Krell DT-10 CD transport Measurements
Sidebar 3: Measurements
This is the first CD-transport review to include a measurements section: We can now measure how much jitter appears in a transport's digital output using the UltraAnalog jitter analyzer described in the October and November 1993 issues of Stereophile. For comparison measurements on other products, see the November '93 feature story.
Fig.1 shows the DT-10's jitter (from the AES/EBU output) with three test signals: digital silence (solid trace), a 90dB, 1kHz sinewave (heavy dashed trace), and a full-scale, 1kHz sinewave (lightly dotted trace). The RMS levels, measured over a 30kHz bandwidth, were 34picoseconds (silence), 127ps (90dB sinewave), and 39ps (full-scale 1kHz sinewave).
Fig.1 Krell DT-10, jitter in AES/EBU data signal, 20Hz50kHz, when transmitting digital silence (bottom solid trace), a 1kHz sinewave at 90dB (top, dashed trace), and a 1kHz sinewave at 0dBFS (middle, light dotted trace) (vertical scale, 1ps2ns, 100µV = 1ps).
Using the same high- and low-level music signals described in last November's jitter article, I plotted the DT-10's jitter, shown in fig.2. The RMS levels were 67ps (low-level music, solid trace) and 43ps (high-level music, dotted trace). Incidentally, the DT-10's jitter levels were about twice as high from the coaxial output as from the AES/EBU output. (The test results presented here were all measured at the AES/EBU output.)
Fig.2 Krell DT-10, jitter in AES/EBU data signal, 20Hz50kHz, when transmitting Sheffield Lab Firebird (solid) and Steve Morse (dashed) (vertical scale, 1ps2ns, 100µV = 1ps).
Although these RMS levels are moderate, note that the traces are smooth rather than spiky. (The peak between 7 and 8kHz in fig.1's digital silence jitter trace is due to the subcode signal. The peaks at 1kHz and its harmonics in fig.1 are signal-correlated, due to the format of the AES/EBU data stream.) This indicates that the DT-10's jitter is the more sonically benign random jitter instead of the more musically degrading periodic jitter. Note also the complete absence of jitter at the power-line frequency of 60Hz.
I must again caution readers not to jump to conclusions about a transport's sound quality based on the jitter measurements. The RMS levels and curves shown here represent the transport's jitter performance into the UltraAnalog jitter analyzer's input circuit, and may not reflect the transport's performance into different D/A converters. Moreover, the technique is so newlast November's article was the first published presentation of transport jitter measurements anywherethat we don't yet understand the correlation between sound quality and transport jitter.
Finally, the DT-10's jitter performance is theoretically a moot issue when the DT-10 is used with Krell processors having the Time Sync input. If the Time Sync clock output is jitter-free (a much easier task than achieving low jitter in the interface), the jitter seen in the measurements won't appear at the DAC's word clock and thus cannot corrupt the sound. Nevertheless, the DT-10's sonic character remained identifiable even with the Time Sync engaged.
I also tested the DT-10's tracking ability by playing the data dropouts deliberately encoded on the Pierre Verany two-CD test set. Each track in the series has a longer dropout: the higher the track number the transport will play without skipping, the better its tracking ability. Most players and transports begin skipping at about track 34, with good players reaching track 38. To my surprise, the DT-10 played through every track, without skipping, to track 50the highest in the series. When searching the higher track numbers, the DT-10 took a few seconds to settle down, but then played through the track without a hitch. This is the best tracking performance I've seen, by a wide margin.Robert Harley