Post by manueljenkin on Jan 28, 2020 7:31:54 GMT
Hi Everyone (and solderdude),
I went through the review and measurements of the SRH1540 and SRH1840 Headphones. A big thank you to solderdude for taking the time and effort to make these headphone measurements.
Now I personally haven't tried SRH1840, but I own the SRH1540 and have heard SRH940. My friends have tried 1840 and were impressed as well. While it is quite easy to look at measurements and examine aberrations, I find that a deeper inspection of the chain and measuring gear is necessary.
First up, all of these headphones are very hard to drive. Not volume wise, but gain-bandwidth wise (if you can relate it to op-amps). I have been able to get my SRH1540 to make loud enough sound through my surface book which barely has 10 milliwatts of output, and have also had it plugged into my Burson fun tapping 1.5 Watts on the knob. No I haven't even remotely blasted my ears. It does sound warm and fluffy when under-driven. It is easy to think that enough power has been delivered, but trust me, with a source chain that has enough power and bandwidth, the SRH1540 is much brighter than the HD800 (I own both). If you have a look at tyll's video, you'd see him mentioning that it was too bright for him at normal volumes. With a proper amplification, the highs are a lot more present and very well extended.The biggest change you hear with amplification are dynamics and bandwidth (highs and lows extension), and all of those are related to what I mentioned above - Gain-Bandwidth product. The explanation lies in how the amplifier is able to handle the back emf of the transducer (the amplifier has to do work against magnetic field of the headphones).
This is not only exclusive to the shures, but most pro monitor headphones around share a similar design. Sony has it's cd900 for which companies make special amplifiers, and people drive their Qualias out of speaker amps. I am not a mechanical engineer and haven't opened it up, but just peeking at the impedance plot and judging by their interaction with amplifiers, it is clear that they are not suspended in a very high intensity magnetic field (unlike say, a HD800), but rather in a region of uniform low magnetic field almost throughout their excursion. You don't see a huge inductive impedance spike, and their impedance plot is almost planar like, especially the 1840. The result of such a design is that, while they take a lot of power to get to the desired bandwidth, they perform the task of a truly transparent system, getting out of the way and letting you hear the recording (and your dac if it has any coloration). Because of this, they are also very sensitive of filters, and you will get to notice band-passing (also referred to as grain in audiophile circles) in a lot of recordings. Depending on your dac, you will also be able to hear dither noise for 16 bit music (sounds like a layer of coarseness on top of the sound). They are also sensitive to aggressive low passing in amplifiers. On question of why you don't hear that in other headphones, if they are already filtered/low passed in their response, they wouldn't be able to resolve this difference anyway. Low passing always sounds pleasant and can filter out noise in recordings but at the expense of dynamics/transients.
Now the second point, about the 8khz spike on the SRH1840. If you take harman as reference, yep, they do not recommend the 8khz boost. However, if you take the diffuse field target, the 8Khz response is "correct". SRH1840 is a diffuse field tuned headphone. Most of Harman target's differences from diffuse field are meant to cleave to compensate for the perceptual lack of weight due to lack of crossfeed. If you have music that has content that is very side panned, the tone of SRH1540 that is somewhat Harman targetesque (though its a mild u shape on top of it, bump in 100-300hz and bump in mid treble) will feel more realistic in timbre. However it'll sound boomy when the same data is set at the centre. On the 1840, it will be vice versa. Center panned music will be more accurate in timbre while side panned music will sound dry. If you listen to mostly music mixed on speakers, SRH1540, HD600 etc will sound more in line with what was intended. If you listen to binaural, live, or even mono, the tone of SRH1840, SRH940, HD800, DT880 will be more correct. If you can simulate crossfeed, again the latter would be more correct sounding. This is the reason why most "live" monitors have a 8khz spike since they monitor mostly in mono. You can explore further by searching for articles on the "missing 6db". An example - rinchoi.blogspot.com/2012/05/on-case-of-missing-6-db-effect.html
A binaural track for you to try out with different tones : vimeo.com/184459967
The third point, the distortion measurements. " The Shure SRH-1540 is a warmish neutral headphone with somewhat elevated and tactile bass ". If you drive the SRH1840 with enough power and bandwidth, it will also have the same tactile bass, in fact even more than the 1540. You can actually feel the pads vibrating the skull. I can feel it with my srh1540, and to an extent with my HD800. It is quite easy to look at a chart and say that they distort. However if you look closer, you will see that there is a discrepancy. There is discrepancy in CSD as measured in solderdude's rig as compared to few other rigs : www.changstar.com/www.changstar.com/1840l.txt4210.jpg?action=dlattach;topic=456.0;attach=2083;image . It measured very similar at goldenears, and SRH940/1840 was the fastest CSD measured there, faster than HD800 by quite a margin (Referring to corrected HD800 measurements, the first one they made had an issue). 3rd order and 5th order is measured high but Csd is wink fast. Odd order harmonics means a system has stored energy and overtly impulsive (sines are becoming squares or triangles), which contradicts what the Csd shows where it just decays in an instant showing zero hint of stored energy. There is very good chance of the rig showing some false aberration. If a headphone transducer is physically breaking, the distortion should keep rising as the frequency reduces especially in the lowest registers when it ventures below f0. Not stabilize like you see in charts. Also, If a system is mechanically breaking it should also be accompanied by significant amount of even order harmonics as well. If you look at the pattern of raising and stabilizing thd in innerfidelity charts, you can see a pattern in most of them. The graph in other tactile stuff, like ZMF, and especially some specialised tactile stuff like Skullcandy crushers trace a similar pattern. What could have happened is the headphone could have rattled the rig due to its tactility and the mic might have picked an impulsive shock noise from the rig (which shows up as odd order harmonics). Since the tool only does math calculation, it could have interpreted it as distortion.
