Details

Parent Category: Products

Category: Headphone Measurements

Created: 10 July 2018

Reviewed: SoundStage! Solo, Brent Butterworth, July 2017

I measured the ATH-ADX5000s using a G.R.A.S. Model 43AG ear/cheek simulator/RA0402 ear simulator, a Clio 10 FW audio analyzer, a laptop computer running TrueRTA software with an M-Audio MobilePre USB audio interface, a Musical Fidelity V-CAN amp, and an Audio-gd NFB-1AMP for distortion measurements. On the Model 43AG, I used the original KB0065 simulated pinna for most measurements as well as the new KB5000 pinna for certain measurements, as noted. These are “flat” measurements; no diffuse-field or free-field compensation curve was employed.

Although the ATH-ADX5000s’ frequency response shows that its tonal balance is fairly similar to those of typical competitors, it does reveal some interesting idiosyncrasies. First is the midrange bump at about 1.3kHz. Then there’s the lack of the usual peak between 2 and 3kHz, which is generally considered necessary for headphones to deliver a sound approximating that of speakers in a room. However, there are strong peaks at 3.4 and 6.3kHz. I can’t remember seeing a response quite like this before, so I can’t confidently predict how it will sound, but if someone showed me a response curve like this and asked me to interpret it, I’d guess it would sound fairly flat but just a little bass-shy.

Note that this is the best match I was able to get between the left and right channels, after about 20 minutes of experimentation. Getting these measurements to match is always a challenge because slight changes in headphone position change the measured response, and the left and right simulated pinnae (which are based on averages of molds made from hundreds of human ears) aren’t perfect mirror images of each other (neither are your ears, for that matter). So I’m always reluctant to criticize headphones for channel matching. I didn’t notice a mismatch when listening, but still, this doesn’t impress me.

This chart shows the ATH-ADX5000s’ right-channel frequency response measured with the old KB0065 pinna (which I’ve used for years) and G.R.A.S.’s new KB5000 pinna, which I’ll be switching to because it more accurately reflects the structure and pliability of the human ear. (I include this mostly for future reference rather than as something you should draw conclusions from; I intend to show both measurements in every review until later this year, when I begin using only the new pinna.)

Here you can see how the ATH-ADX5000s’ tonal balance changes when they’re used with a high-impedance source, such as a cheap laptop or some cheap professional headphone amps. It’s not a big difference, largely because the ATH-ADX5000s’ high impedance means that changes in the source impedance don’t have as big an effect. But the headphones’ uneven impedance curve (see below) does result in a bass boost of about 1dB and a reduction of about 1dB in the lower treble, between 2 and 3kHz. Not a huge difference, but still an audible one.

This chart shows how the ATH-ADX5000s compare with several competing open-back headphones: the Audeze LCD-Xes, the HiFiMan HE1000 V2s, and the Sennheiser HD 800 Ses. While the overall balance is similar in all of these models, the ATH-ADX5000s’ competitors clearly have less energy between 1 and 2kHz, and more between 2 and 3kHz -- and that’s right in the “sweet spot” of human hearing, so the Audio-Technicas will certainly sound a little different.

The ATH-ADX5000s’ spectral decay (waterfall) chart shows no high-amplitude resonances, but many high-Q (i.e., narrow), very-low-amplitude (about -40dB) resonances between 1 and 5kHz. This is typical of open-back models; it doesn’t seem to affect their tonal balance, but I speculate that it’s part of what gives them a more spacious sound. Incidentally, I put 4” of denim insulation on top of the headphones when I take this measurement, so the resonances don’t represent room reflections of the sound coming off the back of the headphones.

The measured total harmonic distortion (THD) of the ATH-ADX5000s is low at any sane listening level. At 90dBA, which is quite loud, the THD rises to only about 1.5% at 20Hz. At the extremely loud level of 100dBA, it’s about 1.5% between 100 and 400Hz, and rises to about 4.7% at 20Hz.

