Digital Processor Reviews

I measured the iFi ZEN CAN's and ZEN DAC Signature's performance with my Audio Precision SYS2722 system (see the January 2008 As We See It).

Looking first at the ZEN CAN, I measured two versions. One, labeled "6XX," with the serial number 5501000735, included equalization for specific Sennheiser headphones. The other, labeled "HFM," with the serial number 5312001087, included equalization for specific HiFiMan headphones.

Other than the equalization, the two versions of the ZEN CAN measured identically. The maximum gain for unbalanced input signals was 0dB, 6dB, 12dB, or 18dB, as specified, from the single-ended headphone output and 6dB higher in each case from the balanced headphone output. Both the balanced and unbalanced outputs preserved absolute polarity (ie, were noninverting). The unbalanced input impedance of both samples was >400k ohms at 20Hz and 1kHz, dropping to a still high 78k ohms at 20kHz. (I wasn't able to measure the balanced input impedance, as the Pentaconn adapter cable I bought from Amazon had the wrong gender XLR plugs, and I didn't have time to order another adapter before the review samples had to be returned to Julie Mullins.) The ZEN CAN's unbalanced output impedance was a very low 0.75 ohm from 20Hz to 20kHz. The balanced output impedance was 1 ohm, again from 20Hz to 20kHz.

With equalization bypassed, both versions of the ZEN CAN had an identical frequency response: flat from 10Hz to 50kHz from both balanced and unbalanced outputs (fig.1, cyan and magenta traces), with the ultrasonic rolloff reaching –1dB at 150kHz. The 6XX and HFM versions had different responses with the equalization active. The 6XX equalization (fig.1, blue and red traces) featured a narrow 3.8dB boost at 6kHz and a gently rising response above 10kHz. The HFM equalization (green, gray traces) had a broader 4.1dB boost centered on 2kHz, and while both equalizations had a rising response below 200Hz, that of the 6XX version was a little higher in amplitude over most of the bass region.

Fig.1 was taken with the ZEN CAN versions' volume controls at the maximum setting and the gain set to 18dB. The channel matching was within 0.02dB and the frequency response didn't change at lower settings of the control. However, because JMu had reported hearing differences in the level of the channels at lower settings of the volume control, I investigated the HFM version's channel matching across its volume control's range. With the volume control set to –10dB and –30dB, the right channel was 0.5dB higher than the left and the right channel was 1dB higher than the left at –40dB. The lowest volume control setting I could investigate was –45dB, which was with the volume control set to 7:00. At that setting, the right channel was 3.3dB higher in level than the left. I understand that iFi recommends not using the volume control lower than 11 o'clock, which is equivalent to an attenuation of 20dB and where the channel levels matched to within 0.03dB.

The iFi preamp's channel separation was >70dB in both directions below 1kHz but decreased in a linear manner as the frequency increased, reaching a disappointing 33dB at 20kHz. (Both versions behaved identically in this respect.) The level of the ZEN CAN's noisefloor varied with the gain setting. The blue and red traces in fig.2 show the spectrum of the HFM version's balanced output with the gain set to 18dB; the green and gray traces show the spectrum with it set to 0dB, but with the 1kHz input signal increased in level to give the same output. The random noisefloor drops by 10dB at the lower gain setting, but the supply-related spuriae at 60Hz, 180Hz, 300Hz, and 540Hz are reduced by approximately the same 18dB as the reduction in gain.

The wideband, unweighted signal/noise ratio, measured at the balanced output with the unbalanced input shorted to ground, the gain set to 0dB, and the volume control set to its maximum, was a very high 91.2dB ref. 2V output (average of both channels, which were very similar). Restricting the measurement bandwidth to the audioband increased the S/N to a superb 110.8dB, while switching an A-weighting filter into circuit further improved the ratio to 113.5dB. With the amplifier's gain set to its maximum, these ratios decreased by 18dB but were still excellent. These S/N ratios were taken with the HFM version; the 6XX version's ratios were approximately 6dB lower in both gain settings.

Figs.3 and 4 respectively plot the percentage of THD+noise in the ZEN CAN's balanced output against the output voltage into 300 ohms with the gain respectively set to 18dB and 0dB. We specify the clipping voltage as being when the THD+N reaches 1%, which is 15V in fig.3 and 7.6V in fig.4. The unbalanced output's clipping voltages were half those of the balanced output's, but even into 30 ohms (fig.5), the ZEN CAN's unbalanced output delivered 7.2V at 1% THD+N. These clipping voltages are much higher than will be necessary with headphones having normal sensitivities, which implies that the THD+N will be very low at typical output levels. I therefore measured how the ZEN CAN's distortion changed with frequency at 2V. The THD+N percentage was extremely low throughout the audioband into 300 ohms (fig.6, blue and red traces) and only rose slightly into 30 ohms (cyan and magenta traces), though there was a very slight increase in the top audio octave.

I looked at the spectrum of the distortion at 3V into 300 ohms (fig.7), because no distortion harmonics were visible above the noisefloor at lower levels. The only harmonic that can be seen is the third harmonic at –117dB (0.00014%), which is below the level of the supply-related spuriae. Tested for intermodulation distortion with an equal mix of 19 and 20kHz tones at the same peak voltage level, the second-order difference product at 1kHz lay below –120dB, and the higher-order products were just below –110dB (0.0003%) (fig.8).

Overall, the ZEN CAN offered superb measured performance.—John Atkinson