A simple way to cross check this is, just to open up the recording of the wave used for measurements and listen back if you can hear some spurious noise that attributes to be an impulsive sound. On a deeper note, I find tactility to be another dimension to music, and a vibration sensor in addition to the recording mics can quite reliably measure that. If you still have the headphone with you, I would be highly obliged if you can see to repeat the measurements (along with vibration sensor if possible) on a more high power high bandwidth amplifier (and a dac that also has same performance).
I would also like to recommend one more test to measure "dynamics". It is quite unreliable to measure reproduction of higher frequencies directly with sine sweeps due to the number of variables involved and how the math can misinterpret phase issues. An easier way to do would be to send 3khz or 5khz square wave bursts. Their harmonics will have higher frequency components. And it is essential to check from the very first pulse. Just a burst of 5 to 10 pulses would be enough to see how the system performs. Innerfidelity square waves snapped at the middle measure stable state behavior of headphones and not the transient behavior.
Thanks,
Manuel Jenkin.
I went through the review and measurements of the SRH1540 and SRH1840 Headphones. A big thank you to solderdude for taking the time and effort to make these headphone measurements.
Now I personally haven't tried SRH1840, but I own the SRH1540 and have heard SRH940. My friends have tried 1840 and were impressed as well. While it is quite easy to look at measurements and examine aberrations, I find that a deeper inspection of the chain and measuring gear is necessary.
First up, all of these headphones are very hard to drive. Not volume wise, but gain-bandwidth wise (if you can relate it to op-amps). I have been able to get my SRH1540 to make loud enough sound through my surface book which barely has 10 milliwatts of output, and have also had it plugged into my Burson fun tapping 1.5 Watts on the knob. No I haven't even remotely blasted my ears. It does sound warm and fluffy when under-driven. It is easy to think that enough power has been delivered, but trust me, with a source chain that has enough power and bandwidth, the SRH1540 is much brighter than the HD800 (I own both). If you have a look at tyll's video, you'd see him mentioning that it was too bright for him at normal volumes. With a proper amplification, the highs are a lot more present and very well extended.The biggest change you hear with amplification are dynamics and bandwidth (highs and lows extension), and all of those are related to what I mentioned above - Gain-Bandwidth product. The explanation lies in how the amplifier is able to handle the back emf of the transducer (the amplifier has to do work against magnetic field of the headphones).
This is not only exclusive to the shures, but most pro monitor headphones around share a similar design. Sony has it's cd900 for which companies make special amplifiers, and people drive their Qualias out of speaker amps. I am not a mechanical engineer and haven't opened it up, but just peeking at the impedance plot and judging by their interaction with amplifiers, it is clear that they are not suspended in a very high intensity magnetic field (unlike say, a HD800), but rather in a region of uniform low magnetic field almost throughout their excursion. You don't see a huge inductive impedance spike, and their impedance plot is almost planar like, especially the 1840. The result of such a design is that, while they take a lot of power to get to the desired bandwidth, they perform the task of a truly transparent system, getting out of the way and letting you hear the recording (and your dac if it has any coloration). Because of this, they are also very sensitive of filters, and you will get to notice band-passing (also referred to as grain in audiophile circles) in a lot of recordings. Depending on your dac, you will also be able to hear dither noise for 16 bit music (sounds like a layer of coarseness on top of the sound). They are also sensitive to aggressive low passing in amplifiers. On question of why you don't hear that in other headphones, if they are already filtered/low passed in their response, they wouldn't be able to resolve this difference anyway. Low passing always sounds pleasant and can filter out noise in recordings but at the expense of dynamics/transients.
