JaZZ
Headphoneus Supremus
To take up a topic broached in this thread...
I guess there are very few people which already have tried to renounce or bypass their headphone amp and connect their headphones directly to their CD player or DAC. Given that the latter's analog output impedance is often clearly below 100 ohm and provides at least the standardized 2 volt, in some cases even considerably more, it must clearly be able to drive any dynamic headphone. And it seems very reasonable to use the shortest signal path possible, which necessarily will provide a virtually unaltered signal to feed the headphones with – in contrast to the headamp operation, which puts some – actually superfluous – amplifier stages in the signal path. Well, all that's necessary to do so is a 250-1000 ohm stereo potentiometer which acts as a pure attenuator (more specifically: as an adjustable bleeder/voltage divider) for loudness control.
Does this configuration have any downsides? Yes: there is a restriction of the source device's output impedance in order not to influence the system's frequency response (and ouput level) too much. I guess anything below 150 ohm will work fine. Besides, this output impedance and the potentiometer's impedance together form varying (depending on the attenuation adjustment) serial and parallel resistances. They cause an electrical interaction with the headphone's impedance curve, so that it may slightly increase the area around its bass resonance and possibly towards highest frequencies, due to the voice coil's inductivity. This happens in the same range as with most tube amps with output impedances between 100 and 300 ohm, while solid-state amps have low impedances of about 20 ohm.
I did compare the affection by serial resistances with my Corda HA-1 which has the extra 120 ohm output jack. It really changed the sound of all my headphones a bit more or less – apart from the volume reduction. The changes restricted themselves to (very) slight balance shifts, mostly in the form of an emphasizing of bass and upper treble. Actually one would expect that low-impedance cans like the Grados are more affected than the 300-ohm HD 600, but that's not the case – the reason is the much flatter impedance curve of the Grado models. And there's no loss of bass precision due to a reduced damping factor, as it would appear with bass speakers. This can clearly be reproduced with the Etymotic ER-4 P to S conversion proceeding or generally by watching the ER-4 system, with its 20 to 100 ohm serial resistors added to the plain 7 ohms of its driver's voice coil to provide a practical impedance – obviously without any harm to the sound.
Some information about the equipment used for this comparison:
– My Sennheiser HD 600 is equipped with a home-made cable made of capillary 0.028 mm magnet wires, plaited to four strands of 400 each; the (dust protection) foam pads which originally cover the drivers are removed. Both modifications move the sound towards considerably greater clarity, forwardness and airiness (no «veiled» sound anymore...).
– My Etymotic ER-4S are equipped with a Fixup replacement cable, with pleasing sound and clearly reduced microphonics compared to the stock cable; originally with two 50 ohm resistors (also called ER-«X»), I reverted to 100 ohm («S»), which has turned out as the better match with my portable Archos MP3 player and performs equally with my home setup. Their foam tips are made of pierced noise protection plugs, impregnated with contact adhesive for air-tight seal and thus stronger bass, leaving the transducer stems flush with the foam surface or even slightly protruding; this necessitates the protection of the thus exposed filter by some nylon gauze folded around the stem. The original filter (membrane) is replaced by some lambswool filaments for an even smoother, less resonant treble reproduction. Additionally this provides the possibility to fine-tune the sound by dosing the lambswool density.
– My Grado SR-225 and SR-325 are dressed with some modded yellow Sennheiser pads; in order not to lose too much treble thereby, I have bended them in such a way that they have a very lose fit on my head, thus low pressure to my ears.
– My AKG K 501 is virtually in original state, except for a black velvet coating of the driver's ring-shaped bare plastic surface, meant to reduce reflections inside the earpieces/-pads. With all of them I can clearly identify the sonic character of the respective amp used.
– My Musical Fidelity X-Cans (primary version), a well-known tube/solid-state hybrid amp, is in original state.
– My Brocksieper Earmax Pro presently is equipped with one Telefunken E801S in the input stage and two Siemens ECC188 in the output stage.
– My Corda Blue (Meier Audio Corda Headamp-1, modified by Head-Fi's famous KurtW) is my solid-state alternative.
– My passive headphone «amp» consists of an «airplane»-plywood housing and a logarithmic 500-ohm (no-name) potentiometer.
– My new Bel Canto DAC2 is a 24 bit/192 kHz upsampling DAC; output impedance: 20 ohm, output voltage: 2 volt.
– Interconnects: all IC cables are of the same kind as described in the HD 600 paragraph: made of ultra-thin magnet wire, i.e. individually isolated 0.028 mm copper wires, about 400 per conductor. This cable design has turned out – to my ears – to provide the best high-frequency extension, a super-clear upper end, assumedly thanks to a highly reduced skin effect...
