[xnor]

Yeah, there are always exceptions. I guess some headphones would even run fine form a current source.
 
While not really high-impedance, I remember a reviewer who complained that the HD555/595 was way too bassy - yes, he used it with a receiver, and returned the headphones. 390 ohms would give you a huge and relatively narrow bass beak. That driver construction is very efficient (more so than some in-ears) at the resonant frequency so the amps control over it is very important. But that's also more of an exception than the rule.

 

[proton007]

Since we're talking about the output impedance design, I'd like to bring something up.
 
The TDK IE800 IEM has an impedance of 10 Ohms @ 1KHz. Its one of the lowest I've seen on an IEM.
 
I wonder why it has been kept so low.
 
Going by the impedance bridging rule, anything below 1.25 Ohms output impedance would be too high. I'm glad I'm using it with my clip zip.

 

[xnor]

My guess would be that the crossover frequency is around 1 kHz and the crossover design causes a dip in the impedance.

 

[proton007]

Quote:

My guess would be that the crossover frequency is around 1 kHz and the crossover design causes a dip in the impedance.

 
And the volume also needs to be pumped higher in order to get them going.
 
Haven't been able to find a freq response or impedance graph. But just taking other dual driver IEMs as reference, it seems the impedance stays flat in the mids/lows and increases beyond 1-1.5k.

 

[Bill-P]

Quote:

Are you sure about that? -0.6 dB at 5 Hz doesn't seem to be audible. There can only be an "inverse" change if the output impedance is negative, which is possible but I've never seen such a headphone amp. This should also answer your over-damping question.

 
I did a lot more reading on impedance, Q-factor, mechanical damping, electrical damping, and etc... and now I realize over-damping is probably not possible for headphones with the amps we have on the market. It's good information.
 
But I think -0.5 dB is definitely audible. I can perceive channel imbalance at that level... and I verified it with an SPL meter.
 
 

Yeah but even new products are released with high output impedance. Those manufacturers just don't care enough about headphone-fidelity it seems.

 
From what I have gathered so far from this thread and some other articles I have read, I guess high output impedance has its own advantage to certain people:
 
1) It attenuates mid and high frequencies, so it has the same effect of "bringing bass forward". In essence, it means the same thing as "more bass impact, body, texture".
 
2) Depending on the characteristics of the mid and high frequencies, the attenuation caused by higher output impedance also serves to reduce glaring peaks (as with the DT880 and DT990) and helps "refine high frequencies". Due to some frequencies being less audible in the presentation, perception of "depth" may be improved.
 
3) Higher output impedance causes voltage to drop more at load, thus the amp must be able to deliver more volts, so a "lack of power" is notable as output impedance increases. If the amp fails to deliver the voltage, it'll clip and cause distortions.
 
4) If the headphone's impedance is sufficiently high, the damping factor would be high enough to prevent any distortion from ever taking place, thus higher impedance headphones would distort less with higher output impedance. An output impedance of close to 0 would distort the least regardless.
 
So I guess it's a trade-off between "close to the intended signature" and "how you want to listen to it". The latter seems to be more the case for most audiophiles, as I think... most of us would rather enjoy our gears than try to be as accurate as possible.

 

[xnor]

Quote:

 
I did a lot more reading on impedance, Q-factor, mechanical damping, electrical damping, and etc... and now I realize over-damping is probably not possible for headphones with the amps we have on the market. It's good information.
 
But I think -0.5 dB is definitely audible. I can perceive channel imbalance at that level... and I verified it with an SPL meter.

But we're not talking about channel imbalance here. We're talking about a DC blocking filter to prevent DC offset potentially harming headphones. Human hearing range starts somewhere around 20 Hz. The FR of the O2 is about 0.04 dB down at 20 Hz. This is not audible.
 
This is bass roll-off (Cowon X5L with 16 ohm load):

 
 
 
Quote:

From what I have gathered so far from this thread and some other articles I have read, I guess high output impedance has its own advantage to certain people:  
1) It attenuates mid and high frequencies, so it has the same effect of "bringing bass forward". In essence, it means the same thing as "more bass impact, body, texture".

Yup, similar to a bell/peaking EQ filter.
 
