Differential Balanced Sound Quality

The term is commonly used both in text books and the professional recording environment as shorthand for the inverted signal in a balanced pair..

It is and many people are confused by it! I apologize if you see this as being patronized, that was not my intent! When I am writing anything here, I am assuming that others are reading along. For that reason I try to avoid getting too technical.

If you do a quick google search on differential amplifier you will see that by far the most common use for this term is exactly what I have described and this is what is used in the majority of equipment with balanced inputs.

OK. Here is just exactly that:
https://www.google.com/search?q=differential+amplifier&client=ubuntu&hs=WBe&channel=fs&a...

Take a look at the first hit (this is an images search). The first image shows both what you’ve described **and** what I described in the same image, the latter of which does not have the Johnson noise issue.


If the device has a transformer at its input, this noise won’t occur. If an opamp is used, it might or might not be used that way- depending on if the input is more of the instrumentation variety rather than a single opamp device (and the reason for not doing that would be if the internal circuitry of the preamp or whatever is fully differential; at that point you’ll need opposing outputs from the first stage which a single opamp can’t provide). We recently had a McCormick amplifier come through the shop; its balanced inputs were executed as I described in one of my prior posts (and as you see in that first Google image), using dual matched input transistors, probably similar to a MAT-12
https://www.analog.com/media/en/technical-documentation/data-sheets/MAT12.pdf
with a common emitter circuit.


Our conversation started with you stating that a balanced input would be noisier due to the resistor noise:

The bit that a lot of people are unaware of is that balanced inputs are actually noisier than single ended (RCAs etc.) in that they utilise relatively high value resistors that introduce their own (johnson) noise into the signal.

IME the statement is only true if opamps are used *and* the internal circuitry is single-ended (in this case, if succeeding opamps are used but with the signal only applied to one input, I would regard that as single-ended. Another way to look at that is look at the volume control; if it has one deck for each channel then its single-ended. This thread is about ’Differential Sound Quality’ so I have to assume that circuits/products that are in fact fully differential are included.).

So if I start with a conventional amplifier block with an inverting and non-inverting input and a single output. A single ended amplifier input might be a 100Ω series resistor followed by an RF filter & DC blocking capacitor, in non-inverting mode the input impedance is set by the resistors to ground at the input so it’s not difficult to maintain a high input impedance alongside a low thermal noise from the series resistor. The actual impedance of the non-inverting input is so large that it can be pretty much ignored.

If you take the above example and feed the cold signal into the inverting input, the series resistor on the cold input will dictate the maximum input impedance as the current will be flowing into the virtual ground at the summing point. So 100Ω is now out of the question. You might for example choose to go with 10kΩ series resistors on both inputs, that’s 20dB more thermal noise than 100Ω.

@pragmasi 


There's no such thing as 'cold signal'. There's non-inverting and inverting. What you're describing (as seen in your link) has to do with an opamp which has differential inputs but a single-ended output. A differential amplifier always has dual outputs. Look again at the diagram I linked:
http://www.atma-sphere.com/en/resources-understanding-our-circuits.html

When I look at your amplifier I see that you have two outputs and I suspect that is the source of confusion... at what point does the cold signal get inverted?.. or does it connect to the negative speaker terminal?

Really, I think this 'cold signal' thing is confusing you. Both inverted and non-inverted signals are 'hot'. They must both be treated the same way. If you are referring to the inverted signal (for example the minus output of a phono cartridge) it gets inverted at the output of the device to which it was applied as an input.

A related question that has been bothering me. If your circuit isn't perfectly duplicated after signal splitting, doesn't this introduce timing errors upon recombination?

@cal3713 

No, at least not at audio frequencies and beyond. Timing becomes an issue at radio frequencies but we have bandwidth to 400KHz in our line stages and it does not seem to be a problem there.

I really take issue with the term 'signal splitting'. That's probably because I don't see that happening. A differential amplifier does have two halves; these are intimately coupled together in a tube circuit at the cathodes, in a transistor circuit, at the emitters, and in an FET circuit, the sources.