I can't pretend to have achieved an exactly equal loudness level by simply adjusting it by naked ears, but I have done it with all the care that I could apply, and the various sessions with individual settings finally have effected consistent results.
With my DAC's line-out signal feeding a headphone directly, just attenuated by some resistors (or the potentiometer, respectively), the sound is more direct (not necessarily spacially), more accurate and also more meager and «dry» than with any of the comparative amps: X-Cans, EMP and Corda Blue. Details are more easily audible and clearer, especially in the treble (I think that's naturally the range where details are located). Each time an amp is in the signal path, the sound seems to gain in coherence and warmth, on the other hand lose some of the clarity, transparency, transient speed and resolution the puristic connection provides. The effect is very much the same independent of the headphone, but with different results in terms of my personal preferences.
With the Etys, the fascination of the full resolution the direct path provides is barely restricted by any lack of coherence or any sensation of coolness or dryness. Their innate character has a signature similar to the direct path's: from the CD directly to the brain... without the detour via pinna and auditory canal directly to the eardrum. The ear-catching lack of smearing effects is cultivated that way to an equally intellectual and sensual/esthetic delight: the beauty of the pure, unaffected sound.
With the EMP, there's just a trace less attack in the mids and lower highs, a flimsy veil, but of a very pleasing manner, exposing – on the other hand – the extreme highs a bit, creating a great beauty of its own kind with a touching musical flow and the impression of a subtle sheen. The soundstage is virtually unaltered, just a tiny loss of focus. – The Etys sound equally impressive and credible with the EMP like with the direct path, sort of milder, but equally lively presentation of the same musical image, maybe comparable with an ultra-fine TFT and a very sharp and fine flat CRT display. And the over-all character/balance is rather similar.
With the Corda Blue, the most conspicuous difference is its slight mat coloration. And like with the EMP, there is a certain minor loss of clarity and attack, accompanied by a kind of breathy timbre. It has its special charm, and – after a certain (short) familiarization – I perceive it as virtually free of coloration, but of an impressive 3-dimensionality and focus, even more than the direct path, and an intimate atmosphere. Of the three connection variants, this is my least preferred one, because of its slight lack of brilliance, but nevertheless the sound it provides is very good.
I still didn't mention the bass. The reason is: it's very similar with all three variants; there's maybe just a tiny bit more control with the direct path, on the other hand the EMP has the same tiny bit more low bass volume. All in all it's negligible.
With the HD 600, the direct path needs getting used to. And I'm ready to do so... the sound is extremely resolved that way. The separation of instruments is that clear that I'm tempted to call the soundscape empty, not to say hollow sometimes. Of course this scenario is pretty much overdrawn and the fruit of intensive A/B comparisons with multiply-repeated passages, but I can't but state that the direct sound has a slightly artificial timbre for that reason. Nevertheless I'm ready to get into it, and actually I'm on the way to find it rather pleasing, although lacking a bit of the warmth and smoothness the indirect paths provide. The highs are ultra-clear and brilliant, but not overly glossy, very detailed and unsmeared; the mids are lean and neutral; the bass is tight, fast and extended. But to be honest: it's not easy to rate single frequency ranges, because the sound is that fast and feather-light in every regard that it's hardly tangible.
With the EMP, this «problem» is virtually solved: it's quite easy to perceive the slight loss of lower and mid-treble (though it's very slight!) and the (thus?) pronounced upper treble, which gives the sound a charming touch, a dash of smoothness and warmth (actually a slight smearing) to the cool clarity as well as an increased perception of coherence, forming together a beautiful musical flow and an always interesting character with lots of details. – The sonic difference to the above mentioned direct sound is much smaller than it possibly appears by reading this paragraph, but the esthetical appearance is indeed considerably altered, ostensibly in favor of the EMP.
The Corda Blue has a very neutral, fluid and transparent midrange. Like the original HA-1, it has a certain preference for the upper bass and the lowest mids – not so much with sustained tones but more with transients. This is even clearer with the HD 600 than with the Etys and may be the main reason for its pronounced «contour» – be it sonically or spacially. This gives it a sense of «grip» and substance, at the expense of a trace of roughness, although of a very fine grain (maybe rather called «matness»). Sometimes the decay of cymbals slightly tends to sound like pink noise. (These characterizations express the differences to the direct path sound; in fact there is barely such a roughness perceivable without this reference.) This character doesn't harm its great ability to draw a credible 3-dimensional soundstage with very accurate instrument placement. And anyhow its refined (fine-grained), slightly rounded highs remind me of analog sound, though originating from a DAC...