Quote:

2) Depending on the characteristics of the mid and high frequencies, the attenuation caused by higher output impedance also serves to reduce glaring peaks (as with the DT880 and DT990) and helps "refine high frequencies". Due to some frequencies being less audible in the presentation, perception of "depth" may be improved.

Overall, the peaks may get attenuated a bit, but peaks in the FR often go hand in hand with peaks in the impedance curve. And we know that peaks in the impedance curve cause a "boost" in the frequency response with high output impedance. As a result, some peaks actually become worse.
 
Anyway, I wouldn't call a low damping factor a refinement.
 
Quote:

3) Higher output impedance causes voltage to drop more at load, thus the amp must be able to deliver more volts, so a "lack of power" is notable as output impedance increases. If the amp fails to deliver the voltage, it'll clip and cause distortions.

Higher output impedance causes a higher portion of the voltage to drop across the internal resistance instead of the load (headphones). It depends on the implementation of the amp if it clips at full volume, or not. But generally we see an increase in distortion if we turn up the volume, and also if we increase the output impedance.
 
Quote:

4) If the headphone's impedance is sufficiently high, the damping factor would be high enough to prevent any distortion from ever taking place, thus higher impedance headphones would distort less with higher output impedance. An output impedance of close to 0 would distort the least regardless.  
So I guess it's a trade-off between "close to the intended signature" and "how you want to listen to it". The latter seems to be more the case for most audiophiles, as I think... most of us would rather enjoy our gears than try to be as accurate as possible.

I wouldn't say the heapdhone's impedance needs to be sufficiently high, but the output impedance needs to be sufficiently low. We've had power amps with very low output impedance for years, why not headphone amps? But yeah both linear and non-linear distortions will be reduced to a level where it's not audible anymore.
 
A zero ohm output impedance doesn't guarantee that you'll get close to the intended sound signature. Some headphones are designed for ~100 ohms, some manufacturers don't even care about this point. Some prefer headphone X with the extra bass, others hate it because of that. Things would be a lot easier with a zero ohm standard.

Anyway, my recommendation would be to get an accurate DAC and amp, headphones that are as close to the sound signature that you need/want/prefer and do some final tweaking with an EQ. That way there's no trade-off.

 

[Bill-P]

Quote:

But we're not talking about channel imbalance here. We're talking about a DC blocking filter to prevent DC offset potentially harming headphones. Human hearing range starts somewhere around 20 Hz. The FR of the O2 is about 0.04 dB down at 20 Hz. This is not audible.
 
This is bass roll-off (Cowon X5L with 16 ohm load):

 
I can actually "hear" the O2 having "less bass" than my Fiio E7, C421, Cmoy, and E11, all without bass boost. All out of the same source/DAC.
 
But that's subjective. I haven't done any measurement to really tell how much of a roll-off it actually is in raw numbers.
 
And the creator of the O2, NwAvGuy, also wrote on his blog admitting that the O2 has a slight bass roll-off.
 
 

Overall, the peaks may get attenuated a bit, but peaks in the FR often go hand in hand with peaks in the impedance curve. And we know that peaks in the impedance curve cause a "boost" in the frequency response with high output impedance. As a result, some peaks actually become worse.
 
Anyway, I wouldn't call a low damping factor a refinement.
 

 
I'd think the peaks that are most likely to be "boosted" are just in the low frequency range? Since higher frequencies are more likely to get attenuated, I don't think higher output impedance would cause any peak in high frequencies. Or am I missing something here?
 
And I was referring to the "loss" in treble peaks as a "refinement" rather than the lower damping factor. For really high impedance headphones (250 Ohm, 600 Ohm), any output impedance under 30 Ohm is already >8 for damping factor. I wouldn't consider that "low", but it may be just me.
 
 

Higher output impedance causes a higher portion of the voltage to drop across the internal resistance instead of the load (headphones). It depends on the implementation of the amp if it clips at full volume, or not. But generally we see an increase in distortion if we turn up the volume, and also if we increase the output impedance.

 
I guess that's more amp-design specific than output impedance specific. I have found that portable amps are more likely to distort than desktop, for one.
 
 

I wouldn't say the heapdhone's impedance needs to be sufficiently high, but the output impedance needs to be sufficiently low. We've had power amps with very low output impedance for years, why not headphone amps? But yeah both linear and non-linear distortions will be reduced to a level where it's not audible anymore.
 
