To balanced or balanced not ?
On paper, a headphone (each side) is a “2 pin device”. Between the two pins, there is a “load”, impedance. One can impose a desirable voltage (signal) between the two pins, and Ohm’s law rules – the current that flows in the can (the load) is the voltage divided by the impedance. No one can argue with that fact.
The driving voltage is the voltage DIFFERENCE between the pins, and of course that voltage is changing all the time. But the point here is that if both pins are given the same amount of additional unwanted voltage (for any reason), the difference is UN affected, and thus there is no undesirable outcome.
That is the COMMOM reason for balanced operation: With balanced hardware, both wires leading to the two pins occupy the same environment, thus both sides tend to pick the same electrical undesirable interference, and the balanced connection offers a cancelation of the disturbance.
However, with short cables such as headphone cables (only a few feet), and even more importantly, with such low impedance on the driver side, as well as a low impedance on the headphone side, the cable pickup issue is not a real world issue. So the balanced connection for headphones is NOT about eliminating electromagnetic pickup.
To prove it, unplug the headphones and listen. Do you hear anything? Probably not. If you do, place a resistor (say 100 Ohms) between the Tip and RING of the headphone plug (that represents“average” driver source impedance. This time surly you will not hear a thing!
So since it is not interference pickup, what could it be? Let’s “dig into it”.
The difference between balanced and unbalanced configurations is:
In balanced connection, one has 2 wires per can, thus 4 wires in total.
In unbalanced connection, one has 3 wires. One of the wires is a COMMON wire to both cans. This is the standard way to use headphones. So is there a real difference?
If the common wire is a perfect conductor (zero Ohms impedance), then there is no difference between the schemes. But say the wire is NOT zero Ohms. Then, the current that runs through the left can will introduce some voltage drop on that wire. And that voltage drop is SUBTACTED from the signal aimed at the right headphone. Of course the "opposite" is also true (right subtracted from left).
One can test for it by applying a MONO sound to say the left headphone while listening ONLY to the right headphone (keep the left headphone “blocked” and as far away as possible). Better yet, "replace" that left can with a "dummy load" - a resistor with a value of the headphone impedance. With no signal to the right headphone, anything you hear is due to the “leakage” from the driven left headphone into the right can. The mechanism would be that common return path wire…
Of course you can reverse the left and right and redo the experiment.
A designer will of course calculate the extent of the potential issue. In fact, it is a worthwhile thing to do, because it teaches us something. Say the “common wire” is 10 feet long. Say it is 24AWG (the cross section area, the equivalent “diameter” of the copper conductors). Then we are dealing 0.25 Ohms per 10 feet. Of course at 18AWG (thicker copper) the same 10 feet will be less resistive - .0604 Ohms. So how much does it matter?
We have voltage divider issue. Say the can is 600 Ohms, and the wire is 18AWG 10 feet long. Then the voltage drop expressed in dB is -80dB. But say the headphone impedance is 30Ohms, and the wire is 10 feet of 24AWG, then the drop in the wire is around -41.5dB
Clearly the drop in the common wire depends on the wire resistance (wire material, length and diameter), and even more important is the impedance of the cans. The amount of that drop in a high impedance can is much lower than the loss in the wire for low impedance can.
I have not quantified too well how much impact there is at all sorts of combinations of wires and cans, but I tend to think that at 300Ohms, the common wire impact is relatively tiny (if any), but with low impedance, I am not ready to argue that there is no audible impact.
So what sort of impact would it be? We are subtracting some tiny amount of the left signal from the right signal, and also subtracting some tiny amount of the right signal from the left. We are making the signals less just slightly different, more alike. That means, we are impacting the stereo field, making it slightly narrower.
This should really be a blessing, because headphones tend to provide more stereo separation then should be. I explained it in my description of the DA11 PIC™ (playback image control). In fact, the realization that we hear too wide an image with headphones was THE REASON why I designed the PIC™. I want listeners to hear the music with the proper stereo image, normally made for speaker listening.
But people get used to whatever they use most, and many listeners of headphones are used to the extra wide image. I am not surprised to find out that some set the PIC to wider setting, which is extra wide stereo. I too find it enjoyable at times (not often); it makes the music more “airy” and “softer”. But the extra wide image is not the real thing. The real thing is what one hears acoustically in the original performance before the recording… I expect that people that go to a lot of acoustic performance would better appreciate narrowing of the image when listening to head phones.
So is that all that can be said regarding balanced? Of course not. One can dig in and look for more. I would "dig in" for answers in the circuits. In fact, a truly balanced driver circuit can reduce (cancel) EVEN harmonics by a few dB. This is not at all a well known fact, but it is an easy to prove fact. That does not make balance a necessary thing - one can make a very good unbalanced driver, but it is worth noting, as another possible cause. (It is unfortunate that a balanced driver does not do a thing for ODD harmonics).
One of the solutions to minimize the common wire impact, would be to replace the headphone wires with thicker wires (lower AWG), and if possible make the wires shorter. I noticed that some of the balanced cans do have pretty thick wires – well, thick wires will improve the unbalanced a lot more then balanced…
I guess if I offer the DA11 with the PIC™ as another solution, with more adjustment range, both for wider and narrower image, it will be taken as self serving. Indeed it is. But the fact is: I made the PIC™ as a reaction to that balanced vs. unbalanced issue. It was around 3 years ago. I gave a lecture at the head-fi in silicon valley, and someone asked me about that balanced issue. I did answer it with the common wire explanation as above, and when I went home, I started work on the playback image control, the PIC™.
I am pretty sure that some people would rather not take the time to dig into the details, and rather just make an announcement such as “balanced is better”. I would suggest that much depends on specific implementation and I would hope that anyone that want to listen and seriously compare balanced to unbalanced will take the time to make sure that the levels are matched to within 0.1dB, and the matching should be done with test gear prior to listening. Do not match the levels by ear! In fact, I would suggest that for ALL listening comparisons