Interestingly with my former Theta DAC I even slightly preferred the Corda Blue to the EMP because of its ability to create such a credible and intimate ambience, but now, with the Bel Canto, it's rather reversed.
Not to get out of hand, I'm abstracting the results with the rest of my equipment. The X-Cans have an astonishingly fine synergy with the HD 600 as well as with the K 501 – almost or in fact equal to the EMP, depending on the recording and my mood. Despite its seemingly reduced midrange transparency and -volume and a tendency to a sharp treble, it offers a great deal of naturalness, sense of space and listening pleasure with these two cans, while the direct sound gives mixed impressions: great clarity and resolution, a slight «digital» timbre and occasionally even shrillness. The same applies to both Grados, of which especially the SR-225 with its mediocre midrange resolution is worth a try with the direct path, but also the more detailed, but less neutral SR-325, at least with rock music.
Summary
There's only one headphone which sounds at least equally good using the direct path via potentiometer: the Etymotic ER-4S. With all other cans I prefer the use of a dedicated headphone amp so far. This seems not to be reasonable. But I have always trusted my ears first. That doesn't mean I think an amp does anything better than a plain cable with some resistors in-between. Rather I suspect that the signal smoothing effect an amp applies masks the technical sound in a way that it becomes agreeable to the ears. But is this hypothetical technical sound due to digital processing or even due to solid-state amplification in most of the digital source devices? The most important thing to me is how astonishingly close the Earmax Pro is to the direct path sound. Are tubes in fact more true than transistors, despite their opposite reputation? On the other hand: what can a further solid-state amplifier stage (like in my case the Corda Blue) change to the better from what a solid-state stage before (the DAC's output stage) has made wrong? (Double-solid-state smoothing – sort of «upsampling»...?) And – referring to this – if a headamp's job is primarily to apply some smoothing effects to the signal, why have the «best sounding» ones to be so damned expensive?
My own headamp arsenal is too small to draw final conclusions. I'd like to kindly invite other Head-Fi members to a fruitful discussion and maybe own direct-path experiments.
JaZZ
BTW: those who have plugged their headphones in their soundcard's line out jack can't really expect «better» sound from additional headphone amps after all – at least some kind of smoothing effect.
I guess there are very few people which already have tried to renounce or bypass their headphone amp and connect their headphones directly to their CD player or DAC. Given that the latter's analog output impedance is often clearly below 100 ohm and provides at least the standardized 2 volt, in some cases even considerably more, it must clearly be able to drive any dynamic headphone. And it seems very reasonable to use the shortest signal path possible, which necessarily will provide a virtually unaltered signal to feed the headphones with – in contrast to the headamp operation, which puts some – actually superfluous – amplifier stages in the signal path. Well, all that's necessary to do so is a 250-1000 ohm stereo potentiometer which acts as a pure attenuator (more specifically: as an adjustable bleeder/voltage divider) for loudness control.
Does this configuration have any downsides? Yes: there is a restriction of the source device's output impedance in order not to influence the system's frequency response (and ouput level) too much. I guess anything below 150 ohm will work fine. Besides, this output impedance and the potentiometer's impedance together form varying (depending on the attenuation adjustment) serial and parallel resistances. They cause an electrical interaction with the headphone's impedance curve, so that it may slightly increase the area around its bass resonance and possibly towards highest frequencies, due to the voice coil's inductivity. This happens in the same range as with most tube amps with output impedances between 100 and 300 ohm, while solid-state amps have low impedances of about 20 ohm.
I did compare the affection by serial resistances with my Corda HA-1 which has the extra 120 ohm output jack. It really changed the sound of all my headphones a bit more or less – apart from the volume reduction. The changes restricted themselves to (very) slight balance shifts, mostly in the form of an emphasizing of bass and upper treble. Actually one would expect that low-impedance cans like the Grados are more affected than the 300-ohm HD 600, but that's not the case – the reason is the much flatter impedance curve of the Grado models. And there's no loss of bass precision due to a reduced damping factor, as it would appear with bass speakers. This can clearly be reproduced with the Etymotic ER-4 P to S conversion proceeding or generally by watching the ER-4 system, with its 20 to 100 ohm serial resistors added to the plain 7 ohms of its driver's voice coil to provide a practical impedance – obviously without any harm to the sound.
