Are headphone balanced cables snakeoil?

[SIY]

However, simply balancing the impedances does not make two single-ended inputs differential.

Of course not. That's my point- something can be differential and not balanced or balanced and not differential.

 

[SIY]

Here's an example that might clarify things. Let's say we have a single ended opamp with essentially zero impedance to the signal ground in its box. We put a resistance in series with its output, let's say 300R and call that output +. At the same time, we put a 300 ohm resistor between another output called - and the same ground.

Now we have another box with a differential input to which we connect the + and - outputs of the first box. There is no voltage drive to the - output of Box 1, but the line is still balanced and the second box's CMR will be fully realized.

 

[tomchr]

Of course not. That's my point- something can be differential and not balanced or balanced and not differential.

I think we agree that those examples would be pretty bad circuit design.

I appreciate the final example (Post 22), though. I had not thought about that case. That should also mean that we get the full CMRR even if we make both impedances on the driver equal to zero (so no series resistance on the opamp and the "-" output goes directly to ground). That'd be like using what I've been calling a pseudo-differential cable.

Tom

 

[Deleted member 2944]

Tom,

Check out Bill Whitlock's paper Jensen AN-003. (Example 2.4)
The resistor to ground technique described by SIY is a common scheme and used in a variety of professional equipment.
Bill describes it as a "true balanced output." I think you would probably disagree with that label, but there it is.

Dave.

 

[tomchr]

That's great. Now watch what happens as the series R goes to zero and the series C goes to infinity. You end up with the Typical Unbalanced Output shown on page 1. It's still a balanced circuit according to Whitlock's definition as the impedance to ground is zero both for the (+) output (output of the opamp) and the (-) output (ground). My question is whether you still get the CMRR if you connect to a differential-and-balanced receiver in that case.

Also note that the input impedances of Whitlock's "Typical Balanced Input" are not balanced. Neither is the "TYP PRO INPUT" shown on page 2 of the app note. Recall that the input impedance of the non-inverting input of an opamp is near infinite, whereas that of the inverting input is near zero due to the virtual ground (caused by the loop gain and negative feedback).

Tom

 

[March Audio]

OK, so you have a couple of things going on here. I think you're trying to apply some concepts that relate to differential signalling at line level to your headphone case. Let me see if I can tease that apart. I'll start with line-level signalling between two pieces of equipment and treat the headphone case at the end.
Note that I use the terms single-ended = unbalanced and differential = balanced interchangeably.

Single-ended/unbalanced connections between pieces of equipment (say a DAC and an amplifier) suffer from one problem: Ground impedance. When you connect multiple boxes together, any difference in their ground potential will cause an error current to flow in the ground connection between the boxes. This sets up an error voltage (V = I*R) across the ground impedance. This is a problem because a single-ended/unbalanced input cares about the voltage difference between the signal conductor and ground, so when the ground potential is different between the two pieces of equipment due to the error voltage, the error voltage is treated as signal. For more detail, read Bruno Putzeys' article on the subject: https://www.diyaudio.com/archive/bl...d1460406090-bruno-putzeys-micropre-g-word.pdf

The solution to this is to transmit the signal as a voltage difference between two conductors and leave the ground as a shield. An ideal differential/balanced input only cares about the voltage difference between the two wires in the signal pair. In reality, the voltage on the signal pair has to be within some voltage of ground (common-mode voltage range), but this range is typically determined by the power supply of the differential input, hence, on the order of ±15 V.
Because the differential input only cares about the voltage difference between the two wires in the signal pair, any voltage that is imposed equally on each wire in the pair will be rejected (common-mode rejection, CMRR). This is how a differential input rejects hum, for example. Not there are limits to this. A good differential input will have around 80-100 dB of CMRR, so if you somehow manage to get 1 V of hum injected on to the wires, you'll get 10-100 uV of hum on the output of the differential receiver.
Using a twisted pair further minimizes the hum by ensuring that any electric field that couples to the wire is imposed equally on both conductors. The twisted pair was figured out and patented by Alexander Bell back in 1881 for use in phone systems.
There are some more modern cable configurations that are even better than Bell's original twisted pair. Canare's Star Quad, for example. See demo here:

The bottom line is that differential signalling is a really good way to ensure good signal integrity, in particular in noisy environments. That's why you see differential signalling used in pro setups with long wire runs. It would not surprise me if residential setups are operating with the mains hum just below audible. At least I found subjective improvements in sound quality when I converted to differential signalling on all my gear.

