Krell Cipher SACD/CD player Measurements

Sidebar 3: Measurements

To measure the Krell Cipher, I used Stereophile's loan sample of the top-of-the-line Audio Precision SYS2722 system (see and the January 2008 "As We See It"); for some tests, I also used my vintage Audio Precision System One Dual Domain. It wasn't possible to measure the player's performance from its current-mode CAST output, but I did do so from its balanced and unbalanced outputs.

To test the Cipher's performance as an SACD player, I used the "provisional" Sony Test SACD. Although Krell doesn't mention DVD playback on their website in their lists of features and specifications for the Cipher, I found that it would play DVD-A discs. So as well as 16-bit test signals burned to a CD-R, I used 24-bit files burned as a DVD-A on a DVD-RW. While the Cipher would play commercial SACDs and DVD-As without problem, it would sometimes not recognize the DVD-RW, presumably because of that medium's low reflectivity compared with pressed discs.

Navigating the Pierre Verany Test CD, the Cipher demonstrated excellent error correction, coping with gaps in the data spiral without glitches in the analog output until those gaps reached 2mm in length. The player muted its output when the gaps were 2.4mm long. The maximum output level at 1kHz with CDs and DVDs was 3.835V from the balanced jacks and 1.918V from the unbalanced RCAs, sourced from impedances of 350 and 175 ohms, respectively. With SACDs, the maximum output level depended on the filter selected: Filter 1 gave the lowest levels, at 1.8V/900mV, balanced/unbalanced outputs; Filter 3 gave the highest levels, at 3.237V/1.62V. Both the balanced and unbalanced outputs preserved absolute polarity (ie, were non-inverting), the XLRs being wired with pin 2 hot.

With CD and DVD playback, the Cipher's frequency response depended on which of the two filter options was in use. At all sample rates, Filter 2 gave an earlier high-frequency rolloff than Filter 1 (fig.1). As with earlier Krell SACD/CD players, both the ultrasonic response rolloff and the absolute level varied considerably with which of the four filters was being used (fig.2; see also fig.1 here). All filters had responses that were flat within the audioband, but Filter 3 was 5.1dB higher in level than Filter 1, Filter 2 was 0.55dB higher, and Filter 4 was 3.3dB higher. These level differences are sufficiently large to invalidate any listening tests comparing the filters unless compensated for with the preamplifier's volume control. Channel separation (not shown) was superb at >125dB in both directions below 5kHz, and still 110dB at 20kHz.

Fig.1 Krell Cipher, PCM frequency response at –12dBFS into 100k ohms with data sampled at 44.1 and 96kHz with: Filter 1 (left channel blue, right red), Filter 2 (left green, right gray) (0.5dB/vertical div.).

Fig.2 Krell Cipher, SACD frequency response at –3dBFS into 100k ohms with: Filter 1 (left channel blue, right red), Filter 2 (left cyan, right magenta), Filter 3 (left green, right gray), Filter 4 (left blue, right red) (2dB/vertical div.).

Testing the Cipher's resolution by sweeping a 1/3-octave bandpass filter from 20kHz down to 20Hz while it played a dithered 1kHz tone at –90dBFS from CD (fig.3, top pair of traces below 5kHz), all that can be seen is the dither noise used to encode the signal. The bottom pair of traces in fig.3 were taken with 24-bit DVD data representing this signal; the noise floor drops by around 16dB at high frequencies, which suggests that the Krell has almost 19-bit resolution in this region, less at low frequencies. The middle pair of traces in fig.3 were taken with SACD data. As with the Krell Evolution, which Fred Kaplan reviewed in September 2008, the Cipher's conventional outputs are a little noisier than the best players I've tested with SACD, resulting in only a modest increase in resolution compared with CD. However, I performed this test with the Cipher set to Filter 1, which, on Krell's recommendation, was how FK had listened to SACDs; the low output level of this Filter slightly compromises absolute resolution.