A zero ohm output impedance doesn't guarantee that you'll get close to the intended sound signature. Some headphones are designed for ~100 ohms, some manufacturers don't even care about this point. Some prefer headphone X with the extra bass, others hate it because of that. Things would be a lot easier with a zero ohm standard. 

Anyway, my recommendation would be to get an accurate DAC and amp, headphones that are as close to the sound signature that you need/want/prefer and do some final tweaking with an EQ. That way there's no trade-off.

 
Well, since I specifically wrote "high impedance headphones", I guess it's more appropriate to consider the headphone. Agreed that low output impedance for headphone amps is desirable, but I haven't run into a situation where it becomes a real issue. Plus not all modern audiophile headphones are low impedance and susceptible to distortion due to high output impedance.
 
It may take a while.
 
Also after a while in the hobby, I've sworn off EQs. They create even more distortions to my ears.

 

[Eisenhower]

Louder sounds better to most people, within a limit. What mean by "lack of an amp" is really "lack of a dedicated amp". Some built-in amplifiers do not have much clean headroom, and can clip at high volumes. But many built-in amplifiers are quite sufficient.
Other than that, high impedance headphones should not effect the amp much, compared to low impedance headphones which can suck more current than the amp has to offer.

 

[xnor]

Quote:

 
I can actually "hear" the O2 having "less bass" than my Fiio E7, C421, Cmoy, and E11, all without bass boost. All out of the same source/DAC.
 
But that's subjective. I haven't done any measurement to really tell how much of a roll-off it actually is in raw numbers.
 
And the creator of the O2, NwAvGuy, also wrote on his blog admitting that the O2 has a slight bass roll-off.

Where did you read that? He wrote that
- output impedance "should be less than 2 ohms to provide the correct bass damping even with 16 ohm headphones"; the O2 has 0.5 ohms.
- frequency response should "be flat to within +/- 0.1 dB form 20 Hz to 20 kHz" and not use capacitor coupled outputs which cause bass roll-off. The O2 produces a flatter output than that and has no electrolytic caps in the signal path.
- "If you can hear -0.04 dB at 20 hz in a blind test I'll give you $10,000.00! The O2's "bass drop off" is WAY past being inaudible. Even ten times that, -0.4 dB at 20 hz is still very inaudible."
 
Anyway, this is a safety and even performance increasing feature. DC offset is bad.
A cmoy with gain 5 he measured produced 8 times the DC offset of the O2. The E7 is down -1 dB at 5 Hz, more than the O2. The E11 also has a DC blocking capacitor in the signal path.
 
 
 
Quote:

I'd think the peaks that are most likely to be "boosted" are just in the low frequency range? Since higher frequencies are more likely to get attenuated, I don't think higher output impedance would cause any peak in high frequencies. Or am I missing something here?  
And I was referring to the "loss" in treble peaks as a "refinement" rather than the lower damping factor. For really high impedance headphones (250 Ohm, 600 Ohm), any output impedance under 30 Ohm is already >8 for damping factor. I wouldn't consider that "low", but it may be just me.

Yeah, overall mids and treble get attenuated, but some headphone drivers have impedance peaks where there are also peaks in the frequency response. These are of course a lot smaller than the peak at the resonant frequency, but still, they're there.
The boost doesn't have to be around 100 Hz, it's just that many dynamic headphones happen to have the resonant frequency roughly around there. There are exceptions. Some headphones have flat impedance that is rising towards 20 kHz ->  boosted upper treble. Others are pretty much flat across the entire 20 Hz - 20 kHz range (mostly non-dynamic headphone drivers).
 
A loss in treble is only a refinement if the headphones produce too much treble to begin with, isn't it? I mean, if you're happy with the stock frequency response a peak in the bass would be a change for the worse.
 
 
Quote:

I guess that's more amp-design specific than output impedance specific. I have found that portable amps are more likely to distort than desktop, for one.

I didn't mean different amps. I meant if you take one amp and gradually increase the output impedance, for example by adding resistors.
 
That's actually another reason why I like low output impedance amps. You usually cannot reduce output impedance without making the amp unstable or whatnot. But you can always add resistors if you really need to.
 
