As an accompaniment to the subjective tests in this month’s PAR Session Trial, we made inquiries into the methods used for quality-control assessment by manufacturers, resulting in an invitation to utilize Audio-Technica resources at its Stow, OH headquarters. The headphones tested were a second set of the makes and models reviewed by Rob Tavaglione, each fresh out the manufacturers’ packaging.
Headphone measurement starts with a reference microphone, but requires additional physical interface for coupling and isolation. Audio-Technica employs a Bruel & Kjær 4153 Artificial Ear, essentially a flat plate with a central mounting for the B&K microphone element. The mic is coupled to an Audio Precision System One test set via a B&K 2608 measuring amplifier. A custom software program controls the System One. Each headset was manually centered on the artificial ear and then the tests run producing a frequency-response plot and measuring sensitivity (output in dB SPL with a 1 mW 1 kHz tone) and impedance. The tests were run on both left and right earpieces. The artificial ear was then moved into an anechoic chamber where isolation was measured relative to frequency with a 94 dB SPL loudspeaker source. For this test, the mic is calibrated for a flat response prior to covering the mic element with a given headset — the measurement being the passive attenuation of outside stimulus.
Plot 1: Frequency response sweeps and measured performance. I brought my Prism Sound dScope III test set along as well. The dScope analog outputs drove a Rane HC6 headphone amp for 10 dB of gain; the headphone output was measured using the same microphone interface used for the frequency-response tests. The resulting distortion measurements thus included the performance of the HC6 and should be considered comparatively, not as absolute numbers.
The B&K ‘Artificial Ear’ test microphone and physical interface with a pair of Audio-Technica MTH-50Ss mounted for evaluation. I have personally, as yet, only listened to one of these specific headphone models, but over time I have listened to other models from each of the brands represented. Based on those prior experiences, I expected that there would be significant differences in frequency response between the headphones tested based on design and components used and/or due to deliberate voicing. Despite those expectations, I was still surprised by the degree of both relative differences and the degree of deviation from flat performance. Printed specifications for headphones are generally vague with little (if any) qualification against reference levels or other qualifying data.
A-T’s test procedure for frequency response (and later, isolation) produced a log plot of the measured data, but also output the raw measurement data. Despite not being able to create logarithmic plots of the measurements, I imported the raw data to Microsoft Excel so that the test data could be graphed in a single illustration [see Plot 1]. Even though the Ultrasone Edition 8s should probably be considered as outside the test group, I left its data in with the rest. The general shapes of the curves tend to grossly coincide: boosted low end, relatively flat from about 200 to 2,500 Hz, plus an HF bump (or bumps).
Within that trend, there’s a lot of variation and a lot of deviation from flat response. Comparing right-channel performance only, the AKG K702s were the flattest to about 3 kHz (+4/-2.7 dB) though having one of the highest HF peaks of +13.5 dB around 6,800 Hz. The Shure SRH840s have the smoothest overall response (less smaller peaks and valleys) and the flattest from 250 to 2,000 Hz, though they have significant low-end shelf around +5 dB up to 70 Hz, a broad peak of +9.5 centered just above 100 Hz and a major peak on the higher end of +12.4 dB at 9,100 Hz. Across a 30 Hz to 20 kHz bandwidth, using the cumulative maximum positive to maximum negative deviation as the measure [see Chart 1], the Shures have the lowest total deviation at 12.8 dB, followed by the Ultrasone HFI- 680s, then the Sony, AKG, Audio-Technica models in that order. The ultra HiFi Ultrasone Edition 8s scored last in this test, with a voicing that looks designed for a wow factor.
Plot 2: Testing passive isolation vs. frequency. Looking at left-channel data highlights inconsistency in many of the left/right driver pairs as tested, most significantly with the HFI-680s, where an anomalouslooking sharp valley was measured, -17.5 dB down centered at 4,750 Hz. Comparing each headphone set’s left and right channel performance atop one another, the Sonys and A-Ts (followed pretty closely by the AKGs) track the closest together on the high end, where you might expect the relative left/right performance to affect imaging. All in all, these are some pretty eyeopening response curves.
The Edition 8s had the greatest isolation by far — instantly noticeable even before measurement if you simply put the phones on your head. The AKG K702s are open-air phones, so they can’t be expected to be a competitor in this test. The Shures performed the best of the studio headphones, with the A-T and Ultrasone HFI sets’ isolation versus frequency graphs intertwining [see Plot 2]. The Sony’s joined to make it a three-way, neckand- neck competition above about 600 Hz.
Plot 3 graphs the relative output level and the THD+N (Total Harmonic Distortion + Noise) of the various phones with a 1 kHz input signal swept from 0 to 20 dBu. The top group of six lines is generally diagonal (indicating a linear relationship from input to relative output) until the headphones begin to overload, with the bottom group of traces showing THD+N, which begins to climb as the headphone performance becomes nonlinear. The Ultrasone headphones stand out as having the highest initial distortion levels and a more rapid increase in distortion as the tone got louder; the rest of the pack is generally similar, distortion remaining at nominal levels until the onset of clipping, the differences in the curves representing different sensitivity and maximum output/input handling.
Plot 3: Relative output levels (top traces) and THD+N performance with a level swept 1 kHz input. Color code: AKG-blue, Audio-Technica-red, Shure-green, Sony-orange, Ultrasone (HFI)- violet, Ultrasone (Edition 8)-olive. The Sony and Audio-Technica models get the nod here for the best combination of output level/efficiency and distortion performance, the Sony’s getting somewhat louder, the A-Ts besting the pack on distortion performance. I also ran FFT plots of the phones with 0 and +10 dBu drive levels out of the Rane HC6, the Audio- Technica and AKG phones showing little aberration but baseband noise at the lower level, the Sony, Shure and Ultrasone HFI sets a bit of 2nd and 3rd harmonic. The Ultrasone Edition 8s had significant 2nd harmonic component (typically perceived as euphonic) even at the lower drive level.
For help in facilitating our tests, a big thank you is due Audio-Technica’s Jeff Firzlaff quality assurance manager, and quality department group leader Mark Shaw.
Frank Wells, formerly a radio broadcast and recording studio technician was the editorial director of Pro Audio Review at the time of this publication.