Now the second point, about the 8khz spike on the SRH1840. If you take harman as reference, yep, they do not recommend the 8khz boost. However, if you take the diffuse field target, the 8Khz response is "correct". SRH1840 is a diffuse field tuned headphone. Most of Harman target's differences from diffuse field are meant to cleave to compensate for the perceptual lack of weight due to lack of crossfeed. If you have music that has content that is very side panned, the tone of SRH1540 that is somewhat Harman targetesque (though its a mild u shape on top of it, bump in 100-300hz and bump in mid treble) will feel more realistic in timbre. However it'll sound boomy when the same data is set at the centre. On the 1840, it will be vice versa. Center panned music will be more accurate in timbre while side panned music will sound dry. If you listen to mostly music mixed on speakers, SRH1540, HD600 etc will sound more in line with what was intended. If you listen to binaural, live, or even mono, the tone of SRH1840, SRH940, HD800, DT880 will be more correct. If you can simulate crossfeed, again the latter would be more correct sounding. This is the reason why most "live" monitors have a 8khz spike since they monitor mostly in mono. You can explore further by searching for articles on the "missing 6db". An example - rinchoi.blogspot.com/2012/05/on-case-of-missing-6-db-effect.html
A binaural track for you to try out with different tones : vimeo.com/184459967
The third point, the distortion measurements. " The Shure SRH-1540 is a warmish neutral headphone with somewhat elevated and tactile bass ". If you drive the SRH1840 with enough power and bandwidth, it will also have the same tactile bass, in fact even more than the 1540. You can actually feel the pads vibrating the skull. I can feel it with my srh1540, and to an extent with my HD800. It is quite easy to look at a chart and say that they distort. However if you look closer, you will see that there is a discrepancy. There is discrepancy in CSD as measured in solderdude's rig as compared to few other rigs : www.changstar.com/www.changstar.com/1840l.txt4210.jpg?action=dlattach;topic=456.0;attach=2083;image . It measured very similar at goldenears, and SRH940/1840 was the fastest CSD measured there, faster than HD800 by quite a margin (Referring to corrected HD800 measurements, the first one they made had an issue). 3rd order and 5th order is measured high but Csd is wink fast. Odd order harmonics means a system has stored energy and overtly impulsive (sines are becoming squares or triangles), which contradicts what the Csd shows where it just decays in an instant showing zero hint of stored energy. There is very good chance of the rig showing some false aberration. If a headphone transducer is physically breaking, the distortion should keep rising as the frequency reduces especially in the lowest registers when it ventures below f0. Not stabilize like you see in charts. Also, If a system is mechanically breaking it should also be accompanied by significant amount of even order harmonics as well. If you look at the pattern of raising and stabilizing thd in innerfidelity charts, you can see a pattern in most of them. The graph in other tactile stuff, like ZMF, and especially some specialised tactile stuff like Skullcandy crushers trace a similar pattern. What could have happened is the headphone could have rattled the rig due to its tactility and the mic might have picked an impulsive shock noise from the rig (which shows up as odd order harmonics). Since the tool only does math calculation, it could have interpreted it as distortion.
A simple way to cross check this is, just to open up the recording of the wave used for measurements and listen back if you can hear some spurious noise that attributes to be an impulsive sound. On a deeper note, I find tactility to be another dimension to music, and a vibration sensor in addition to the recording mics can quite reliably measure that. If you still have the headphone with you, I would be highly obliged if you can see to repeat the measurements (along with vibration sensor if possible) on a more high power high bandwidth amplifier (and a dac that also has same performance).
I would also like to recommend one more test to measure "dynamics". It is quite unreliable to measure reproduction of higher frequencies directly with sine sweeps due to the number of variables involved and how the math can misinterpret phase issues. An easier way to do would be to send 3khz or 5khz square wave bursts. Their harmonics will have higher frequency components. And it is essential to check from the very first pulse. Just a burst of 5 to 10 pulses would be enough to see how the system performs. Innerfidelity square waves snapped at the middle measure stable state behavior of headphones and not the transient behavior.
Thanks,
Manuel Jenkin.