Some information about the equipment used for this comparison:
– My Sennheiser HD 600 is equipped with a home-made cable made of capillary 0.028 mm magnet wires, plaited to four strands of 400 each; the (dust protection) foam pads which originally cover the drivers are removed. Both modifications move the sound towards considerably greater clarity, forwardness and airiness (no «veiled» sound anymore...).
– My Etymotic ER-4S are equipped with a Fixup replacement cable, with pleasing sound and clearly reduced microphonics compared to the stock cable; originally with two 50 ohm resistors (also called ER-«X»), I reverted to 100 ohm («S»), which has turned out as the better match with my portable Archos MP3 player and performs equally with my home setup. Their foam tips are made of pierced noise protection plugs, impregnated with contact adhesive for air-tight seal and thus stronger bass, leaving the transducer stems flush with the foam surface or even slightly protruding; this necessitates the protection of the thus exposed filter by some nylon gauze folded around the stem. The original filter (membrane) is replaced by some lambswool filaments for an even smoother, less resonant treble reproduction. Additionally this provides the possibility to fine-tune the sound by dosing the lambswool density.
– My Grado SR-225 and SR-325 are dressed with some modded yellow Sennheiser pads; in order not to lose too much treble thereby, I have bended them in such a way that they have a very lose fit on my head, thus low pressure to my ears.
– My AKG K 501 is virtually in original state, except for a black velvet coating of the driver's ring-shaped bare plastic surface, meant to reduce reflections inside the earpieces/-pads. With all of them I can clearly identify the sonic character of the respective amp used.
– My Musical Fidelity X-Cans (primary version), a well-known tube/solid-state hybrid amp, is in original state.
– My Brocksieper Earmax Pro presently is equipped with one Telefunken E801S in the input stage and two Siemens ECC188 in the output stage.
– My Corda Blue (Meier Audio Corda Headamp-1, modified by Head-Fi's famous KurtW) is my solid-state alternative.
– My passive headphone «amp» consists of an «airplane»-plywood housing and a logarithmic 500-ohm (no-name) potentiometer.
– My new Bel Canto DAC2 is a 24 bit/192 kHz upsampling DAC; output impedance: 20 ohm, output voltage: 2 volt.
– Interconnects: all IC cables are of the same kind as described in the HD 600 paragraph: made of ultra-thin magnet wire, i.e. individually isolated 0.028 mm copper wires, about 400 per conductor. This cable design has turned out – to my ears – to provide the best high-frequency extension, a super-clear upper end, assumedly thanks to a highly reduced skin effect...
I can't pretend to have achieved an exactly equal loudness level by simply adjusting it by naked ears, but I have done it with all the care that I could apply, and the various sessions with individual settings finally have effected consistent results.
With my DAC's line-out signal feeding a headphone directly, just attenuated by some resistors (or the potentiometer, respectively), the sound is more direct (not necessarily spacially), more accurate and also more meager and «dry» than with any of the comparative amps: X-Cans, EMP and Corda Blue. Details are more easily audible and clearer, especially in the treble (I think that's naturally the range where details are located). Each time an amp is in the signal path, the sound seems to gain in coherence and warmth, on the other hand lose some of the clarity, transparency, transient speed and resolution the puristic connection provides. The effect is very much the same independent of the headphone, but with different results in terms of my personal preferences.
With the Etys, the fascination of the full resolution the direct path provides is barely restricted by any lack of coherence or any sensation of coolness or dryness. Their innate character has a signature similar to the direct path's: from the CD directly to the brain... without the detour via pinna and auditory canal directly to the eardrum. The ear-catching lack of smearing effects is cultivated that way to an equally intellectual and sensual/esthetic delight: the beauty of the pure, unaffected sound.
With the EMP, there's just a trace less attack in the mids and lower highs, a flimsy veil, but of a very pleasing manner, exposing – on the other hand – the extreme highs a bit, creating a great beauty of its own kind with a touching musical flow and the impression of a subtle sheen. The soundstage is virtually unaltered, just a tiny loss of focus. – The Etys sound equally impressive and credible with the EMP like with the direct path, sort of milder, but equally lively presentation of the same musical image, maybe comparable with an ultra-fine TFT and a very sharp and fine flat CRT display. And the over-all character/balance is rather similar.
With the Corda Blue, the most conspicuous difference is its slight mat coloration. And like with the EMP, there is a certain minor loss of clarity and attack, accompanied by a kind of breathy timbre. It has its special charm, and – after a certain (short) familiarization – I perceive it as virtually free of coloration, but of an impressive 3-dimensionality and focus, even more than the direct path, and an intimate atmosphere. Of the three connection variants, this is my least preferred one, because of its slight lack of brilliance, but nevertheless the sound it provides is very good.
