Details

Parent Category: Products

Category: Headphone Measurements

Created: 01 September 2014

I measured the KEF M200 earphones using a G.R.A.S. RA0045 ear simulator, a Clio 10 FW audio analyzer, a laptop computer running TrueRTA software with an M-Audio MobilePre USB audio interface, and a Musical Fidelity V-Can headphone amplifier. Measurements were calibrated for drum reference point (DRP), the equivalent of a earphone’s response at the surface of your eardrum. This is a “flat” measurement; no diffuse-field or free-field compensation curve was employed. I used the medium-size silicone tips that I received with the review samples. Because of the large diameter of the M200s’ sound tubes, I had to press lightly with a fingertip to get a good seal in the ear simulator.

There’s some disagreement about what constitutes a good frequency-response measurement for earphones, but I think all experts would agree that this one looks unusual. That dip in the midrange centered at about 900Hz is fairly common, but here it’s about -6dB lower than I usually measure. The treble response above 4kHz is cleaner than I’m used to seeing, with just one strong, narrow peak at 8kHz instead of the usual multiple, spread-out peaks.

Adding 70 ohms output impedance to the V-Can’s 5-ohm output impedance to simulate the effects of using a typical low-quality headphone amp does affect the M200s’ performance, but only in a band of about one-third of an octave centered at 13kHz, where the response drops about -4dB with the high-impedance source. This would be audible to most people, although whether you’d perceive it as an improvement in or a degradation of the sound would depend on your hearing and taste.

This comparison of the M200s with NAD’s Viso HP20 and Bowers & Wilkins’ C5 earphones suggests that the KEFs are the least likely to be perceived as having a flat response, thanks to that big midrange dip. To my ears (and those of many other reviewers), the Viso HP20s sound fairly flat, the C5s a little on the bassy side.

The spectral-decay (waterfall) plot looks clean except for one very strong resonance at 4.8kHz. But given the narrowness of this resonance, I expect it would be audible only with certain pieces of music, and then only fleetingly.

The total harmonic distortion (THD) at 100dBA is a little high relative to the best earphones I’ve measured, hitting about 3% at 1kHz, but given that 3% isn’t such a high distortion level in transducers, and that 100dBA is an extremely loud playback level, I doubt you’d encounter this flaw in normal listening.

In this chart, the external noise level is 75dB SPL; numbers below that indicate the degree of attenuation of outside sounds. Thanks probably to its big, fat 6.8mm sound tubes, the M200s deliver good isolation from outside sounds. In the key band between 100Hz and 1kHz, the reduction ranges from -11dB at 100Hz to -27dB at 1kHz, and even better at higher frequencies.

The impedance magnitude is almost dead flat (if unusually low) at 12.5 ohms; the impedance phase is also almost entirely flat.

The M200s’ average sensitivity from 300Hz to 3kHz at the rated 12 ohms measures 96.7dB, which is -7 to -10dB lower than I measure with typical earphones. That’s because of the big midrange dip. With the M200s, you’ll probably need to turn your smartphone up to nearly maximum volume.

. . . Brent Butterworth
brentb@soundstagenetwork.com

Details

Parent Category: Products

Category: Headphone Measurements

Created: 01 August 2014

I measured the performance of the HP50 headphones using a G.R.A.S. 43AG ear/cheek simulator, a Clio FW audio analyzer, a laptop computer running TrueRTA software with an M-Audio MobilePre USB audio interface, and a Musical Fidelity V-Can headphone amplifier. Measurements were calibrated for ear reference point (ERP), which is roughly the point in space where, with your hand pressed against your ear, your palm intersects the axis of your ear canal, and roughly the place where the front of the headphone’s driver grille will sit when you wear the ’phones. This is a “flat” measurement; no diffuse-field or free-field compensation curve was employed. I experimented with the position of the earpads by moving them around slightly on the ear/cheek simulator, and settled on the positions that gave the best bass response and the most characteristic result overall.