Quote:

Well, since I specifically wrote "high impedance headphones", I guess it's more appropriate to consider the headphone. Agreed that low output impedance for headphone amps is desirable, but I haven't run into a situation where it becomes a real issue. Plus not all modern audiophile headphones are low impedance and susceptible to distortion due to high output impedance.

I've run into the problem. I've seen others run into the problem. I've seen headphone jacks that range to about 470 ohms output impedance. Luckily the bass "boost" doesn't increase linearly with rising output impedance, i.e. doubling the output impedance doesn't double the problem/"boost".
 
Quote:

Also after a while in the hobby, I've sworn off EQs. They create even more distortions to my ears.

I've also had bad experiences with "broken" headphone amps. Doesn't mean I won't use properly implemented ones or build my own.

 
I don't know what music you listen to but there are people that would never use an EQ in their playback chain but listen to genres where the music is horribly compressed, EQ'd and processed in ways they couldn't even imagine. Showing them that speakers/headphone drivers, amps and DACs produce more nonlinear distortion than a properly implemented EQ causes quite some cognitive dissonance.

 

[Bill-P]

Where did you read that? He wrote that
- output impedance "should be less than 2 ohms to provide the correct bass damping even with 16 ohm headphones"; the O2 has 0.5 ohms.
- frequency response should "be flat to within +/- 0.1 dB form 20 Hz to 20 kHz" and not use capacitor coupled outputs which cause bass roll-off. The O2 produces a flatter output than that and has no electrolytic caps in the signal path.
- "If you can hear -0.04 dB at 20 hz in a blind test I'll give you $10,000.00! The O2's "bass drop off" is WAY past being inaudible. Even ten times that, -0.4 dB at 20 hz is still very inaudible."

 
Yeah, but low output impedance, again, is only important for low impedance headphones. For something with over 120 Ohm impedance, I don't think the output impedance would change much (provided it's still significantly smaller than the impedance of the headphone itself), but... feel free to correct me on that.
 
Here is the frequency response graph of O2 directly from his website. It rolls off more below 20 Hz.
 

 
The problem is not that I can't "hear" 20 Hz, but that most headphones aren't capable of producing 20 Hz and below. Some headphones, like the DT880 and DT990 are rated for 5Hz to way past 20KHz, and I genuinely believe that I can at least "perceive" under 20 Hz, if I can't "hear" them. This is getting into subjective realm again, but I think it's worth noting that the O2 is not completely flat since it starts rolling off from 20 Hz downward.
 
I admit -0.04 dB at 20Hz won't be noticeable, but try -0.5 dB at 5 Hz on a headphone that can reproduce it.
 

A loss in treble is only a refinement if the headphones produce too much treble to begin with, isn't it? I mean, if you're happy with the stock frequency response a peak in the bass would be a change for the worse.

 
 
Yeah, and I'd assume you haven't heard about DT990 + DT880 producing too much treble? Common "problem". I'm mostly just commenting on that.
 

I didn't mean different amps. I meant if you take one amp and gradually increase the output impedance, for example by adding resistors.
 
That's actually another reason why I like low output impedance amps. You usually cannot reduce output impedance without making the amp unstable or whatnot. But you can always add resistors if you really need to.

 
Yes, but if you are using a high impedance headphone, again, something with more than 120 Ohm impedance, usually in the range of 250 Ohm, 600 Ohm, or even 2000 Ohm, would a higher Ohm output really make that much a difference? Provided it's not astronomically high like 470 Ohm.
 

I've also had bad experiences with "broken" headphone amps. Doesn't mean I won't use properly implemented ones or build my own. 

 
I don't know what music you listen to but there are people that would never use an EQ in their playback chain but listen to genres where the music is horribly compressed, EQ'd and processed in ways they couldn't even imagine. Showing them that speakers/headphone drivers, amps and DACs produce more nonlinear distortion than a properly implemented EQ causes quite some cognitive dissonance. 

 
I'm not a purist... so I'm not entirely against EQ, but from my experience, EQ is only good if you are "reducing" things rather than "boosting" them.
 
And it doesn't have anything to do with the amp or DAC. I've often found that the distortion is caused by the headphone itself. Some headphones are simply not capable of producing sound a different way.
 