I still didn't mention the bass. The reason is: it's very similar with all three variants; there's maybe just a tiny bit more control with the direct path, on the other hand the EMP has the same tiny bit more low bass volume. All in all it's negligible.
With the HD 600, the direct path needs getting used to. And I'm ready to do so... the sound is extremely resolved that way. The separation of instruments is that clear that I'm tempted to call the soundscape empty, not to say hollow sometimes. Of course this scenario is pretty much overdrawn and the fruit of intensive A/B comparisons with multiply-repeated passages, but I can't but state that the direct sound has a slightly artificial timbre for that reason. Nevertheless I'm ready to get into it, and actually I'm on the way to find it rather pleasing, although lacking a bit of the warmth and smoothness the indirect paths provide. The highs are ultra-clear and brilliant, but not overly glossy, very detailed and unsmeared; the mids are lean and neutral; the bass is tight, fast and extended. But to be honest: it's not easy to rate single frequency ranges, because the sound is that fast and feather-light in every regard that it's hardly tangible.
With the EMP, this «problem» is virtually solved: it's quite easy to perceive the slight loss of lower and mid-treble (though it's very slight!) and the (thus?) pronounced upper treble, which gives the sound a charming touch, a dash of smoothness and warmth (actually a slight smearing) to the cool clarity as well as an increased perception of coherence, forming together a beautiful musical flow and an always interesting character with lots of details. – The sonic difference to the above mentioned direct sound is much smaller than it possibly appears by reading this paragraph, but the esthetical appearance is indeed considerably altered, ostensibly in favor of the EMP.
The Corda Blue has a very neutral, fluid and transparent midrange. Like the original HA-1, it has a certain preference for the upper bass and the lowest mids – not so much with sustained tones but more with transients. This is even clearer with the HD 600 than with the Etys and may be the main reason for its pronounced «contour» – be it sonically or spacially. This gives it a sense of «grip» and substance, at the expense of a trace of roughness, although of a very fine grain (maybe rather called «matness»). Sometimes the decay of cymbals slightly tends to sound like pink noise. (These characterizations express the differences to the direct path sound; in fact there is barely such a roughness perceivable without this reference.) This character doesn't harm its great ability to draw a credible 3-dimensional soundstage with very accurate instrument placement. And anyhow its refined (fine-grained), slightly rounded highs remind me of analog sound, though originating from a DAC...
Interestingly with my former Theta DAC I even slightly preferred the Corda Blue to the EMP because of its ability to create such a credible and intimate ambience, but now, with the Bel Canto, it's rather reversed.
Not to get out of hand, I'm abstracting the results with the rest of my equipment. The X-Cans have an astonishingly fine synergy with the HD 600 as well as with the K 501 – almost or in fact equal to the EMP, depending on the recording and my mood. Despite its seemingly reduced midrange transparency and -volume and a tendency to a sharp treble, it offers a great deal of naturalness, sense of space and listening pleasure with these two cans, while the direct sound gives mixed impressions: great clarity and resolution, a slight «digital» timbre and occasionally even shrillness. The same applies to both Grados, of which especially the SR-225 with its mediocre midrange resolution is worth a try with the direct path, but also the more detailed, but less neutral SR-325, at least with rock music.
Summary
There's only one headphone which sounds at least equally good using the direct path via potentiometer: the Etymotic ER-4S. With all other cans I prefer the use of a dedicated headphone amp so far. This seems not to be reasonable. But I have always trusted my ears first. That doesn't mean I think an amp does anything better than a plain cable with some resistors in-between. Rather I suspect that the signal smoothing effect an amp applies masks the technical sound in a way that it becomes agreeable to the ears. But is this hypothetical technical sound due to digital processing or even due to solid-state amplification in most of the digital source devices? The most important thing to me is how astonishingly close the Earmax Pro is to the direct path sound. Are tubes in fact more true than transistors, despite their opposite reputation? On the other hand: what can a further solid-state amplifier stage (like in my case the Corda Blue) change to the better from what a solid-state stage before (the DAC's output stage) has made wrong? (Double-solid-state smoothing – sort of «upsampling»...?) And – referring to this – if a headamp's job is primarily to apply some smoothing effects to the signal, why have the «best sounding» ones to be so damned expensive?
My own headamp arsenal is too small to draw final conclusions. I'd like to kindly invite other Head-Fi members to a fruitful discussion and maybe own direct-path experiments.
BTW: those who have plugged their headphones in their soundcard's line out jack can't really expect «better» sound from additional headphone amps after all – at least some kind of smoothing effect.