The HP50s’ frequency response shows a strong peak centered at 2.8kHz (as one often finds in headphone response measurements), and a broad, strong boost between 6 and 10kHz. This is similar to the typical diffuse-field equalization employed in many headphones. Adding 70 ohms output impedance to the V-Can’s 5-ohm output impedance, to simulate the effects of using a typical low-quality headphone amp, produces a slight increase (about 1dB) in bass response between 40 and 90Hz.

Compared with most of the headphones I’ve measured, the HP50s’ response looks pretty flat. The treble does show some boost, but it’s slight -- typically, 1 to 2dB relative to the bass -- and very broad, covering the entire band between 2 and 8kHz. So the treble should be pretty much uncolored, and that rise could have the result of making the HP50s sound a tad bright (or the bass a tad damped). Note the slight difference in bass response, probably due to the fit of the different earpieces on the ear/cheek simulator; these are the best results I was able to achieve.

When I tried increasing the source impedance from 5 to 75 ohms, to simulate the effect of using a low-quality source device such as a typical laptop computer, there was no notable change in frequency response.

You can see from the chart above how similar the HP50s are to Paul Barton’s other model of passive over-ear headphones, the PSB M4U 1. The HP50s produce a bit more low bass, but the difference you’ll probably notice most is the HP50s’ roughly -3dB dip in response around 1kHz. According to Barton, this results in not an audible dip in midrange sound, but an increased sense of the headphones sounding like real speakers in a real room. In comparison, the Bowers & Wilkins P7 headphones produce less bass but have a relatively strong peak around 2.8kHz, which means they should sound somewhat brighter, with less bottom-end kick.

The spectral-decay (waterfall) plot shows a couple of strong but very narrow (and thus probably inaudible) resonances, at 1.8 and 2.9kHz.

The total harmonic distortion (THD) is extremely low at 100dBA, and remains less than 2% at 20Hz. 

The spectrum of a 500Hz sinewave shows that, even at 100dB, all of the distortion harmonics are well below -70dBFS (0.03%) and are thus inaudible.

For passive headphones, the HP50s do a great job of attenuating external sounds. They reduce external noise by -15dB at 1kHz, and by as much as -40dB at 8kHz. They won’t do much for you on a plane, though; as with most passive closed-back headphones, there’s no significant reduction of noise below 200Hz, which is where jet engines produce most of their noise.

The impedance averages 37 ohms, and the magnitude and phase are both close to flat.

The HP50s’ average sensitivity from 300Hz to 3kHz, at the rated impedance of 32 ohms, measures 106.3dB with a 1mW signal.

. . . Brent Butterworth
brentb@soundstagenetwork.com

Details

Parent Category: Products

Category: Headphone Measurements

Created: 15 July 2014

I measured the Bowers & Wilkins P7 headphones using a G.R.A.S. Model 43AG ear simulator, a Clio 10 FW audio analyzer, a laptop computer running TrueRTA software with an M-Audio MobilePre USB audio interface, and a Musical Fidelity V-CAN headphone amplifier. Measurements were calibrated for the ear reference point (ERP) -- roughly the point in space where, with one palm pressed against your ear, the axis of your ear canal intersects with your palm. This is a “flat” measurement; no diffuse-field or free-field compensation curve was employed.

The P7s’ frequency-response curve correlates fairly well with what’s generally accepted as a subjectively flat headphone response. There’s a little extra energy at 2.3 and 7kHz, corresponding to the lower and mid-treble regions, which suggests that the treble response will be strong; you should get a heightened sense of detail.

Adding 70 ohms of output impedance to the V-CAN’S 5-ohm output impedance, to simulate the effects of using a typical low-quality headphone amp, has a noticeable but mild effect on the P7s’ response: a slight bass boost (typically, +1.5dB) when the P7s are used with a higher-impedance source device, such as a typical laptop computer or cheap smartphone. This will result in slightly more perceived bass, and probably a slightly softer, smoother treble response. But considering the P7s’ comparatively strong treble, they should sound good even with low-quality sources.

Compared to two well-regarded over-ear headphones, the ADL H118 and the Focal Spirit Classic models, the P7s have a flatter response below 1kHz, but a notably stronger treble response.

