Digital Processor Reviews

Sidebar 2: Measurements

Both Kal Rubinson and Michael Lavorgna report in this issue on the positive effects UpTone Audio's USB Regen7 had on the sound quality of D/A processors fed USB data. Kal loaned me his review sample for a few days so I could see if it made any measurable differences to a D/A processor's analog output signal.

Unlike AudioQuest's passive JitterBug, which I reviewed in the September issue, the Regen is an active device that regenerates the USB datastream. The JitterBug is intended to reduce the level of radio-frequency (RF) noise on the balanced data connection, the 5V power bus, and the ground connection. In my measurements of the JitterBug for my review, I used four different test signals with three different D/A processors and one, two, or no JitterBugs. Only in two of the large number of test conditions did I find any measurable improvement in the reconstructed analog signal. Even then, the measured improvement was so small that it was hard to see that it could have had anything to do with the improvement in sound quality I heard the JitterBug make with the Mac mini I use as a music server.

One thing that emerged after the review was published was that Gordon Rankin, one of the JitterBug's designers, suggested using a very long USB cable to bring the measurable difference out from the DAC's noise floor. The long cable will marginalize the USB transmission to the point that the effect of the JitterBug will be more easily measurable. So to measure the effect of the Regen, I took a generic 3m USB cable and lengthened it with a 1.5m USB cable extender. This brought the total length of the cable close to the USB's maximum cable length of around 12'.

Using this long cable and my MacBook Pro running on battery power, I tested the Regen with a PS Audio PerfectWave DirectStream DAC and a Meridian Prime D/A headphone amplifier, using my Audio Precision SYS2722 analyzer and the same test signals I'd used for the JitterBug review. Sad to report, I found that the Regen made absolutely no difference in the D/A processors' analog output signals. With one exception: the levels of the power-supply–related spuriae in the Prime's output when powered by its wall wart were slightly higher in level with the Regen than without it.

However, I did find one idiosyncrasy that will affect A/B comparisons of the Regen's effect. I used Pure Music 2.0 to play my source files. When I disconnected the D/A processor from the MacBook Pro's USB port to insert the Regen then reconnected it, while Pure Music still sent the file's data to the USB port for playback, it no longer automatically changed the Mac's CoreAudio parameters to match the file. If I played a 16-bit/44.1kHz file without the Regen, then disconnected the DAC, inserted the Regen, reconnected the DAC, and played a 24/96 file, that file was truncated to 16 bits and downsampled to 44.1kHz. This is something that will be readily audible.

The correct procedure to perform A/B comparisons with the Regen using Pure Music was to deselect the DAC in use as the output device, relaunch Pure Music, disconnect the DAC, insert the Regen, reconnect the DAC, reselect the DAC with Pure Music, then relaunch Pure Music. Then, and only then, could a legitimate A/B test be performed.

Kal and Michael were careful to observe the correct protocol when performing their comparisons with the USB Regen. But I do wonder how many of the A/B tests published on the Internet were not done correctly.—John Atkinson