Now for your headphone case: Marketeers have caught onto balanced = good, so now anything "high-end" has to be balanced. "Back in the day" when amplifiers had 1% distortion, you could lower the distortion considerably by converting to a balanced design. I have a few issues with this: 1) You only cancel the even order harmonics (those that many find pleasing to the ear) leaving the odd order harmonics (that many find to sound harsh) in the signal. That doesn't sound like a wise thing to do. 2) Most cancellation circuits, in my experience, end up making the performance worse rather than better. This is especially true at the ultra-low distortion levels of modern opamps. 3) There is really no compelling technical reason to use cancellation schemes to lower the THD as modern parts already deliver THD that's orders of magnitude below audible (OPA1612 0.000015% anyone?)
Now before someone cries, "but THD isn't everything!" ... that's true. THD is not everything. But by improving the THD you usually end up with a lot of other improvements as well (IMD and multi-tone IMD spring to mind).

Where balanced is relevant to the headphone case is for crosstalk or coupling between channels. If you share the ground connection between the two headphone drivers, you will have some of the left channel signal mixing with the right channel signal. Even the shared ground in a 1/4" plug can cause crosstalk. For example, I measure 115 dB channel separation in my HP-1 when using the XLR output and about 95 dB when using the 1/4" output. The difference is due to the shared ground in the 1/4" plug. Now, both are below audible and I've never detected any difference in subjective listening tests, but it is definitely a measurable effect and is backed up by theory (Ohm's Law, to be specific). Both outputs on the HP-1 are single-ended. In the XLR output, I just ground the (-) side of the XLR plug. I see no advantage of a balanced amplifier for reasons outlined above.

So to answer your question: If the only shared ground in the connection from the amp to your headphones is in the 1/4" plug, I doubt you'll notice any difference between 1/4" jack and XLR connections to the headphones.
Now, if you go out and spend $2k on a pair of Super Duper Balanced Bling headphone cables, I guarantee you that you will hear a difference. But that has to do with cognitive psychology and not any difference in the stimulus that reaches your ears. Just make sure you read the manufacturer's marketing babble so you know what to expect and which improvements to hear with the $2k bling cables.

I hope this answers your questions and furthers your understanding.

Tom

My measurements with this (a while back - not with my forthcoming HPA1 amp) indicated this crosstalk was less of a problem than many indicate. It depends on the headphone cable configuration and ground impedance in the amp.

If the headphones have separate ground to the drivers commoned at the plug end, and the the ground of the circuit is low impedance I dont consider it a "real world" problem. Measurable yes, significant probably no.

The measurements I performed were not particularly optimised but I saw 115dB fall to 105 dB with a 32 ohm resistive load. I dont think many would hear that level of difference, although I am sure there will be some that claim night and day difference.

Wired directly no 6.3mm plug and socket

Wired through 6.3mm plug and socket no load

wired through 6.3mm plug and socket 32 ohm resistive load

 

[solderdude]

To elaborate the above post a bit more (as well as other posts)

Advantages could be a higher available output voltage (= power) in high impedance headphones.
This is interesting for battery fed devices as one can (almost) quadruple the output power.
You get double the output voltage (as you basically have 2 amplifiers in counter-phase) and since P= U2/R.

For lower impedance headphones there might be no 'power' benefits when the current limits of each amplifier stage is reached.
Say each amp 'section' is limited to say 50mA and these 2 amp sections are in series this simply is the maximum current.
There cannot flow more current and since R is constant the power does not increase for low impedance headphones.

Then of course there are conditions where the current limit is higher in that case there is some increase in output power.

For desktop equipment creating enough output voltage and having enough current is no problem.
The most common problem is voltage rails which could be limited by the usual op-amps used.
There are plenty tricks for that + discrete designs.

Still we can find enough DT equipment sporting 'balanced' (as in a 4-wire headphone cable + 4 wire connector)
The reason for that is 'demand' and 'prestige' one could argue. Partly true perhaps.

Balanced is used in studio equipment to lower noise floors and above all influences from 'garbage inducing' surroundings.
Mains cables running alongside cables, lighting cables etc. over long distances could be in close proximity over certain lenghts.
What does this have to do with headphone cables... nothing really as these usually don't pick up something unless you have your phone against your cable and an EMC sensitive amplifier section in there.

So ... what could be the reason for a balanced cable for headphones other than have more power for high impedance headphones when a low power supply voltage rail is present.