I'm going to use the term 'emitter' in place of 'cathode' or 'source' in the following explanation:


In all cases, since the current for both halves is flowing thru the common emitter circuit, if one side of the differential amplifier is turned on, all the current goes thru that side so the other half is forced off and vice versa. It important to understand that this process occurs in real time; there's no 'slight delay'; for one side to turn on the other side **absolutely is** being turned off in perfect tandem.


If both halves are turned half-way on their outputs will be at the same level. At all times the current through the emitter circuit is constant. Because the devices aren't perfect, its advantageous to put a current regulator in the emitter circuit called a 'Constant Current Source' (CCS). The more constant the current in the emitter circuit, the more theoretically perfect the differential effect. To this end the quality of the CCS is arguably as important than the gain of the devices used in the differential amplifier.


If you drove only one half of the differential amp, if it had perfect differential effect, both outputs would be equal and opposite. In practice there are slight differences. But if you have a succeeding differential gain stage these differences go away- they are not exacerbated.


Because there are slight differences when driven single ended, when you drive them balanced the distortion is slightly lower. The higher the CMRR (Common Mode Rejection Ratio, measured in dB) the less this is so.

Differential amplifiers get their name from a simple fact: They amplify what is different between their inputs. If one input is at ground, then they amplify the side that has the signal (single-ended). If both sides have the *same* signal they won't amplify (because that signal is Common to both sides). If the signals applied are opposite phase of each other, then they get amplified. It doesn't matter so much if the two inputs aren't exactly equal; what matters is that they are opposite- the outputs of the differential amplifier will even things out. There's no 'recombient distortion' or any such nonsense.


The variable here is the Common Mode Rejection! If its poor (less than 80dB) what I said in the paragraph above starts to go out the window. If its very high (140dB) it really won't be measurable whether the input is single-ended or balanced.


Achieving a good CMRR value isn't hard. We can do it with 6SN7s.

There is a reason: In most consumer (and professional) amplifiers the balanced signal is dealt with at the input by a differential amplifier necessitating the need for a series resistor on the inverting input. In the worst case the diff amp is the load seen by the source so the resitor needs to be large enough to present a reasonable input impedance (that’s the ’relatively high’ part).

@pragmasi

Any gain stage (differential, balanced or not) might need stopping resistors at its inputs, but I suspect this isn’t what you’re talking about.


We’ve been building differential amplifiers for balanced inputs longer than anyone else in home audio (IOW we’ve introduced balanced operation to home audio with our MA-1 amplifier in 1986 and followed up with the first fully differential balanced preamp in 1989), and using vacuum tubes have nevertheless gotten fairly good CMMR values, in excess of 100dB. Each input (pin 2 or pin 3 of the XLR) sees the same input impedance, which is what you would expect of a balanced input, and both have the same resistance between the XLR connection and the actual grids of the input tube. So far we’ve not seen any such need for a resistor as you describe. For what are you thinking this resistor is needed/what’s its function?


You can see a simplified example of one of our input circuits in the article at this link:

http://www.atma-sphere.com/en/resources-understanding-our-circuits.html

As you can see, the diagram is a textbook example of a differential circuit. I really am mystified by what resistor you’re talking about! Can you explain in greater detail?

Alternatively, the balanced equipment discussed above separately amplifies the positive and negative parts of the electrical signal. This requires double the circuitry because you have to amplify two signals (0-to-positive and 0-to-negative) instead of one (negative-to-positive) as in single-ended equipment.  If you imagine music as a sign wave, then balanced equipment is separately amplifying the upper and lower halves of the wave. The benefit of this added complexity is that any noise introduced into the signal by your equipment will be cancelled out when these two "half" signals are re-combined to make the full-wave signal that drives your speakers. This is because any injected noise will appear positive going in one amplification circuit and negative going in the other... add them together and you get automatic noise-rejection because they'll be in exact opposition.