The P7s have a nice, clean decay, the only significant resonance showing up in the vicinity of 900Hz. And even that’s at -30 to -40dBFS, and lasts only about 10 milliseconds.

The P7s’ total harmonic distortion (THD) at 90 and 100dBA is generally fairly low, but, as with many headphones, it rises below 100Hz, reaching 8.5% at 20Hz at 100dBA. Note, though, that this is a very loud listening level.

In this chart, the external noise is at an SPL of 75dB; the numbers below that level indicate the attenuation of external sounds. For an over-ear, sealed-back headphone, the P7s’ isolation is a little better than average, reducing noise at 1kHz by about 14dB, and by 25 to 30dB at higher 

The P7s’ impedance is basically flat, with a little rise in the bass. That’s why its bass response in the 5 vs. 75 ohms response chart shows a slight bump upward with high-impedance (75 ohm) source devices. The impedance phase response is also close to flat.

The B&W P7s’ sensitivity, measured with a 1mW signal calculated for the rated 22 ohms impedance, is 101.7dB. That’s moderate, suggesting that the P7s will play loud, but not real loud, from most smartphones and portable audio players.

. . . Brent Butterworth
brentb@soundstagenetwork.com 

Details

Parent Category: Products

Category: Headphone Measurements

Created: 15 June 2014

To measure the Focal Spirit Classic headphones, I used my usual rig: a G.R.A.S. 43AG ear/cheek simulator, a Clio FW audio analyzer, a laptop computer running TrueRTA software with an M-Audio MobilePre USB audio interface, and a Musical Fidelity V-Can headphone amplifier. Measurements were calibrated for ear reference point (ERP): roughly, the point in space where your palm intersects with the axis of your ear canal when you press your hand against your ear; and, roughly, the place where the front of the driver grille will sit when you wear the headphones. This is a “flat” measurement; no diffuse-field or free-field compensation curve was employed. I experimented with the positions of the earpads by moving them around slightly on the ear/cheek simulator, and settled on the positions that gave the best bass response and the most characteristic result overall.

Although there’s still no broad agreement on what measurement of headphone frequency response would correspond to a perceived flat response, the Spirit Classics’ measured response pretty much squares with what I’ve found most people to perceive as flat. There’s a typical response bump at 3.1kHz, to accommodate the natural resonance of the human ear canal, and a strong measured response out to 9kHz. Only the bass looks a bit deficient, but this curve is actually quite similar to measurements I’ve taken of several respected audiophile models.

Adding 70 ohms to the V-Can’s output impedance of 5 ohms, to simulate the effects of using a typical low-quality headphone amp, produces no significant difference in response. Considering that the Spirit Classics’ sensitivity is fairly high, you should be able to get plenty of output and decent sound from anything with a headphone jack.

Compared to the ADL H118 and Bowers & Wilkins P7 headphones (both shown in the accompanying chart), the Spirit Classics have a little more bass and a little less treble, but their tonal balance looks flatter than either competitor’s.

The Focals’ spectral decay (waterfall) plot shows a very strong resonance at 800Hz -- which happens to correspond with a dip in the measured frequency response right at that frequency. However, this is a fairly narrow resonance, so I suspect it would be only occasionally audible.

Total harmonic distortion (THD) is practically nonexistent, even at 100dBA.

The spectrum of a 500Hz sinewave shows that the most audible distortion artifact at 100dB is the third harmonic (1500Hz), at about -56dB (about 0.16%).

The Spirit Classics attenuate external sounds pretty well for passive, closed-back headphones, reducing outside noise by -14dB at 1kHz and by as much as -32dB at higher frequencies. There’s no reduction in the “jet-engine band” below 200Hz, though.

The Focals’ impedance is mostly flat, averaging 33 ohms below 10kHz.

The average sensitivity from 300Hz to 3kHz at the specified impedance of 32 ohms measures 104.8dB with a 1mW signal.