The answer is simple.... Crosstalk.
No... not crossfeed. In fact it is closer to the opposite of crossfeed.
And then ONLY for 3-pin (and being cheap also 3-wire) headphone cables.
It is a bigger 'issue' for lower impedance headphones.

The reason is the ratio between the headphone impedance and cable+headphone plug+ internal wiring (in some cases) of the RETURN wire only.
The resistance of the signal wires does not matter here.
I can make a long post here with pictures but wrote it down in this article already.

The other advantage of balanced cables for headphones, aside from more power (depends on situation) is improved stereo separation due to the lack of a common return path due to usage of cheap 3-wire cables.
The 4 wire 'balanced' signal thus has no common return path.
Those headphones that already have 4 wires (usually the dual entry headphones) only to combine the return wires in the 3.5mm or 6.3mm plug won't really benefit from 'balanced'. Well aside from the available output voltage increase when played VERY loud.

In short: common return wire gone in 3-wire cables and higher maximum output voltage (depends on circumstances) for 'balanced' operation.

 

[Frank Dernie]

hey everybody,please lets keep this discusion about headphone cables.I understand balanced cables can be good in studio recording and other things,dont know much about phono,not my thing but I want this thread to be about your typical audiophile sitting in front of PC or on chair listening to his headphones,I would like if this thread was about headphone cables only.

In that case the answer to your question is you don't need balanced cable for headphones, selling them as an audible improvement is a scam.

 

[Graph Feppar]

But what about crosstalk? I read from tomchr that 6.35mm jack have crosstalk 90db down,I also saw one measurement where 3.5mm had crosstalk that have risen to -40db at 20 KHz.

I wonder how various cable types,lenghts and jacks contribute to crosstalk.Also I wonder how headphone impedance affect crosstalk,I have read that higher impedance headphones cause less crosstalk.
,I think its becose there is less current running through so less magnetic coupling.

The problem is when measuring balanced vs single ended crosstalk in cables and jacks is that you cant just use the single ended and balanced outputs of amp becose they are different so the measurement would be different,these things would need to be tested with signal generator.

 

[SIY]

Higher impedance headphones will draw less current, so the crosstalk induced by the shared common will indeed measurably decrease; the key factor is IR drop, not "magnetic coupling." But for any realistic cable length, crosstalk is pretty negligible from an audibility standpoint. Although -50 dB (as an example) might seem alarming, it's seriously doubtful you'd hear the difference between that and -100 dB. To put things in perspective, crosstalk of even the best phono cartridges is maybe -30dB at midband, and much worse at high and low frequencies.

-40dB at 20kHz could potentially be an issue for dolphins.

 

[watchnerd]

QED: the technical factors are of trivial importance

ERGO: balanced headphone cables for home use are for audio epeen

 

[Jorj]

I really like Tom's earlier explanation, and he mentioned something important. If you're of the mindset that subjective claims are valid, then read the manufacturer's drivel and spend your money and enjoy the cognitive dissonance!

That makes me wonder, given the rationalist\objectivist bent of this forum, how many of us are atheist? It would be interesting to do an anonymous poll and see if the numbers skew as I believe they probably do. No, I'm not starting a religious discussion, and will not engage further on the topic, just interested in the data and the correlation between various types of what I believe are examples of cognitive dissonance.

 

[Graph Feppar]

But doesnt the crosstalk caused by common ground also cause some nasty intermodulation distortion?

 

[SIY]

But doesnt the crosstalk caused by common ground also cause some nasty intermodulation distortion?

No. That requires nonlinearity. Wires are linear.

 

[Graph Feppar]

No. That requires nonlinearity. Wires are linear.

But tomchr wrote it does cause intermodulation.I imagine that the ground of one channel is changing depending on what the other channel is playing,I think that could cause intermodulation but I am not expert.

 

[SIY]

But tomchr wrote it does cause intermodulation.

I don't think he said that. If there is no nonlinearity, and for all intents and purposes wires have no nonlinearity, there's no intermodulation.

 

[sergeauckland]

But tomchr wrote it does cause intermodulation.I imagine that the ground of one channel is changing depending on what the other channel is playing,I think that could cause intermodulation but I am not expert.

It doesn't cause intermodulation because as the ground voltage is changed by the signal in one channel, all that happens is that the other channel sees this ground voltage change and so affects its own output. This is entirely a linear process, so there is no intermodulation, just crosstalk.