@cal3713 The bit about positive and negative parts is problematic. Usually for balanced operation differential circuits are used, and its probably easier to understand that they amplify **opposites** rather than 'halves'. And it does not require double the circuitry- this is because in a differential circuit, the cathode (or emitter, or source) circuit has all the parts used in it in common. As far as noise or distortion goes, the benefit of cancellation accrues with each stage, not just at the loudspeakers. Finally, for a given single-ended gain stage, if done differentially will have theoretically 6dB less noise generated. This can be significant from one end of a circuit (like a preamp) to the other. Two stages, each with 6dB lower noise, that makes 12dB... So you don't need as many gain stages. In our full function preamps there are only three stages of gain from phono input to main output and they can work with LOMC cartridges; contrast that with a typical single-ended preamp which will have at least 4 stages of gain (unless an SUT is employed) or more to do the same thing and you can see that the parts count myth is just that.

The bit that a lot of people are unaware of is that balanced inputs are actually noisier than single ended (RCAs etc.) in that they utilise relatively high value resistors that introduce their own (johnson) noise into the signal.

This statement is problematic. There’s no reason why any such ’high value resistors’ be used in a balanced circuit that aren’t also in a single-ended circuit.

Because it is so obviously the wrong way to go. The number one factor in quality sound is quality parts. Going balanced requires twice as many parts. A whole duplicate circuit.

@millercarbon This statement is false and is a common myth: it ignores the bit about ’differential’. If the circuit employs differential amplifiers, then the parts count is about 50% higher rather than double. There is a distinct advantage to differential amplifiers (which are often used in amps with single-ended inputs, as they are often used as phase splitters); they have much more power supply noise rejection and if the circuit is fully differential, then even ordered harmonic distortion is cancelled at each stage throughout the circuit, rather than compounded (which can increase higher and odd-ordered content).


There is an advantage to this- when the circuit is fully differential, the primary distortion harmonic is the 3rd, which is treated by the ear the same as the 2nd in that it contributes to ’bloom’ and ’warmth’. But in such a circuit, the 3rd will often be at a 10th of the value that is seen in a circuit that has a 2nd harmonic as the primary distortion component. In mathematical terms, the former has a cubic non-linearity whereas the latter has a quadratic non-linearity. When there is a cubic non-linearity, succeeding harmonics drop off at a much higher rate than they do if the non-linearity is quadratic. In a nutshell, fully differential circuits are much lower distortion and so sound more neutral. This is not subtle either- its easy to hear. So there is a distinct advantage to all those extra parts if the designer took advantage of all the aspects of differential operation!

Balanced is totally designed for professional use. Runs of hundreds of meters. Not tens of feet.

This statement is also false. The benefits of balanced line connections are there even if the connection is only 6 inches. I can go into it further if you like; FWIW this idea of ’only for long distances’ is also a common myth. Think about this- you don’t have to know the technical side; the simple fact that you can run such long distances without coloration is a tip off that those colorations won’t be in a 6 foot run either! If you’ve heard differences between single-ended cables, then you know what I’m talking about when I use the word ’colorations’.

I’m not a fan of preamp to amp balanced connections. I’ve always preferred single ended overall. YMMV

@geof3 The reason balanced operation might not seem to bring home the bacon in high end audio is that there is a standard for balanced line operation, which most high end audio manufacturers do not support (and don’t seem to realize how this degrades the sound). When the standard isn’t supported, the cable construction becomes audible like it does when running single-ended. Balanced operation isn’t two single-ended signals running together in a cable! Its a single signal traveling in a twisted pair with both sides ignoring ground. Please Note that I am not arguing with you here- I’m sure that what you’ve experienced is very real. But what you experienced was not how balanced operation is supposed to work. BTW, the balanced line standard is known as AES48. You can always ask the manufacturer if their equipment supports AES48; if they don’t know or don’t know what it is, then its a very safe bet their gear doesn’t support the standard and so will not demonstrate all the advantages of balanced operation.


Those advantages are:
1) lower noise (blacker backgrounds)
2) neutral, transparent interconnect operation regardless of length (more detailed, less colorations resulting in no need for expensive interconnections)
3) no ground loop noise or intermodulations (not as harsh)

But if the standard isn’t supported these things go away and as you’ve heard, might even be worse.