The Jitter Game Page 5

The last processor measured was the Bitwise Musik System Zero (reviewed in September 1992). At full scale the Zero had a smooth spectrum, with very few spikes and an overall jitter level of between 2.0 and 2.8ns. It maintained this performance until -50dBFS (worst case), where the number and amplitude of the periodic jitter components increased dramatically. The Zero's jitter spectrum is shown when driven by a full-scale signal (fig.27) and at -50dB (fig.28).

Fig.27 Jitter Spectrum, Bitwise System Zero processing 1kHz sinewave at 0dBFS (linear frequency scale, 0dB = 226.8ns). Medium level of mainly random jitter (but note components at approximately 19khz and 20kHz).

Fig.28 Jitter Spectrum, Bitwise System Zero processing 1kHz sinewave at -50dBFS (linear frequency scale, 0dB = 226.8ns). Reducing the data level brings up the overall level of jitter.

Once I got going with the LIMD, I found all sorts of things to measure. First, I wanted to see if there was a measurable difference in the clock jitter at the DAC when a processor was driven via a coaxial cable or Toslink optical interconnect. Critical listeners agree that the Toslink sounds substantially inferior to coaxial—an assertion that brings scorn from the "bits is bits" mentality. Fig.29 is the Musik System Zero's jitter spectrum when driven by data representing a full-scale, 1kHz squarewave (a test signal that exacerbates the number and level of discrete-frequency jitter components) via a coaxial interconnect. Fig.30 shows the jitter spectrum under the same measurement conditions, but when the processor was driven through the lower-bandwidth (6MHz) Toslink optical interconnect. Note the significant reduction in both the number and amplitude of periodic jitter components in fig.29. This is the first "objective" evidence I've seen that digital audio interfaces sound different (footnote 8).

Fig.29 Jitter Spectrum, Bitwise System Zero processing 1kHz squarewave at 0dBFS via coaxial datalink (linear frequency scale, 0dB = 226.8ns). Note presence of data-related components at squarewave odd-harmonic frequencies.

Fig.30 Jitter Spectrum, Bitwise System Zero processing 1kHz squarewave at 0dBFS via Toslink optical datalink (linear frequency scale, 0dB = 226.8ns). Note significant increase in level of data-related, odd-harmonic jitter components with the lower-bandwidth data connection.

Finally, I measured the effect of the Audio Alchemy Digital Transmission Interface (DTI) on clock jitter. The DTI is an inexpensive $349 reclocking device inserted between a CD transport and a digital processor that reportedly reduces jitter in the data stream. Using the Sumo Theorem driven by a full-scale, 1kHz squarewave, I compared the jitter spectrum without (fig.31) and with (fig.32) the DTI. As claimed, the DTI does indeed reduce jitter in the data stream, evinced by the reduction in periodic jitter component amplitude of an astounding 25dB. In addition, the number of jitter components rising higher than 20dB above the overall level was reduced from 11 components without the DTI to just three components with the DTI. In short, the Audio Alchemy Digital Transmission Interface produces a measurable reduction in a digital processor's clock jitter where it counts—at the DAC. A full review of the DTI will appear shortly.

Fig.31 Jitter Spectrum, Sumo Theorem processing 1kHz sinewave at 0dBFS (linear frequency scale, 0dB = 226.8ns). Note presence of data-related components at squarewave odd-harmonic frequencies.

Fig.32 Jitter Spectrum, Sumo Theorem processing 1kHz squarewave at 0dBFS via Audio Alchemy DTI (linear frequency scale, 0dB = 226.8ns). Note significant decrease in level of data-related, odd-harmonic jitter components.

The Meitner LIM Detector opens many additional measurement possibilities that were beyond the scope of this article. For example, it is now possible to explore how sensitive a particular digital processor is to transport quality, measure the jitter between different combinations of transports and processors with different interconnects, assess jitter-reduction devices (the Audio Alchemy DTI, for example), and measure the effects—if any—of such CD tweaks as rings, green paint, the Laser Illusions Spatial Filter, and other products.

Summing up
Minute timing variations in a digital audio system produce an analog-like variability in the final analog signal's fidelity. The belief that if the ones and zeros are the same, the sound must be the same, is thus exposed as, at best, naïive.

Moreover, there seems to be a broad correlation between a digital processor's measured jitter performance and certain aspects of its musical presentation. This measurement technique is a powerful tool for exploring the reasons why things like interconnects between transport and processor sound different. Because we can measure differences in a well-established source of sonic degradation—clock jitter at the DAC—with different digital interface types (coax vs Toslink, for example), we are well on the way toward scientific justification of the subjective differences we hear.



Footnote 8: I recently encountered a past president of the Audio Engineering Society to whom I send Stereophile. When I asked him if he'd been reading the magazine, a bemused look came over his face. The source of his amusement, he told me, was reports in Stereophile about differences in sound between digital interfaces and interconnects, an idea he considered the height of subjectivist absurdity. When asked if he had ever listened for himself to different digital interfaces, he admitted he hadn't, and had no plans to.—Robert Harley

X