For instance, I tried to EQ the DT990 Pro, and no matter what I did, I can never get its midrange to be lusher, smoother, less grainy, warm, or engaging. And I can never make it retrieve and present as much detail in high frequency as the DT880 does. The headphone just wasn't capable of delivering that sort of performance.
 
If you own headphones that can properly reproduce every frequency linearly... as in... a perfectly "flat" and "linear" headphone, the "ideal" headphone, then yeah, I can see in that case how EQ may help.

 

[liamstrain]

Quote:

I admit -0.04 dB at 20Hz won't be noticeable, but try -0.5 dB at 5 Hz on a headphone that can reproduce it.

 
 
 
A half dB at a frequency that no human can hear (and when strong enough to "perceive" is damaging to internal organs) is a non-issue. Entirely. 

 

[Bill-P]

Quote:

A half dB at a frequency that no human can hear (and when strong enough to "perceive" is damaging to internal organs) is a non-issue. Entirely. 

 
Humans can definitely "perceive" under 20 Hz given that it's actually reproduced in the headphone... and that the level is high enough.
 
http://www.noiseandhealth.org/article.asp?issn=1463-1741;year=2004;volume=6;issue=23;spage=37;epage=57;aulast=Moller
 
And it won't cause damage to internal organs. It's not like they're playing back 20KHz at 200 dB...

 

[Steve Eddy]

What on earth are people listening to that has content down to 5Hz?
 
se

 

[stv014]

It would be best to try some experiments on music that you think is audibly affected by the AC coupling in the O2 (can you upload/post links to any samples ?). These include:
- simulating the filter in the O2, and comparing the result against the unfiltered audio using the foobar ABX plugin
- filtering out everything above 20 Hz (also other frequencies like 10 Hz and 5 Hz), and listening to what is left
 
Note that while your headphones may have been specified by the manufacturer to have a frequency response of something like 5-35000 Hz, these specs are usually very optimistic, and the headphones themselves roll off (and also distort) infrasonic audio much more than the O2.

 

[xnor]

Quote:

 
Here is the frequency response graph of O2 directly from his website. It rolls off more below 20 Hz.

That's what a DC blocker does.
 
Quote:

The problem is not that I can't "hear" 20 Hz, but that most headphones aren't capable of producing 20 Hz and below. Some headphones, like the DT880 and DT990 are rated for 5Hz to way past 20KHz, and I genuinely believe that I can at least "perceive" under 20 Hz, if I can't "hear" them.

Yes and thats a big marketing joke.
 
 
Quote:

This is getting into subjective realm again, but I think it's worth noting that the O2 is not completely flat since it starts rolling off from 20 Hz downward.

This has nothing to do with subjectivity. It is flat (+/- 0.1 dB) from below 20 Hz to far beyond 20 kHz which is what counts. All the other amps you listed aren't dead flat down to 5 Hz either, not that it matters.
 
 
Quote:

I admit -0.04 dB at 20Hz won't be noticeable, but try -0.5 dB at 5 Hz on a headphone that can reproduce it.

This is a joke, right? What headphone do you propose? You do realize that the hearing threshold for such low frequencies is far beyond any headphones' reproduction capabilities?
 
We're talking ~110 dB SPL here to reach the hearing threshold. You know what's about 0 dB SPL at 1 kHz.
 
Also, what noises are you listening to? This can't be music created for humans.
 
 
Quote:

Yeah, and I'd assume you haven't heard about DT990 + DT880 producing too much treble? Common "problem". I'm mostly just commenting on that.

I've heard them producing a nasty treble peak which can be cut with a bell EQ filter, not with high output impedance because the peak would still stand out.
 
 
Quote:

Yes, but if you are using a high impedance headphone, again, something with more than 120 Ohm impedance, usually in the range of 250 Ohm, 600 Ohm, or even 2000 Ohm, would a higher Ohm output really make that much a difference? Provided it's not astronomically high like 470 Ohm.

That's been answered before. 1/8th is a good rule of thumb.
 
Quote:

If you own headphones that can properly reproduce every frequency linearly... as in... a perfectly "flat" and "linear" headphone, the "ideal" headphone, then yeah, I can see in that case how EQ may help.

It also can help a lot with non-ideal headphones. I take a small increase (boost) or decrease (cut) in distortion over nasty treble peaks, bass peaks or dips any time.