As has been shown above, the crosstalk change is miniscule and won't be in any way audible. I use 10m of thin three core mains cable as a headphone extension, with no ill effects whatsoever. With that relatively high amount of common impedance one would expect the crosstalk to be greatly increased, and yet, so what? As was mentioned above, pickups have barely -30dB of crosstalk, FM radio at best -40dB, so it's a complete non-issue.

 

[tomchr]

But tomchr wrote it does cause intermodulation.

Um. No. What I did write is that some designers claim that a balanced circuit is needed for good performance. I provided a few specific examples of why I disagree with that claim. Further, I elaborated that THD and IMD often go hand in hand. I.e. optimizing for one will often bring improvements in the other. None of this has anything to do with which connector is on the front panel of the amp. I perceived your original question as being rather broad and answered accordingly.

The crosstalk in a 1/4" jack is caused by the I*R drop across the shared ground. As March Audio pointed out above, the slight increase in crosstalk with the 1/4" connection is not a concern in the real world. You will also get some crosstalk from capacitive coupling between the conductors in the headphone cable. Headphones are relatively low-impedance loads, so this is also not a concern in the real world.

The fact that something is not a concern in the real world does not prevent customers, manufacturers and marketeers from believing that it is. I'm not implying any ill intent there. It's more of a self-fulfilling prophecy. Customer requests a feature. Manufacturer implements feature. Marketing guy touts the feature as the greatest thing since sliced bread. More customers now want that feature. Rinse. Repeat.

Tom

 

[AnalogSteph]

Essentially what shared return resistance does is attenuate L+R but not L-R. Given a mostly resistive driver impedance, this is equivalent to a more or less slight amount of stereo widening.

Let's assume a case of 16 ohm (mostly resistive) drivers and a cable with 1 ohm each in L, R and GND. (Not too outlandish a scenario - I found about an ohm of shared ground return on my Soundmagic E10 in-ears.)
L-R sees an attenuation of (16+16) ohms / (16 + 16 + 2) ohms = 16 / 17
L+R sees an attenuation of (16|||16) / [(16+1)||(16+1)+1] = 16 / 19
Hence, (L+R)/(L-R) = 17 / 19 ~= 0.895

I set stereo width in Rockbox to 0.90 to (mostly) compensate. (Reasons to use Rockbox, #285.)
Crosstalk in this example would come out as 1 / [(16+1)+1] = 1 / 18 = -25.1 dB.

Note that the severity of the effect diminishes as output series resistance increases. A hypothetical 16 ohm headphone with no cable resistance except for 1 ohm shared return would be at (L+R)/(L-R) = 8/9 ~= 0.889 (or -1.0 dB, crosstalk @ -24.6 dB), while adding 47 ohm to L and R gives 63/65 ~= 0.969 (-0.27 dB, crosstalk @ -36.1 dB). Assuming a largely ruler-flat impedance response, the latter would probably be sounding better...

As you can probably tell from my example, it takes some rather high return resistance to cause any real trouble, enough for the most likely troublemakers to be the cable and bad contact at the jack or plug. My rule of thumb for desired crosstalk has been -40 dB or better for years - and I mean at the headphone drivers, so an amp should be a good bit better than that to leave some slack for some shared cable return (if any). I'd say by the time you hit 50-60 dB @ 16 ohm, you should generally be safe. That is quite doable even in an unbalanced circuit with a 3.5mm jack (the O2 was somewhere around 60 dB if memory serves).

 

[Graph Feppar]

Um. No. What I did write is that some designers claim that a balanced circuit is needed for good performance. I provided a few specific examples of why I disagree with that claim. Further, I elaborated that THD and IMD often go hand in hand. I.e. optimizing for one will often bring improvements in the other. None of this has anything to do with which connector is on the front panel of the amp. I perceived your original question as being rather broad and answered accordingly.

The crosstalk in a 1/4" jack is caused by the I*R drop across the shared ground. As March Audio pointed out above, the slight increase in crosstalk with the 1/4" connection is not a concern in the real world. You will also get some crosstalk from capacitive coupling between the conductors in the headphone cable. Headphones are relatively low-impedance loads, so this is also not a concern in the real world.

The fact that something is not a concern in the real world does not prevent customers, manufacturers and marketeers from believing that it is. I'm not implying any ill intent there. It's more of a self-fulfilling prophecy. Customer requests a feature. Manufacturer implements feature. Marketing guy touts the feature as the greatest thing since sliced bread. More customers now want that feature. Rinse. Repeat.

Tom

I believe I read it from your post in the other balanced cable thread,I am too lazy to search for it lol
Well anyway,SIY already explained to me that no intermodulation is going on in the cables.