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

Where we discuss new analog design ideas for Pro Audio and modern spins on vintage ones.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

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

Post by mediatechnology »

And the winner is: No Build-Out.

I recently updated the headphone amplifier post I did at DIY Audio that's based on the THAT1646 with current boost transistors.
People over there were wanting to use the DRV134 and it requires a simple mod.

One of the forum members asked about removing the build-out resistor.
At about the same time I received an e-mail directing me to John Siau's recent whitepapers at Benchmark:


Note: Updated broken links 12/18.

https://benchmarkmedia.com/blogs/applic ... ifications
https://benchmarkmedia.com/blogs/applic ... -amplifier
https://benchmarkmedia.com/blogs/applic ... ers-part-2

I just finished duplicating some of John's work and found another reason to eliminate the build-out resistor in headphone amplifiers that will allow it.

The DCAO current-boosted THAT1646 is an open-loop Class-A output and can run into capacitive and headphone loads with no build-out.
Here's the RMAA comparison:


Image
Dual Class-A Output Headphone Driver both with and without build-outs driving Sony MDR-7506 headphones.
The 0R figures are pretty darn close to the Soundcard's loopback performance.


The improvements in distortion performance driving an actual headphone load both with and without build-out resistors clearly points to not using them.

John and others have also documented the frequency response errors.

Image
Dual Class-A Output Headphone Driver frequency response both with and without build-outs driving Sony MDR-7506 headphones.

What I didn't expect to find was the huge improvement in crosstalk performance.

Image
Dual Class-A Output Headphone Driver crosstalk both with and without build-outs driving Sony MDR-7506 headphones.

My working theory is that the lower output impedance reduces the inter-conductor induced currents between the "tip" and "ring" conductors in the headphone cable itself and thereby reduces crosstalk.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

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

Post by JR. »

Crosstalk at -60dB for stereo program is not a big deal. A 1/2 dB rise is barely audible in the midrange, and surely inaudible at frequency extremes.

The 5 dB higher noise level could be audible and the 20 dB worse IMD is interesting while .2% is down >50dB so fairly difficult to hear.

Of course better is always better, as long as it is stable driving electrostatic cans...(capacitive?) it's all good.

JR

PS: Sorry I didn't read the links I ASSume these measurements are of the voltage driving the cans, and ignoring the transfer function of the headphone transducers?
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

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

Post by mediatechnology »

Sorry I didn't read the links I ASSume these measurements are of the voltage driving the cans, and ignoring the transfer function of the headphone transducers?
Yes.
Crosstalk at -60dB for stereo program is not a big deal. A 1/2 dB rise is barely audible in the midrange, and surely inaudible at frequency extremes.

My phones are rated at 63 Ohms. It might be far worse with lower-impedance phones so I wouldn't generalize just based on these 'phones.
The MDR-7506 don't have a reputation to be all that difficult to drive. Others do.

John S. provides this fairly compelling example:

Image
Benchmark 0 Ohm Headphone Amplifier Figure 2

I run out of CPU power doing Visual Analyzer LF FFTs around 32 Kpts.
My tests revealed on 100 Hz THD-3 about 20 dB reduction with the 63 Ohm Sony MDR-7506. At 100 Hz I can't see THD-2 all that well.

If you can reduce the Zout by simple means, such as lowering or eliminating a resistor or placing it within a feedback loop, it's probably worth doing in a headphone amp.

WRT load capacitance: On the DCAO I can drive 1 uF with a 10 kHz squarewave at 10V p-p. It doesn't like driving that amount of current, but it will and remain stable.
100 nF is no problem.

I'm not sure the extreme "far more is better" on the damping factor holds since the headphone cable resistance appears in series and limits it several orders of magnitude over what is available at the amp output.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

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

Post by JR. »

Not to be more contrarian than usual he is talking about distortion spurs down at -70dB.. for chuckles how about a distortion overlay for the transducer?

JR
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

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

Post by mediatechnology »

Not to be more contrarian than usual he is talking about distortion spurs down at -70dB.. for chuckles how about a distortion overlay for the transducer?
Ah, there's the rub...

He does admit that acoustic measurements may show more distortion...
All measurements shown in this paper are taken at the headphone inputs. These are not measurements of the headphone acoustic output. Acoustic measurements may show significantly higher levels of distortion. But, the acoustic measurements can never show less distortion than the measurements taken at the headphone inputs.
Still, if it only requires making a 30 Ohm resistor a 0 Ohm resistor (open loop example) or putting the same 30 Ohm resistor inside a feedback loop (closed loop example) then why not do it?
I realize in the closed loop example another capacitor may be needed.