. . . Brent Butterworth
brentb@soundstagenetwork.com

Details

Parent Category: Products

Category: Headphone Measurements

Created: 01 June 2014

I measured the Audiofly AF140s using a G.R.A.S. RA0045 ear simulator, a Clio 10 FW audio analyzer, a laptop computer running TrueRTA software with an M-Audio MobilePre USB audio interface, and a Musical Fidelity V-Can headphone amplifier. Measurements were calibrated for drum reference point (DRP), the equivalent of the headphones’ response at the surface of the eardrum. This is a “flat” measurement; no diffuse-field or free-field compensation curve was employed. I used the medium-size silicone tips supplied, and also tried the medium-size foam tips. There was only a slight difference in response, so except for the isolation measurement, I stuck with the silicone tips.

Compared with other earphones I’ve measured, the Audiofly AF140 has a relatively (but not grossly) strong peak at 2.5kHz, and a little bit (relatively speaking) of a dip in the mids at around 800Hz, but overall, the response looks fairly flat.

Adding 70 ohms to the V-Can’s output impedance of 5 ohms, to simulate the effects of using a typical low-quality headphone amp, has a big effect on the AF140’s performance. While the balance of bass to treble is pretty much flat with a low-impedance (i.e., high-quality) source device, the tonal balance will radically change if a high-impedance source (e.g., a cheap smartphone, tablet, or computer) is used: the sound will become much bassier and much duller in the highs. This is due to the big impedance swing you can see elsewhere in these measurements.

This comparison of the AF140 with the NuForce Primo 8 (a multidriver, balanced-armature earphone) and the Sony XBA-H1 (a hybrid design with one dynamic and one balanced-armature driver) shows that the Audioflys’ overall tonal balance looks fairly flat; their only real anomaly is that midrange dip.

Except for a long but very low-level resonance at 4kHz and a fairly strong but brief resonance at 13kHz (which also shows up in the frequency-response curves), the AF140’s decay looks pretty clean.

The AF140s’ total harmonic distortion (THD) at 90 and 100dBA is relatively high. Granted, these are very loud levels, but most of the earphones I’ve measured produce no more than a few percent THD on these tests. Even at 90dBA, a loud but still realistic listening level, the distortion hits 3% at 800Hz. This will be very audible, given that the main distortion harmonics will be at 1.6 and 2.4kHz, where human hearing is very sensitive. At 100dBA -- a level impracticable for listening but that, in these measurements, does tend to separate great from so-so products -- the distortion hits 14% at 800Hz.

The spectrum of a 500Hz sinewave confirms that AF140s’ distortion is relatively high. Even at the loud but not crazy-loud level of 90dBA, the levels of second- and third-harmonic distortion are almost equal, at about -28dB.

In this chart, the external noise level is at an SPL of 75dB; the numbers below that indicate attenuation of outside sounds. The AF140s don’t deliver much isolation in the “jet engine” band down low, but their isolation is fantastic at higher frequencies. Between 100Hz and 1kHz, the reduction ranges from -5dB at 100Hz to -17dB at 1kHz. But look at the 4kHz result: down -45dB! So while the AF140s probably won’t do a lot to damp the drone of jet engines, they’ll definitely block much of the hissing of the ventilation system, and probably help quiet screaming kids (for you, at least).

The AF140s’ impedance curve is a wild ride. That’s not uncommon for balanced-armature earphones, but still, dropping from 74 ohms at 20Hz to 16 ohms at 11kHz is a pretty big swing. However, the impedance phase is fairly flat.

The AF140s’ average sensitivity from 300Hz to 3kHz at the rated 38 ohms measures 106.2dB -- enough to play loudly with practically any source device.

. . . Brent Butterworth
brentb@soundstagenetwork.com 

Details

Parent Category: Products

Category: Headphone Measurements

Created: 15 May 2014

I measured the performance of the Audeze LCD-X headphones using a G.R.A.S. 43AG ear/cheek simulator, a Clio FW audio analyzer, a laptop computer running TrueRTA software with an M-Audio MobilePre USB audio interface, and a Musical Fidelity V-Can headphone amplifier. Measurements were calibrated for ear reference point (ERP), which is roughly the point in space where your palm intersects with the axis of your ear canal when you press your hand against your ear, and the place where the front of a driver grille sits when you wear the headphones. This is a “flat” measurement; no diffuse-field or free-field compensation curve was used. I experimented with the positions of the earpads by moving them around slightly on the ear/cheek simulator, and settled on the positions that gave the best bass response and the most characteristic result overall.