In this chart the external noise level is 85dB SPL; the numbers below that indicate the degree of attenuation of outside sounds. Like the other open-back models (the Audeze LCD-Xes and the HiFiMan HE1000 V2s), the ATH-ADX5000s provide no significant isolation. It’s interesting, though, to see how the ATH-ADX5000s and HE1000 V2s -- both of which have very open backs -- deliver less isolation than the LCD-Xes. If you want isolation, you’ll have to go with a closed-back model such as the Audeze LCD-XCs (also shown).

To the best of my memory, all over-ear and on-ear dynamic headphones I’ve tested have an impedance bump at the driver resonance (always in the bass); nonetheless, the ATH-ADX5000s’ is pretty extreme. The nominal impedance is about the same as the rated impedance of 420 ohms, but the impedance breaks 1400 ohms at 90Hz. Fortunately, because the overall impedance is high, this doesn’t cause major changes in tonal balance if a high-impedance source device is used. The phase response is pretty flat, though.

The sensitivity of the ATH-ADX5000s, measured between 300Hz and 3kHz with a 1mW signal calculated for 420 ohms impedance, is 101.8dB, which is 1.8dB higher than specified. This means that even though they weren’t designed for the purpose, the ATH-ADX5000s will work reasonably well with low-quality source devices.

. . . Brent Butterworth
brentb@soundstagenetwork.com

Details

Parent Category: Products

Category: Headphone Measurements

Created: 15 June 2018

I measured the Marshall Mid A.N.C.s using a G.R.A.S. Model 43AG ear/cheek simulator/RA0402 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 amp. On the Model 43AG, I used the original G.R.A.S. KB0065 simulated pinna for most measurements, as well as the new KB5000 pinna for certain measurements, as noted. For tests in Bluetooth mode, I used a Sony HWS-BTA2W Bluetooth transmitter to send signals from the Clio 10 FW to the headphones. These are “flat” measurements; no diffuse-field or free-field compensation curve was employed.

The Mid A.N.C.s’ frequency response (shown here with NC on, and a wired connection) looks fairly standard, except that the 3kHz peak found in most headphones is unusually strong. In my experience, that extra couple dB is likely to make headphones sound just a bit on the bright side.

This chart shows the right-channel frequency response of the Mid A.N.C.s measured in some of its various operating modes: wired passive (NC off), wired (NC on), and wired Bluetooth (NC on). They’re pretty close to each other, which is a good thing -- the Mid A.N.C.s should sound pretty similar no matter what mode you’re in. The Bluetooth mode looks as if it might sound a bit softer, but the gating required for the Clio analyzer to compensate for Bluetooth’s latency might be contributing to that effect in the measurement.

This chart shows the Mid A.N.C.s’ measured right-channel frequency response in wired mode with NC on, measured with the old KB0065 pinna (which I’ve used for years) and G.R.A.S.’s new KB5000 pinna (which I’ll eventually switch to because it more accurately reflects the structure and pliability of the human ear). I include this mostly for future reference rather than as something you should draw conclusions from; I intend to show both measurements in every review until I completely switch to the new pinna.

This chart shows the Mid A.N.C.s’ measured right-channel frequency response compared with three other noise-canceling headphones: the PSB M4U 8s, the Sony WH-1000X Mk.2s, and the Bose QC25s. Clearly, the Mid A.N.C.s have a little more energy between about 600Hz and 3.5kHz, with no extra bass to balance it out, so they’re likely to sound just a bit bright.

This spectral-decay (waterfall) chart shows the results in wired mode; the latency introduced by Bluetooth prevented me from getting a reliable measurement in that mode. While the resonance in the bass is unusually low, there’s a somewhat strong resonance centered at about 4kHz that doesn’t seem to correspond with the frequency-response measurements. Will this cause a slight brightness? Possibly.

