Headphone Reviews

Sidebar 2: Measurements

The HeadRoom amplifier was non-inverting for all control settings, and its input impedance was a little lower than specification at 21k ohms at 1kHz (footnote 1). Channel separation in Bypass mode was better than 70dB at low frequencies, worsening to 37.5dB at 20kHz due to capacitive coupling between the channels, probably between the parallel pcb tracks carrying the signals to the volume control. The HeadRoom's background noise was a little higher than I expected, its S/N ratio measuring about 62dB (unweighted, 22Hz-22kHz, ref. 1V). (An earlier sample of the Supreme offered about 6dB lower noise.) Nevertheless, I was never bothered by audible noise, even with the volume control full up and the CD player in Pause.

The HeadRoom's output impedance was very low, at 0.4 ohms. (Contrary to what the block diagram implies, negative feedback appears to be taken from the socket end of the current-limiting 8 ohm output resistors.) Its maximum voltage gain into a high-impedance load with Process and Filter bypassed was 16.6dB, which, given its measured maximum voltage output into a 150 ohm load of 7.5V, corresponds to a sensitivity of 1.1V. Fig.1 shows the changes in the THD+noise percentage for different RMS output voltages into 150 and 40 ohms. The negative slope of the curve to the left of this graph reveals that the reading is dominated by noise; as the output level increases, the constant level of noise becomes a smaller percentage of the signal. The sharp knee of the traces is when the true distortion starts to rise due to the amplifier running out of the ability to swing sufficient volts. The extra current demanded from the amplifier by the 40 ohm load cuts the clipping voltage (defined as 1% THD+noise) from 7.5V to 4.4V. (These are equivalent to power levels of 375mW and 129mW, respectively.) The latter is still plenty enough to produce deafening sound levels.

Fig.1 HeadRoom Supreme, distortion+noise vs output voltage into 150 ohms, 40 ohms, and 40 ohms in parallel with 0.22µF.

Also shown in fig.1 is a trace produced when I added a 0.22µF capacitor in parallel with the 40 ohm resistor to see whether the HeadRoom would misbehave with highly capacitive loads. With the exception of a marginally higher distortion level between 2V and 3.6V, this trace exactly overlays the one made with a pure resistive load. Given that this load represents the HeadRoom driving several power amplifiers in parallel at the end of about 700 yards of AudioQuest interconnect cable, this minimal change in clipping behavior would suggest that a good headphone amplifier also makes rather a good line stage (as Melos has discovered).

That the amplifier's intrinsic distortion is very low in level is confirmed by looking at the spectrum of its output when reproducing a low-frequency tone at very high level into 150 ohms (fig.2). The only harmonics popping their heads up above the -100dB level are the second at -95.6dB (0.0015%) and the fourth at -99.3dB (0.001%)—completely inconsequential at these levels, particularly given the sonically benign nature of these two specific harmonics. (As track 21 on Stereophile's Test CD 2 will inform you, even 1% of second harmonic is very hard to hear!) The higher current demand of the 40 ohm load at this 4V level brings up the distortion somewhat (fig.3), the second harmonic rising to -80.5dB (0.01%) and the fourth to -91.8dB (0.0025%). The third and fifth also now make appearances above -100dB [polite applause from the gallery], but I doubt very much if any of this has any impact on subjective quality. "Big volts, little loads—don'tcha just luv 'em!" says the HeadRoom.

Fig.2 HeadRoom Supreme, spectrum of 50Hz sinewave, DC-1kHz, at 4V into 150 ohms (linear frequency scale). Note that the second harmonic at 100Hz is the highest in level, 95.6dB below the level of the 50Hz fundamental (0.0015%).

Fig.3 HeadRoom Supreme, spectrum of 50Hz sinewave, DC-1kHz, at 4V into 40 ohms (linear frequency scale). Note that the second harmonic has now risen to -80.5dB (0.01%).

This predominantly second-harmonic-plus-noise signature is confirmed by fig.4, which shows the distortion waveform with the amplifier working hard driving 1kHz at 4V into 40 ohms. (I averaged 32 individual measurements to produce this graph, dropping the signal's noise content by 15dB to enable its harmonic content to emerge.)

Fig.4 HeadRoom Supreme, 1kHz waveform at 4V into 40 ohms (top); distortion and noise waveform with fundamental notched out (bottom).

---

Footnote 1: Unless stated otherwise, all the measurements refer to the HeadRoom Supreme amplifier fitted with the 30k/5600pF EQ circuitry appropriate for the Sennheiser '580 cans.