half-amps are identical. Gains can be matched with discrete components only up to certain point. Fully balanced amps have some advantages like no output ground current, cancellation of even harmonic, doubled slew rate etc. but have also many disadvantages. One of them is the fact that signal goes true many more stages (bad for clarity), output impedance is doubled while the cost (and complexity) is much higher. |
Cleeds, yes, balanced operation is used in professional audio in form of balanced cables, balanced inputs etc. I’m all for it. I just don’t see how fully balanced amp can have better noise rejection than plain amp with balanced input. Instrumentation amps, like one used in my Rowland amp have 90dB noise rejection at 60Hz. That would imply gain matching of two independent amps to 0.003% to achive the same result. I also don’t see how removal of even harmonics only can help. Since outputs have no ground reference (current doesn’t flow thru ground) feedback might cause instability. That requires additional third feedback for common mode.
Again, I don’t see how this scheme can help with noise. In addition even connection with XLR cable is not always the best. If you don’t have a lot of noise you might be better of with plain single ended RCA since additional circuitry to make balanced output affects sound. |
@1 - I've never said that it doesn't. But it will remove cable noise by means of either balanced input in single ended amp or fully balanced amp.
@2 - again, I did not say it removes even harmonics from the signal. It removes even harmonics produced by the amp at any stage by means of cancellation AT THE OUTPUT. These even harmonics often make amplifier sounding warm and euphonic. Yes, the third one is not bad, but higher odd harmonics are quite nasty. Removing only even harmonics while leaving odd harmonic intact will lower total THD but also will make sound colder, brighter.
@3 - Output impedance will double since there will be two output stages - similar situation to bridged amplifier. Complexity is much higher no mater how simple the circuit is. If signal goes thru one gain stage only in single ended design, it will go thru two stages in the balanced one (not to mention two output stages). You might not see a problem with that, I do.
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. I'm sure that Al (our technical guru) would agree with me - that balanced output, because of additional circuitry might not always be the best. I will leave it at that. |
Al, I was looking at overall output impedance that speaker generated EMF sees, and that would be double.
I also stated that fully balanced design is not only unnecessary to provide great noise rejection, but in fact might be worse than one achievable in single ended amp with balanced input. High CMRR would require perfect gain matching of both "legs" of the amp and that is not possible at the level of good instrumentation amp (90dB @60Hz in my amp). It would require 90dB (0.003%) gain matching with discrete components, calling for <0.0015% resistors, that don't even exist. It gets even worse at higher frequencies where both multistage amps/legs would have to have identical frequency characteristics. Same is of course true for instrumentation amp, but to much smaller degree. Resistors are laser trimmed on the same substrate while amplifier's bandwidth can often be much higher (22MHz in my amp). |
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.
|
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. |
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. |