Bluetooth’s latency meant that I had to measure the total harmonic distortion (THD) of the Mid A.N.C.s in wired mode. Distortion is a little higher than the norm, though the problem seems restricted to the bass; remember, most music has very little content below 40Hz.

In this chart, the external noise level is 85dB SPL; the numbers below that indicate the degree of attenuation of outside sounds. The isolation of the Mid A.N.C.s is just about the same as that of another on-ear NC model I recently measured, the AKG N60 NC Wireless. Not surprisingly, neither can touch the over-ear Bose QC35 IIs, but they provide enough NC to make a plane ride much more pleasant.

The Mid A.N.C.s’ impedance magnitude in wired mode is dead flat at 38 ohms, with a nearly flat phase response. In active mode with NC on, it’s dead flat at 880 ohms, with flat phase response.

The sensitivity of the Mid A.N.C.s, measured between 300Hz and 3kHz with a 1mW signal calculated for the specified 32 ohms impedance, is 98.0dB in wired passive mode, and 102.2dB in wired active mode with NC on. They won’t deliver enough volume to blast out your eardrums, but you’ll have plenty enough for watching a movie on an airplane.

. . . Brent Butterworth
brentb@soundstagenetwork.com

Details

Parent Category: Products

Category: Headphone Measurements

Created: 15 June 2018

I measured the HD 1 Frees using a G.R.A.S. Model 43AG ear/cheek simulator plus RA0402 ear simulator and KB5000 simulated pinna, a Clio 10 FW audio analyzer, and a laptop computer running TrueRTA software with an M-Audio MobilePre USB audio interface. I used a Sony HWS-BTA2W Bluetooth transmitter to send signals from the Clio 10 FW to the headphones. These are “flat” measurements; no diffuse-field or free-field compensation curve was employed. Note that some of the measurements I usually perform are not included here; because the HD 1 Frees have only Bluetooth input, my analyzer was unable to measure their spectral decay; and, of course, impedance and sensitivity measurements are irrelevant for wireless-only earphones.

The frequency response of the HD 1 Frees is normal in having peaks around 3 and 6.5kHz, but unusual in that the 6.5kHz peak is so much higher in amplitude than the 3kHz peak. Normally, it’s the other way around. Although it’s tough even for experienced technicians to know precisely how headphones or earphones will sound from their measurements, I speculate from these that the HD 1 Frees will sound a bit recessed in the upper midrange and lower treble, and unusually strong in the mid-treble. The broad bump in the bass is fairly standard for dynamic earphones.

This chart shows the HD 1 Frees’ measured right-channel frequency response with Bluetooth (BT) and noise canceling (NC) on, measured with the stainless-steel coupler included with the RA0402 ear simulator, and with the addition of G.R.A.S.’s new KB5000 pinna, which I’ll eventually be switching to because it more accurately reflects the structure and pliability of the human ear. I include this mostly for future reference rather than as something you should draw conclusions from; I intend to show both measurements in every review until I completely switch to the new pinna.

This chart shows how the HD 1 Frees differ from a few passive earphones I’ve reviewed. (I had no measurements of other Bluetooth earphones available.) Clearly, there’s a little extra energy around 50Hz and 6.5kHz, and a little less than usual around 3kHz.

Because of Bluetooth’s latency problem, I had to measure the total harmonic distortion (THD) of the HD 1 Frees using discrete tones in one-octave steps rather than swept tones; still, the results should be comparable to my usual distortion measurements. The distortion is very low, at less than 0.5% -- so low that I had to adjust my usual Y-axis scale down from its usual maximum value of 50%. Note that I tested a couple of other Bluetooth earphones at the same time, and got distortion numbers in the more usual range of 2-4% -- this measurement does appear to be genuinely excellent, not just a result of the different measurement technique.

In this chart, the level of external noise is 85dB SPL; the numbers below that indicate the degree of attenuation of outside sounds. Clearly, the isolation offered by the HD 1 Frees seems less than average; if this is a concern for you, I recommend replacing the stock eartips with Comply foam tips.