The LCD-X’s frequency response is about what I’m used to seeing from planar-magnetic headphones: dead flat from about 50Hz to 1.2kHz, with a peak at around 2.5kHz, which is typical. The only anomaly is that the treble response is a little less than I’m used to seeing from planar magnetics. Note the difference in bass response between the right and left channels: Each represents the best measurement I was able to get in more than a half-dozen tries; however, because slight differences in the seal of the earpads against the ear/cheek simulator can have huge effects on the measured response, I have no way of knowing if this represents an actual imbalance or a measurement artifact. Regardless, I didn’t notice any difference in the sounds of the left and right channels in my listening to the Audezes.

Because planar-magnetic drivers have an almost purely resistive (i.e., flat) impedance curve, their tonal balance rarely changes with different source devices. I simulated a change in source device by adding 70 ohms output impedance to the V-Can’s 5-ohm output impedance, for a total of 75 ohms, which is typical of the low-quality headphone amps built into laptops and cheap MP3 players. As you can see, there’s no significant difference in the Audezes’ response.

This chart compares the LCD-X with Audeze’s top model, the LCD-3, and with Oppo Digital’s new PM-1 planar-magnetic headphones. You can see that the LCD-X and LCD-3 headphones are practically identical up to 7kHz, but that the LCD-3s have 4 to 6dB more output between 7 and 9kHz. This should, at least in theory, give the LCD-3s a slightly brighter sound, and probably an enhanced sense of “air” and spaciousness. The Oppo PM-1s have a flatter response, but less energy in the lower treble.

The spectral-decay (waterfall) plot shows a couple of mild resonances in the vicinities of 800Hz and 1.4kHz, but their duration and bandwidth are so low that you probably wouldn’t notice them.

This plot of the LCD-Xs’ total harmonic distortion vs. frequency is one of the cleanest I’ve seen. The orange trace is taken at 100dBA, a higher level than most people could stand to listen to for more than a few seconds; even so, distortion is practically nonexistent.

This spectrum plot of the distortion harmonics, again taken at very loud levels, provides still more evidence that these are super-clean-sounding headphones.

Open-back planar-magnetic headphones provide little or no isolation from outside sounds, and the LCD-Xs almost none at all -- along with your music, you’ll hear everything going on around you.

The LCD-Xs’ impedance magnitude is essentially flat at 22 ohms (which matches the specification), and the impedance phase shift is near zero.

The LCD-Xs’ average sensitivity from 300Hz to 3kHz, at the rated 22 ohms, measured 101.5dB, which is very high for planar-magnetic headphones.

. . . Brent Butterworth
brentb@soundstagenetwork.com

Details

Parent Category: Products

Category: Headphone Measurements

Created: 01 April 2014

I measured the NAD Viso HP20s using a G.R.A.S. RA0045 ear simulator, a Clio 10 FW audio analyzer, a laptop computer running TrueRTA software with an M-Audio MobilePre USB audio interface, and a Musical Fidelity V-CAN headphone amplifier. Measurements were calibrated for drum reference point (DRP), the equivalent of the headphones’ response at the surface of the eardrum. This is a “flat” measurement; no diffuse-field or free-field compensation curve was employed. Except as noted, I used the HP20s’ medium standard eartips. I experimented with the fit of the eartips and earpieces by inserting and reinserting them in the RA0045, and settled on the positions that gave the best bass response and the most characteristic result overall.

Earphones don’t always sound like they measure, but the HP20s sure seemed to. (It probably helps that designer Paul Barton and I use similar measurement gear.) There’s a mild bass boost centered at 40Hz -- exactly as I heard -- and a lot of energy between 4 and 6.5kHz, which is surely why I occasionally perceived the sound as bright.

Adding 70 ohms to the V-CAN’s output impedance of 5 ohms, to simulate the effects of using a typical low-quality headphone amp, had no effect on the HP20s’ response above 25Hz. So as you plug them into, variously, your smartphone, your laptop, and your high-end headphone amp, the HP20s’ tonal character shouldn’t change.