And, in doing one of the above, the acoustic performance of the mechanical system might be improved.

Though it still might be orders of magnitude higher than the electrical even after that.
But why not do it anyway as long as there is the same degree of short-circuit protection?

It's at zero cost or almost so.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

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

Post by JR. »

Better is always better,,, Driving magnetic systems like transformers IIRC benefit from lower source impedance, perhaps a transducer's magnetic system also benefits. My distortion overlay is not completely fair, as some distortions are more audible than others... I recall hearing modest amounts of specific kinds of high order distortion over speakers that were generally making much larger amounts of low order distortion.

I remain nervous about 0 ohm build outs, but that's just me... there will always be some series R and L.

JR
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

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

Post by mediatechnology »

Better is always better,,, Driving magnetic systems like transformers IIRC benefit from lower source impedance, perhaps a transducer's magnetic system also benefits. My distortion overlay is not completely fair, as some distortions are more audible than others... I recall hearing modest amounts of specific kinds of high order distortion over speakers that were generally making much larger amounts of low order distortion.
Well, I think many of the damping factor arguments for power amplifiers - to a point - apply here.
I remain nervous about 0 ohm build outs, but that's just me... there will always be some series R and L.
I would prefer some.
I'd like to know where the diminishing return begins.
Once it drops below the copper resistance of the cable the available damping factor is going to reach a maximum.
I'm figurin' that the dual Class-A is going to be about 2 Ohms Zout.
For a damping factor of 100 with a true 0 Ohm source, the total cable resistance and interconnect would have to be 0.63 Ohms.
I don't see that happening.
ricardo
Posts: 262
Joined: Sun Apr 10, 2011 7:24 am

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

Post by ricardo »

JR. wrote:Better is always better,,, Driving magnetic systems like transformers IIRC benefit from lower source impedance, perhaps a transducer's magnetic system also benefits.
Actually moving coil transducers have quite a bit less distortion with current drive. 1k makes a big difference into headphones.

http://www.aes.org/e-lib/browse.cfm?elib=6099

I've done a lot of work on this and related stuff as part of my Powered Integrated Super Sub technology.

"Ribbon" headphones benefit less as most(?) are inherently low distortion with voltage drive. Wharfedale (Rank HiFi) held several patents on their Isodynamic technology. I used the LEAK (Rank) 3000 headphones for some Blind Listening Tests when I was worried about distortion. These had less than 0.1% THD to well above 100dB spl.

But the important question is "what Ro were your headphones designed for?" In the 70's & 80's, most domestic gear used a few 100R in series with the speaker output. This worked well with then common 8R phones up to the Sennheisser HD414 @ 2K.

I've no idea what makers design for these days. If you have a different Ro, the response will not be as the designer intended. IEC 61938 specify 5V @ 120R which I think is too high Ro

There's a much better argument for variable Ro in headphone amps than for speaker amps.

> My working theory is that the lower output impedance reduces the inter-conductor induced currents between the "tip" and "ring" conductors in the headphone cable itself and thereby reduces crosstalk.

Lower Ro reduces the common voltage induced in the Shield wire. You can reduce this by having separate earth wires for each side; 4 wire connection. The common shield resistance sets a lower limit on xtalk. I think Sennheisers had separate leads to the plug at one time. The xtalk is "out of phase" so increases the "stereo" component.

More "depth, ambience, musicality bla bla ... " :lol:
Last edited by ricardo on Wed Jan 11, 2012 8:15 pm, edited 1 time in total.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

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

Post by mediatechnology »

I have often wondered about constant-current drive for phones.

We used it driving spring reverb transducers at MicMix.

WRT Ro some common values are 32 and 63 Ohms. I've seen ranges from 10-200 Ohms. There may be one model at 300R.

AKG re-designed the "most-highest" AKG-240 which was 500-600 Ohms in the old days. IIRC they now come in at around 50R?
ricardo
Posts: 262
Joined: Sun Apr 10, 2011 7:24 am

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

Post by ricardo »

I should also point out that current drive (bigger Build-Outs) reduce xtalk AT the phone transducers; ie where it matters. If you measure at the amp just after the Build-0uts, before the leads & shields, then bigger Build-Outs will show increased xtalk.

But the xtalk introduced by Ro=0R gives more depth, ambience, musicality etc. :lol:
Post Reply