The HD 1 Frees’ range of Bluetooth operation was outstanding: I measured reliable line-of-sight reception at 33’ indoors -- three times farther than some Bluetooth headphones and earphones can manage.

. . . Brent Butterworth
brentb@soundstagenetwork.com

Details

Parent Category: Products

Category: Headphone Measurements

Created: 01 June 2018

I measured the Campfire Comets using a G.R.A.S. Model 43AG ear/cheek simulator (including the RA0402 high-resolution 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 amplifier. I used the RA0402 as a direct coupler for most of the measurements, adding the Model 43AG plus the G.R.A.S. KB5000 simulated pinna for certain measurements, as noted. I used the included medium-size silicone tips for measurements with the coupler, and the large foam tips for measurements with the ear/cheek simulator. These are “flat” measurements; no diffuse-field or free-field compensation curve was employed.

The Comets’ response is unusual for earphones in that it’s much flatter than most; most have a broad boost in the bass and a stronger peak in the 3kHz region. That’s generally considered to give the most satisfying response, but headphones that don’t match that response can also sound great. The peak centered at 3.2kHz is narrower than usual, and followed by a much broader peak stretching from about 6.5 to 10kHz. That peak is probably why I perceived these headphones as sounding detailed and slightly bright, but not harsh or edgy.

This chart shows the Comets’ right-channel frequency response measured using only the RA0402 coupler (which has a stainless-steel tube into which earphones fit), and measured using the Model 43AG ear/cheek simulator with G.R.A.S.’s new KB5000 pinna, which I’ll be switching to eventually for my earphone measurements because it’s a more realistic representation of the acoustical environment presented by the human ear. (I include this for future reference; I intend to include both measurements until I completely switch to the new pinna.)

This chart shows the results of 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. As with almost all balanced-armature earphones, the Comets’ large impedance swings interact with the output impedance of the source device to change the response. So if you use a source with a relatively high output impedance (about 50 ohms or higher), such as a cheap laptop or smartphone, the Comets will likely sound substantially brighter.

This chart compares the Comets’ measured right-channel frequency response with that of two other multidriver earphones: the 1More Quad Drivers and the PSB M4U 4s. Both competitors have a large bump in the bass and much higher average treble energy compared with the Comets, but even that sort of response can sound balanced if the bass bump is in suitable proportion to the treble peak.

The spectral-decay (waterfall) chart shows that the Comets’ resonances are negligible.

The total harmonic distortion (THD) of the Comets is high for earphones. At 90dBA, it rises over a two-octave-wide midrange band to a peak of 2%; at 100dBA, it’s anywhere from 1.5 to 9%. (To be sure of the results, I repeated these measurements three times.) Note: These are extremely loud levels; I heard no distortion when I was auditioning the Comets; and scientific research shows that distortion in headphones is only rarely audible.

In this chart, the level of external noise is 85dB SPL; the numbers below that indicate the degree of attenuation of outside sounds. I included measurements with the silicone and foam eartips because the choice of tip makes such a big difference with the Comets, probably because of the size of their earpieces. My guess is that the tiny earpiece lets the foam get farther into the ear canal, where it can form an exceptionally tight seal -- the best I can remember measuring from an earphone that doesn’t use over-ear cable routing or active noise canceling. My listening impressions match the measured result.

The Comets’ impedance, like that of almost every balanced-armature earphone, varies considerably with frequency, rising from 22 ohms in the bass to over 300 ohms in the treble. The phase shift is also large. I recommend using a low-impedance source with these; e.g., an iPhone, a higher-end Android phone, a decent portable music player, or a portable DAC-headphone amp.

The Comets’ sensitivity, measured from 300Hz to 3kHz with a 1mW signal at the specified 20 ohms impedance, is 107.8dB. You should get plenty of volume from them with any source device.