This comparison of the HP20s with Bowers & Wilkins’ C5 and RBH’s EP-1 suggests that, at least alongside those esteemed competitors, the HP20s’ response is relatively flat, with a more even balance of bass and treble than the two other models. Note the RBHs’ extra bass, and the B&Ws’ relative lack of energy in the treble.

The spectral-decay (waterfall) plot shows a fairly strong resonance at 5kHz and a weaker one at 6kHz, both of which correlate with the response peak in the treble.

The HP20s’ total harmonic distortion (THD) at 90 and 100dBA is very, very low

The spectrum of a 500Hz sinewave suggests that if you push the HP20s really, really loud, you’ll get a roughly equal mix of second- and third-harmonic distortion. But if you play the HP20s at levels high enough to make it audibly distort, you won’t have much hearing left for long.

For reasons I can’t explain, the HP20s delivered superb isolation at the lower frequencies of the audioband, where it really matters (and where jet engines roar): from -10 to -28dB, up to 4kHz. At higher frequencies, however, their isolation was less than the norm.

The HP20s’ impedance magnitude was effectively flat at 16.5 ohms; the impedance phase was also effectively flat.

The HP20s’ average sensitivity, from 300Hz to 3kHz at the rated 16 ohms, measured 106.9dB.

All things considered, nothing in these measurements suggests even the slightest reason for concern.

. . . Brent Butterworth
brentb@soundstagenetwork.com 

Details

Parent Category: Products

Category: Headphone Measurements

Created: 15 March 2014

I measured the performance of the Audeze LCD-3 headphones using a G.R.A.S. 43AG ear/cheek simulator, a Clio FW audio analyzer, a laptop computer running TrueRTA software with an M-Audio MobilePre USB audio interface, and a Musical Fidelity V-CAN headphone amplifier. Measurements were calibrated for ear reference point (ERP), which is roughly the point in space where your palm intersects with the axis of your ear canal when you press your hand against your ear, and the place where the front of the headphones’ driver grilles will sit when you wear them. This is a “flat” measurement: no diffuse-field or free-field compensation curve was used. I experimented with the position of the earpads by moving them around slightly on the ear/cheek simulator, and settled on the positions that gave the best bass response and the most characteristic result overall.

The LCD-3’s frequency response is textbook for planar-magnetic headphones, with essentially flat response below 1kHz, a strong response peak at 2.8kHz, and minor response peaks at 6 and 8.5kHz. This generally conforms to the typical diffuse-field equalization used in many headphones.

Thanks to the resistive impedance of the planar-magnetic driver, adding 70 ohms output impedance to the V-CAN’s 5-ohm output impedance to simulate the effects of using a typical low-quality headphone amp had zero audible effect. I could measure a difference only below 20Hz.

This chart compares the LCD-3 with another highly regarded planar-magnetic headphone, the HiFiMan HE-6, and a respected, new dynamic open-back headphone, the AKG K712. The responses of all three are similar below 1kHz, but between 3.3 and 6.5kHz the HE-6 has a lot more output than the LCD-3, which should make it sound brighter than the Audeze. The K712 should sound substantially different from its planar-magnetic competitors, with less output in the octave between 2.5 and 5kHz.

The spectral-decay (waterfall) plot shows a series of strong but narrow resonances between 2 and 4kHz.

For logistical reasons, I was unable to run a 90dB SPL distortion measurement on the LCD-3, but considering that the 100dB measurement shows near-zero distortion, the 90dB result could only be better.

The LCD-3 being an open-back planar-magnetic headphone, it provides almost no isolation from outside sounds. There is no significant attenuation below 2kHz, and only -5dB of isolation at 5kHz.

The impedance magnitude is essentially flat at 47 ohms, and the impedance phase is at 0 degrees through almost the entire audioband, rising to +5 degrees at 20kHz.

At its claimed impedance of 45 ohms, the LCD-3’s average sensitivity from 300Hz to 3kHz measured 94.5dB.

. . . Brent Butterworth
brentb@soundstagenetwork.com