Headphone Outputs: Build-Out vs. No Build-Out

[mediatechnology]

Sound and Vision magazine http://www.soundandvisionmag.com just ran a review of "6 Top Over-Ear Models."

Though the February/March print issue content has yet to be published to the web with the full test results, I found some useful facts that I thought I'd share.

Sony MDR-Z1000: 28 Ohms avg. 31 max @ 71 Hz.
Grado RS2i: 32 Ohms avg. 45 at 68 Hz.
HiFiMan HE-300: 58 Ohms nominal 94 max @ 68 Hz.
Audio-Technica ATH-AD900: 35 Ohms avg. 45 @90 Hz.
BeyerDynamic DT-990 is available in 32, 250 and 600 Ohms. 32 Ohm tested: 32 Ohm avg. 43 Ohms @ 100 Hz.

They used a Rane HC6 which appears to be a "0" Ohm output and then added 75 Ohm build-out to check for impedance peaks with each model.


Rane HC6S Headphone Amp Output

The role of R8 interests me.
Is it to lower THD or to improve overload recovery with the JRC 4580?

[JR.]

I used lots of 4560s at Peavey and they were decent mid range performers.

The 1k may be some obscure stability improvement (I vaguely recall some opamp specific tricks like that.. the R values look wrong for DC correction. The effect of source R on stability is not simple.

It looks like they are trying to reduce class B distortion by making the opamp supply a bunch of the load current (+/- 33mA or so) before the transistors start helping. The opamp is class AB and rated for 50 mA.

I would be tempted to use a smaller C23, a larger c22, and larger R50, but I would generally add some class A bias to the output transistors.

Note: I didn't do any of that in the simple buffer I used to drive the speakers in my first generation drum tuner (to drive 16 ohms to a few volts). My circuit looked pretty similar this one but I used a different parts and values all around. For my application distortion was not much on issue below a few percent.

JR

[mediatechnology]

The 1k may be some obscure stability improvement (I vaguely recall some opamp specific tricks like that.. the R values look wrong for DC correction. The effect of source R on stability is not simple.

That's what I'm thinking: It was for some stability improvement or to prevent excess current flow and overload recovery issues when the input CM range was exceeded.

the R values look wrong for DC correction.

And that would generally make it wrong for impedance-balancing the input and feedback impedances for THD improvement.
I think I managed to get some THD improvement out of an OPA2604 using a 1K in the input.
But the OPA2604 is a BIFET and the JRC4580 PNP BJT.
I liked the 4560s. Otari used those a lot too and they're a frequent staple in Soundcards.

[ricardo]

Sound and Vision magazine http://www.soundandvisionmag.com just ran a review of "6 Top Over-Ear Models."

Did they do impedance curves?

Rane HC6S Headphone Amp Output
The role of R8 interests me.
Is it to lower THD or to improve overload recovery with the JRC 4580?

R8 is to reduce breakthrough when the pot is at 0. Common in 70' & 80's Calrec stuff.

Yes John. I know this is symptomatic of sewage (in this case, the pot current) down the clean water pipes but sometimes on a big complicated module, it can't be helped. Especially near high current stages like phone amps & main out and evil ground planes. Mea maxima culpa.
________________
On the original subject of build outs, I'm in 2 minds. Looks like good headphones will benefit (slightly) from current drive while bad ones (which buzz / rattle) may be worse.

The only clear cut case is when a phone has been designed for a particular Ro and the response is a lot different with Ro=0. In which case, the circuits in the other thread on defined Ro may be appropriate.

[mediatechnology]

Did they do impedance curves?

Yes. Apparently those measurement curves are to be posted in the future online edition.
They gave the peak values and frequency both with and without 75 Ohm build-outs for some models in the print version.

Forgot the sixth model:

Sennheiser HD598: 60 Ohms avg. 253 Ohms @ 97 Hz. 3.3 dB bump at 97 Hz with 75R build-out.

R8 is to reduce breakthrough when the pot is at 0.

Sorry, don't follow. I believe you but am just clueless.
How does 1K in series with gig-Ohms reduce breakthrough?
What is break-through? Crosstalk?
I realize that there's end-stop and wiper resistance but what does the 1K do to help?
Reduce capacitance effects in the pot?

[ricardo]

R8 is to reduce breakthrough when the pot is at 0.

Sorry, don't follow. I believe you but am just clueless.
How does 1K in series with gig-Ohms reduce breakthrough?
What is break-through? Crosstalk?
I realize that there's end-stop and wiper resistance but what does the 1K do to help?
Reduce capacitance effects in the pot?

I remember having this conversation with Ken Farrar, Tech Dir. and my electronics mentor, more than 30 yrs ago and saying almost the same things as you.

Breakthrough is what you hear when the pot/fader is fully "off". Mainly HF so I think it does reduce the capacitance effects of the pot but this is really long ago. It appears as a hand written comment in my notes of various Calrec circuits.
_______________

BTW, anyone with recommendations for good/accurate/comfortable/cheap headphones? As a beach bum, my stereo is also my speaker source for measuring mikes. So it's rarely set up to listen to music. All the ones I know & like are now Unobtainium.

[JR.]

Still doesn't make any sense to me... but whatever...

JR

[mediatechnology]

Glad I'm not alone.

I've done the same thing (R8) for other reasons and Rane may have too.
Having some 4580 samples to check the CM overdrive performance (or stability) might answer the question as to why Rane did it.
I wonder if the 4580 polarity reverses...

[emrr]

I think I can relate to this 'break-through' description that ricardo makes, but only from a 'field observation' point with certain gear. Have never investigated the tech.

[JR.]

Great we have confirmation... but still no description of what "break through" means or sounds like.

Burst of instability, crosstalk from unrelated signals, what?

We were speaking in the context of a non inverting opamp... I can image sundry possibilities for other topologies, not so much for that one.

JR