. . . Brent Butterworth
brentb@soundstagenetwork.com

Details

Parent Category: Products

Category: Headphone Measurements

Created: 15 May 2018

I measured the Aventho Wireless headphones using a G.R.A.S. Model 43AG ear/cheek simulator/RA0402 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 amp. On the Model 43AG, I used the original KB0065 simulated pinna for most measurements, as well as the new KB5000 pinna for certain measurements, as noted. For measurements using a Bluetooth connection, I used my Sony HWS-BTA2W Bluetooth transmitter to send signals from the Clio 10 FW to the headphones. These are “flat” measurements; no diffuse-field or free-field compensation curve was employed.

The Aventho Wirelesses’ frequency response (shown here in wired mode; I was unable to get consistent measurements from the left and right channels in Bluetooth mode) is a little unusual. A typical headphone response might have a mild, broad boost in the bass between about 50 and 200Hz, with a strong, distinct response peak at around 2.5 or 3kHz, and a weaker peak or two between 5 and 10kHz. The Aventhos instead have a relatively narrow peak at 150Hz, building gradually to a softer peak at about 2.6kHz. I’ve measured few headphones with such a frequency response, so it’s hard for me to predict what these will sound like based on these measurements.

This chart shows the right-channel frequency responses of the Aventho Wirelesses, measured with wired and wireless Bluetooth connections. The Bluetooth connection seems to have a few dB more average output in the bass, as well as weaker output above about 7kHz. This suggests that the Aventhos will sound softer in Bluetooth mode, but the gating required to compensate for Bluetooth’s latency does introduce some uncertainty into this measurement.

This chart shows the Aventhos’ right-channel frequency response measured with the old KB0065 pinna I’ve used for years, and with G.R.A.S.’s new KB5000 pinna, which I’ll be switching to because it more accurately reflects the structure and pliability of the human ear. I include this mostly for future reference rather than as something you should draw conclusions from; I intend to show both measurements in every review until I completely switch to the new pinna later this year.

This chart shows the Aventhos’ measured right-channel frequency response compared with those of another set of on-ear headphones of similar size (the AKG N60 NC Wirelesses, with their noise canceling activated) and a well-known reference for affordable passive headphones (the over-ear NAD Viso HP50s). Clearly, the Aventhos’ response is flatter overall, with a much less prominent peak in the 3kHz range. However, the Beyerdynamics’ bass rolloff may counteract their reduced treble response to create a subjectively flat response.

The Aventho Wirelesses’ spectral-decay (waterfall) chart looks very clean, with no significant resonances above about 600Hz, and lower-than-average resonance in the bass frequencies.

The total harmonic distortion (THD) of the Aventho Wirelesses, measured in wired mode because Bluetooth’s latency prevents my audio analyzer from doing distortion measurements, is a little on the high side, although it’s unlikely to be audible at normal listening levels. At 90dBA, the THD rises to about 1% at 100Hz and 4% at 20kHz. At the extremely loud level of 100dBA, it’s about 3% at 100Hz and 10.5% at 20Hz. That probably would be audible, but 100dBA is so loud that your ears will be strained, anyway.

In this chart, the external noise level is 85dB SPL; the numbers below that indicate the attenuation of outside sounds. The isolation of the Aventho Wirelesses is pretty good for an on-ear model with no active noise canceling, and is comparable to that of the over-ear NAD Viso HP50s. You can see from the isolation traces for the AKG N60 NC Wirelesses and Bose QC35IIs that active noise canceling will improve isolation at frequencies below about 1kHz.

The Aventhos’ impedance magnitude is the same with their power on or off, and ranges between 33 and 39 ohms. The phase response is similarly flat.

The sensitivity of the Aventho Wirelesses in wired mode, measured between 300Hz and 3kHz with a 1mW signal calculated for 32 ohms impedance (my default for internally powered headphones of no specified impedance), is 106.1dB. The Beyerdynamic Aventhos should give you plenty of volume when plugged into an airplane seat’s headphone jack, even when the battery is low.