Fig.3 Krell Cipher, 1/3-octave spectrum with noise and spuriae of dithered 1kHz tone at –90dBFS, with: 16-bit data (top), 24-bit data (middle), DSD data (bottom) (right channel dashed).

Fig.4 repeats the spectral analysis for 16- and 24-bit data with an FFT technique; again, it reveals that the increase in bit depth drops the noise floor by around 16dB above 1kHz, though a few very-low-level idle tones are unmasked by the reduction in noise. Figs.3 and 4 also show that some power-supply spuriae are present, though these all lie at or below –120dB and will therefore be inconsequential. Commendably, with DSD data there was no change in the noise floor as the signal level changed (fig.5).

Fig.4 Krell Cipher, FFT-derived spectrum with noise and spuriae of dithered 1kHz tone at –90dBFS with: 16-bit data (left channel cyan, right magenta), 24-bit data (left blue, right red) (linear frequency scale).

Fig.5 Krell Cipher, spectrum of 1kHz sinewave with DSD data, DC–1kHz, at 0dBFS into: 100k ohms (left channel blue, right red), –40dBFS (left cyan, right magenta), –60dBFS (left green, right gray) (linear frequency scale).

Linearity error with 16-bit data (fig.6) was less than ±1dB above –110dBFS, and was dominated by the recorded dither noise. With an undithered CD signal at exactly –90.31dBFS, the Cipher's excellent linearity and low noise readily allowed the waveform to be reproduced with good symmetry (fig.7). This graph was taken with Filter 1; Filter 2 gave an identical result. DSD data at the same level gave a well-formed sinewave (fig.8).

Fig.6 Krell Cipher, linearity error, 16-bit data (2dB/vertical div.).

Fig.7 Krell Cipher, Filter 1, waveform of undithered 1kHz sinewave at –90.31dBFS, 16-bit data (left channel blue, right red).

Fig.8 Krell Cipher, Filter 1, waveform of dithered 1kHz sinewave at –90dBFS, DSD data (left channel blue, right red).

As expected from a Krell player, the Cipher offered very low amounts of harmonic distortion, with the second and third harmonics highest in level (fig.9). In fact, the distortion is around 10dB lower than with Krell's older Evolution 505 player. This graph was taken into the benign 100k ohm load; dropping the load to 600 ohms actually reduced the level of the third harmonic, leaving the second harmonic unchanged at –110dBFS (0.0003%, fig.10). Tested with an equal, full-scale mix of 19 and 20kHz tones, the Cipher offered very low levels of intermodulation distortion (fig.11), though the fact that I had only a 16-bit version of this test signal on disc means that the noise floor in this graph looks rather hashy.

Fig.9 Krell Cipher, spectrum of 1kHz sinewave, DC–10kHz, at 0dBFS into 100k ohms (left channel blue, right red) (linear frequency scale).

Fig.10 Krell Cipher, spectrum of 1kHz sinewave, DC–10kHz, at 0dBFS into 600 ohms (left channel blue, right red) (linear frequency scale).

Fig.11 Krell Cipher, HF intermodulation spectrum, DC–30kHz, 19+20kHz at 0dBFS into 100k ohms, 16-bit CD data (left channel blue, right red) (linear frequency scale).

The Evolution 505 offered a rather idiosyncratic performance when tested for jitter rejection with CD playback. By contrast, the Cipher's jitter rejection was superb, with no accentuation in its analog output of the odd harmonics of the low-frequency squarewave component of the J-Test signal (fig.12).

Fig.12 Krell Cipher, high-resolution jitter spectrum of analog output signal, 11.025kHz at –6dBFS, sampled at 44.1kHz with LSB toggled at 229Hz: 16-bit data from CD (left channel blue, right red). Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz.

The Krell Cipher's measured performance can be summed up in one word: superb.—John Atkinson

Krell Industries
45 Connair Road
Orange, CT 06477-3650
(203) 298-4000
Share | |

Enter your username.
Enter the password that accompanies your username.