Headphone Cable Questions

[Joe Garfield]

I want to make a new cable for AKG K712 which has a trs to 3-pin mini XLR. I’ve noticed that typical trs, like headphone extensions, use 2-conductor cable with the shield acting as 3rd conductor (ground in this case.) I’ve also noticed people using 4-core microphone cable.

Using a shield as the ground would result in a different resistance/impedance/capacitance than the positive signal wires. Doesn’t that matter?
Initially I didn’t like the idea because grounding a shield at both ends introduces a ground loop, but I guess in this application there isn’t a second ground path, but you’re still intentionally making an antenna and using it as part of the ac signal path.

Quad core is meant for balanced cable situations - two + and two - wound together so that noise is symmetrical between + and - and therefore can be better handled by balancing. This doesn’t seem to make sense for 2-channel headphones. You could use the two pairs as R+L but then you lose the claimed benefit Of low noise. If used for 2-channel with common ground, you’d end up with ground loops. More interestingly it seems like more room for jitter (time based problems) since partial signals could arrive at different times/levels if the paths are not identical.

However it’s very difficult to find 3 conductor shielded cable in the proper diameters for headphones.

Mogami’s headphone extension is really a balanced mono extension. Hosa’s headphone extension is 2-core with shield as the ground, and they even tell you the shield is less oxygen free than the cores. But audio is AC so why does that make sense?

[mcandmar]

Try Canare L-4E6S, has two pairs and a shield inside, and more importantly its a very soft and flexible cable.  http://www.canare.com/ProductItemDisplay.aspx?productItemID=53

For the XLR to TRS adapters i just tied the two returns and shield connections together to the outer of the connector.  As a point of reference, the attached photo is a stock HD800 cable where Sennheiser have done exactly that onto a TRS plug.

I don't believe you need a shield for headphones, or can hear any difference with or without so i really wouldn't worry about it.

[Joe Garfield]

For dual entry headphones like the HD800, it at least makes sense to use quad core wire to reduce cross talk and have 2 separate circuits for the longest possible distance. In that case it's like a star ground as opposed to a looped common ground.

I actually already ordered Canare 4E5C - before reading and thinking about this, since that's what people seem to use and is similar to Mogami's "Headphone Extension". Even Grado's headphone extensions use TRS connectors with 4-core wire. It just seems that wiring 2 cores plus a shield all to ground at both ends makes a lot of chance for ground loops and seems pretty unnecessary.

It looks like Gotham suggests that using the shield grounded at one end, plus one conductor grounded at one end and spiraled over the rest, will help with noise reduction (essentially 2 shields directionally grounded.) But I thought that would trade capacitance (tone) for noise.

I just want to understand what is electrically ideal, and why doesn't anyone seem to be doing it.

[Paul Birkeland]

It just seems that wiring 2 cores plus a shield all to ground at both ends makes a lot of chance for ground loops and seems pretty unnecessary.

Well, I wouldn't recommend using shielded cabling for headphones.  If the cable you use has a shield, you could certainly leave it disconnected.  If you are terminating it to a 4 pin XLR, you can terminate the shield to the shield pin on the XLR. 

I'm not seeing how you could end up with a ground loop.  As far as cross talk goes, use Ohm's law to quantify the impedance of the ground conductor and calculate the AC voltage appearing acorss that based on the recommended power levels of your headphones, then calculate the magnitude of that signal vs. the likely fundamental.  Is it more than 100dB down?  You are very unlikely to hear that!

[Joe Garfield]

(-ve) -—-—->==============<————
(L+ve)———>———————————-<————
(L-ve)———->—————————-——<————

Assuming the ——— are single conductors and === is 2 parallel conductors, would that not make a ground loop?

[Paul Birkeland]

Nope, as they are terminated to different places at the driver end of the cable.

[Joe Garfield]

With TRS at one end and 3-pin XLR at headphone end, two (-ve) wires would be terminated together at the sleeve, and also together at Pin 3. Am I missing something?

[Paul Birkeland]

Ah, now I see what you're saying.

You could certainly only use one ground connector if you'd like.  Depending on the amp you plan to use, it may make sense to connect a cable shield you have to ground (I would do this in our amps), but also maybe not (in amps with "active" ground connections, amps that are advertised as "fully balanced" or "differential output).

It is easy enough to hook up your phone to your headphone amp and play a loud 60Hz tone into the amp, then take voltage measurements at the output of your proposed cable to look for issues.  I suspect you will have a very tough time measuring much in the way of a difference between these arrangements.

[Joe Garfield]

Thanks Paul. This is for my AKGs which I don't think will play well with the Crack I'm building, but there's a good chance I'll end up with a S.E.X. In the near term the amp for these cans will be Schiit Vali 2 (tube hybrid, discrete channels but nothing fancy.)

What are your thoughts about using a 3-core wire with one of the cores as (-), versus a 2-core wire and using the shield as (-)?

It seems to me the ideal setup is 3-core with an exterior shield grounded at the amplifier end. Since 3-core wire is tough to find in headphone size/weight/flexibility, I'm wondering what the next best thing might be.

[Paul Birkeland]

What are your thoughts about using a 3-core wire with one of the cores as (-), versus a 2-core wire and using the shield as (-)?

I don't think you'll notice much, if any difference.

It seems to me the ideal setup is 3-core with an exterior shield grounded at the amplifier end. Since 3-core wire is tough to find in headphone size/weight/flexibility, I'm wondering what the next best thing might be.

Get stranded wire and braid it to make your own 3 conductor cable, then cover it with tinned braided sleeving.

[Joe Garfield]

I just built my first Bottlehead amp, and all of a sudden using a headphone extension cable makes a difference. I must be on Crack!

Seriously, with a good DAC, a ‘decent’ amp, and HD650s I tried to listen for a difference with the extension and never could. With Crack it’s more than obvious. Unbelievable!

When you say to braid stranded wire and cover it with a tinned shield, are you suggesting to ground the shield at (-) on the amp, or should I put a TRRS jack in these amps and ground the shield to the chassis?

[Paul Birkeland]

I wouldn't shield a headphone cable at all personally.  The levels at the output of a headphone amp are quite high and noise pickup from a headphone cable is minimal, if not completely absent.

The comment about braiding three wires and covering that with tinned braided sleeving was in reference to what you considered to be an ideal headphone cable. 

[Joe Garfield]

Thanks, that’s much easier =)

[Joe Garfield]

I ordered a custom 20ft Mogami Star quad Extension from Performance Audio last night, and it’s already been shipped. For little more than the cost of components I got the whole package, and always have the opportunity to take it apart and modify it later.

Thanks for helping me think through all this.

[Joe Garfield]

Well unfortunately, out if the box there a loss of quality. The track I use to evaluate gear is a live concert. There a loss of “room” or air, and the high hat loses this delicate, 3d sparkley decay. It’s subtle but noticeable.

Edit: it might have something to do with capacitance. It's a 20ft extension using Mogami Starquad cable. They claim 23 + 3 + 4 pF between conductor pairs/shield, but then say 50 + 54 pF conductors/shield for balanced quad configuration. I don't know how that adds up but it looks to me like possible 3x capacitance depending on how it's wired.