. . . Brent Butterworth
brentb@soundstagenetwork.com

Details

Parent Category: Products

Category: Headphone Measurements

Created: 01 May 2018

I measured the PSB M4U 8s using a G.R.A.S. Model 43AG ear/cheek simulator and RA0402 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 amp. On the Model 43AG I used the original G.R.A.S. KB0065 simulated pinna for most measurements, as well as the new KB5000 pinna for other measurements, as noted. For tests in Bluetooth mode, I used a Sony HWS-BTA2W Bluetooth transmitter to send signals from the Clio 10 FW to the headphones. These are “flat” measurements; no diffuse-field or free-field compensation curve was employed.

The M4U 8s’ frequency response is pretty much in accordance with contemporary ideas of which headphone FR is most pleasing to most listeners: fairly close to the so-called “Harman curve.” The bass is shelved up below 120Hz, and there’s a somewhat larger-than-usual peak at 3.5kHz; normally, this peak might be centered at more like 2.5kHz, and it might be a few dB lower in magnitude. This is almost certainly why I found the treble a bit bright overall. Note that the left channel did not exhibit this effect; while it’s possible that the two channels measure differently (a common occurrence in active headphones, because the acoustics can be different due to the batteries being on only one side, etc.), this headphone’s response is tuned with digital signal processing (DSP), so I suspect that this may be a measurement anomaly. However, none of my attempts to reposition the earpiece on the ear/cheek simulator were able to match the peak in the right channel.

This chart shows the right-channel frequency response of the M4U 8s measured in some of their operating modes: wired passive, wired active, wired active with NC on, and wired Bluetooth with NC off. You can see that only the wired passive mode deviates greatly from the others, which is no surprise -- it bypasses the active circuitry and the DSP used to tune the other modes. Switching on NC seems to boost the bass by a couple dB, and the top and bottom look slightly rolled off in Bluetooth mode -- although Bluetooth’s latency can introduce measurement anomalies, so I’m not 100% confident in this measurement.

This chart shows the M4U 8s’ measured right-channel frequency response in wired mode with NC on, measured with the old KB0065 pinna (which I’ve used for years) and G.R.A.S.’s new KB5000 pinna, which I’ll eventually be switching to because it more accurately reflects the structure and pliability of the human ear. (I include this mostly for future reference rather than as something you should draw conclusions from; I intend to show both measurements in every review until I completely switch to the new pinna later this year.)

This chart shows the M4U 8s’ measured right-channel frequency response compared with those of three other NC headphones: the PSB M4U 2s, the Sony WH-1000X Mk.2s, and the Bose QC25s. The M4U 8s deviate from the norm in only one clear way -- they have less midrange response between about 150Hz and 1.8kHz. In my experience, this doesn’t mean the midrange will necessarily sound recessed, but it does make the treble peak more subjectively prominent, even if it’s not really higher relative to the 500Hz normalization point used for this graph.

This spectral decay (waterfall) chart shows the results in passive mode; the results in active mode are similar, but the latency introduced by the DSP prevented me from getting a chart comparable to those I typically publish. There’s a bit of resonance in the bass, but it’s along the lines of what I’ve measured from many other over-ear, closed-back headphones. There’s also a series extremely high-Q (i.e., narrow bandwidth), low-magnitude resonances in the treble that correspond with the M4U 8s’ measured response peaks; I see this effect in most of the open-back planar-magnetic headphones I measure, but rarely in closed-back dynamic designs. While the response peak is certainly audible, these resonances are narrow, and low enough in magnitude, that I doubt they’d be audible.

Because I had to measure the total harmonic distortion (THD) of the M4U 8s in passive mode, again because of latency problems, this measurement shows only the distortion of the driver. It’s very low, rising to just 1% at 20Hz even at extremely loud listening levels.

In this chart the external noise level is 85dB SPL; the numbers below that indicate the degree of attenuation of outside sounds. (Note that I recently switched from measuring at a level of 75dB to 85dB; this doesn’t change the way the isolation curves look, but an 85dB level allows me to get better measurements of NC headphones, which demand a lower noise floor.) The isolation of the M4U 8s isn’t exceptional, but it does get down into the ballpark of some of the better NC headphones; Sony’s WH-1000X Mk.2s beat it by a hair, and the Bose QC35 IIs by a few dB more, though the Boses (which have the best NC performance I’ve measured from over-ear headphones) are clearly much better -- at the cost of the “eardrum suck” noted in the review.

The M4U 8s’ impedance magnitude in wired passive mode averages about 38 ohms, with a nearly flat phase response. The impedance in the wired active mode was above the Clio 10 FW’s 1500-ohm limit for impedance measurements.

The sensitivity of the M4U 8s, measured between 300Hz and 3kHz with a 1mW signal calculated for the specified 32 ohms impedance, is 108.2dB in wired passive mode, or 110.8dB in wired active mode with NC on. The M4U 8s should deliver ample volume from any source device.

. . . Brent Butterworth
brentb@soundstagenetwork.com

Details

Parent Category: Products

Category: Headphone Measurements

Created: 15 April 2018

I measured the Massdrop x NuForce EDC3 in-ear headphones using a G.R.A.S. Model 43AG ear/cheek simulator (including the RA0402 high-resolution 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 amplifier. On the Model 43AG, I used the original KB0065 simulated pinna for most measurements as well as the new KB5000 pinna for certain measurements, as noted. These measurements used the medium-sized silicone eartips; for isolation measurements, I tried both the silicone tips and the largest of the included foam tips. These are “flat” measurements; no diffuse-field or free-field compensation curve was employed.

I can’t recall measuring a set of earphones with a response like that of the EDC3s. It looks more like the response of a set of large, open-back, planar-magnetic headphones: essentially flat up to about 1kHz, then rising to a broad peak between 1.4 and 3.2kHz, then to a couple of smaller peaks between 4 and 10kHz. That’s not necessarily to say that the EDC3s will sound like big planar-magnetic headphones -- they’re inserted into the ear canals, and thus bypass the acoustical effects of the pinna and the outer portion of the ear canal, which of course affect the sound quality of over-ear headphones.

This chart shows the EDC3s’ right-channel frequency response measured using only the RA0402 coupler (which has a stainless-steel tube into which an earphone fits), and measured using the Model 43AG ear/cheek simulator with G.R.A.S.’s new KB5000 pinna (which I’ll eventually switch to for my earphone measurements because it’s a more realistic representation of the acoustical environment presented by the human ear). I include this for future reference; I intend to show both measurements until I begin using only the new pinna.

This chart shows the results of 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. As with every set of balanced-armature earphones I’ve measured, the EDC3s’ large swings in impedance (see impedance graph below) interact with the output impedance of the source device to change the response. Here, however, the effect is more subtle than I usually see -- you’ll get just a little less bass and a little more treble if you use a lower-quality headphone amp with a relatively high output impedance.

This chart compares the EDC3s’ measured right-channel frequency response with that of three other multidriver earphones: the 1More Quad Drivers, the PSB M4U 4s, and the Shure SE215s. Note how, compared with the EDC3s, all three competitors exhibit a large bump in the bass and a higher, more pronounced peak in the lower treble.

The ECD3s’ spectral-decay (waterfall) chart shows very low resonance overall.

The ECD3s’ total harmonic distortion (THD) is pretty low, and a bit unusual. At the testing levels I use (both are high relative to normal listening levels), there seems to be a flat 1% THD below 1kHz -- yet even at the very high testing level of 100dBA, the distortion doesn’t significantly rise. THD of 1% is common in transducers and generally not noticeable, but this test provides yet another indication that something a bit different is going on in the EDC3s.

In this chart, the external noise level is 85dB SPL; the numbers below that indicate the degree of attenuation of outside sounds. (Note that I recently switched to measuring at a level of 85dB instead of 75dB; this doesn’t change the way the isolation curves look, but an 85dB level lets me get better measurements of noise-canceling headphones, which demand a lower noise floor.) The EDC3s deliver excellent isolation, especially when their included foam eartips are used. Even by the generally high standards of earphones with over-ear cable routing, this is an excellent result.

The EDC3s’ impedance, like that of all the balanced-armature earphones I can remember measuring, has a large swing, rising from 16 ohms in the bass to 53 ohms in the treble. The phase shift, too, is considerable, going from 0° in the bass to +72° at 20kHz. This is the reason for the change in sound quality when you switch from a low-impedance source (such as an iPhone or a typical good headphone amp) to a high-impedance source (e.g., the headphone amps built into most cheap laptops).

The sensitivity of the Massdrop x NuForce EDC3s, measured between 300Hz and 3kHz with a 1mW signal at their specified impedance of 20 ohms, is 109.1dB. This means that you should get plenty of volume from any source device.

. . . Brent Butterworth
brentb@soundstagenetwork.com

Details

Parent Category: Products

Category: Headphone Measurements

Created: 01 April 2018

I measured the effects of Sonarworks’ True-Fi processing using a G.R.A.S. Model 43AG ear/cheek simulator (including the RA040X high-resolution ear simulator and KB5000 pinna), a laptop computer running TrueRTA software with an M-Audio MobilePre USB audio interface, and a Musical Fidelity V-CAN amplifier. Because the test files had to be processed through the Sonarworks app, I had to play a pink-noise file as the stimulus and use TrueRTA, instead of generating and analyzing signals using my usual Audiomatica Clio 10 FW analyzer. These are “flat” measurements; no diffuse-field or free-field compensation curve was employed.

This chart shows the frequency response of the Audeze LCD-X headphones measured with and without True-Fi processing. True-Fi introduces a bass boost of about 7dB centered at 35Hz, and a treble boost of about 5dB maximum and centered at 3kHz, as well as a 4dB boost centered at 7.5kHz.

This chart shows the frequency response of the Bose QC20 earphones with noise canceling on, measured with and without True-Fi processing. True-Fi cuts the bass by about 7dB centered at 40Hz, and introduces treble cuts of about 4dB maximum centered at 3.1kHz, and 9dB centered at 8kHz.

This chart shows the frequency response of the Sony MDR-7506 headphones, measured with and without True-Fi processing. Here the effect is more subtle than with the above headphones and earphones. True-Fi cuts the bass by about 2dB centered at 60Hz, boosts the lower mids by about 4dB centered at 200Hz, and reduces the treble response by about 2dB on average between 4 and 7kHz.

Here you can compare the responses of the Audeze LCD-4, Sony MDR-7506, and Status Audio CB-1 headphones after True-Fi processing. It seems the target response is different for each design; if it were the same, the traces would overlap except for a few spurious variances caused by differences in the physical design of the headphones and the fit of their earpieces on the ear/cheek simulator.

In this chart you can see the effects of True-Fi’s Adjust for Age feature, with the intensity set to 50%, and the correction set for 55-year-old males and females and a 35-year-old man. Interestingly, this control introduces not only a treble rolloff, but also some effects on the bass and on the overall listening level.

What constitutes a “correct” measurement for the results above is very much a matter of debate, but these measurements do indicate that True-Fi’s corrections are not (at least in the cases of these headphone models) extreme or unexpected, and that Sonarworks does seem to have put some serious thought and work into each of the correction curves, rather than merely making each headphone model conform to the same curve.

. . . Brent Butterworth
brentb@soundstagenetwork.com