HRT Music Streamer Pro USB D/A converter Measurements
Sidebar 4: Measurements
I used Stereophile's loan sample of the top-of-the-line Audio Precision SYS2722 system to perform the measurements on the HRT Music Streamer Pro (see www.ap.com and the January 2008 "As We See It,"); for some tests, I also used my Audio Precision System One Dual Domain and the Miller Audio Research Jitter Analyzer. As source, I used my Apple MacBook computer running OS10.6 and Bias Peak Pro 6 to play WAV files of the various test signals I use.
The Music Streamer Pro uses the Texas Instruments TAS1020B USB receiver chip, 390which is hi-rez capable, mounted on a small daughterboard. Apple's USB Prober utility identified the HRT as being the "Music Streamer Pro" from "www.hirestech.com 2010 REV 1.4," and indicated that the HRT accepted up to 24-bit data with sample rates of 32, 44.1, 48, 88.2, and 96kHz. USB Prober also confirmed that the Music Streamer Pro operates in "isochronous asynchronous" mode with a polling interval of 1ms.
The Pro uses a Burr-Brown PCM1794 24-bit DAC, and its output stage appears to be based on two high-performance OPA4131 quadop-amp chips. The maximum level at 1kHz from the balanced outputs was to specification, at 4.5V RMS, and was exactly half that from the single phase used for the TiniQ/RCA adapter supplied with the review sample. The output impedance was a little lower than the specified 200 ohms for each phase of the balanced signal, at 193 ohms, but was constant across the audioband. With both the TiniQ/XLR and TiniQ/RCA adapters and the balanced cable supplied by HRT for the review, the Music Streamer Pro's output was in inverted polarity.
The Pro operated correctly with data having sample rates of 3296kHz, including the important 88.2kHz rate. Fig.1 shows the Music Streamer's frequency response at sample rates of 44.1kHz (cyan and magenta traces) and 96kHz (blue, red). The output is flat within the audioband and rolls off smoothly above 10kHz, reaching 0.8dB at 20kHz with 44.1kHz data and 1.5dB at 40kHz with 96kHz data. A slight ripple at the top of the audioband can be seen in the left channel only (cyan, blue traces). This is most likely subjectively innocuous, but suggests that the digital filter code is not quite the same in the two channels. Channel separation (not shown) was 100dB in both directions at all frequenciesimpressive performance for a USB bus-powered device.
Fig.1 HRT Music Streamer Pro, frequency response at 12dBFS into 100k ohms from balanced outputs with data sampled at: 44.1kHz, 96kHz (left channel blue, right red; 0.25dB/vertical div.).
I tested the Music Streamer's resolution in my usual fashion, feeding it 16- and 24-bit data representing a dithered 1kHz tone at 90dBFS while sweeping the center frequency of a 1/3-octave bandpass filter from 20kHz to 20Hz; the resulting spectra are shown in fig.2. The traces peak at exactly 90dBFS, suggesting very low linearity error, with no distortion-related spuriae visible for 16- or 24-bit data in the right channel (dotted trace). However, the left channel's spectrum with 24-bit data has bumps at 2, 3, 5, and 9kHz, which implies truncation at the LSB level. Fig.3, which shows high-resolution FFT-derived spectra with the same data, confirms this analysis. Again, the left channel is not as good as the right.
Fig.2 HRT Music Streamer Pro, 1/3-octave spectrum with noise and spuriae of dithered 1kHz tone at 90dBFS with 16-bit data (top) and 24-bit data (middle), and of dithered tone at 120dBFS with 24-bit data (right channel dashed).
Fig.3 HRT Music Streamer Pro, 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).
On the positive side, increasing the word length drops the noise floor in both channels in both these graphs by more than 20dB, implying resolution approaching 20 bits, which is excellent. The HRT readily reproduces a tone at 120dBFS (fig.2, bottom traces), but the left channel (solid trace) has a positive linearity error of 8dB, with distortion harmonics evident, and the right channel (dotted trace) has a 2dB negative linearity error and some second-harmonic distortion. This misbehavior was confirmed with FFT analysis (fig.4). While it is probable that this problem will have no audible consequences, and while the Pro works perfectly with 16-bit data, producing a perfect waveform with undithered 16-bit data at exactly 90.31dBFS (fig.5), there are some mathematical implementation problems with 24-bit data and signal levels below 90dBFS in the left channel.
Fig.4 HRT Music Streamer Pro, FFT-derived spectrum with noise and spuriae of dithered 1kHz tone at 120dBFS with 24-bit data (left blue, right red).
Fig.5 HRT Music Streamer Pro, waveform of undithered 1kHz sinewave at 90.31dBFS, 16-bit data (left channel blue, right red).
At high levels, the Music Streamer Pro offers a regular series of distortion harmonics (fig.6) that are unaffected by reducing the load to the demanding 600 ohms. Though the highest-level harmonicsthe second in the left channel and the fifth in both channelsare still low at 84dB (0.006%), this does suggest that bus power with the necessary DC/DC voltage converter is not quite sufficient to produce maximally linear behavior from the Pro's output stage, even though the converter is followed by monolithic ±5V voltage-regulator chips. Dropping the signal level by 10dB (fig.7) significantly reduces the levels of the higher-order harmonics; however, the nonlinear behavior at high levels does lead to disappointing performance in the high-frequency intermodulation test (fig.8). The 1kHz difference product is low, especially in the left channel (blue trace).
Fig.6 HRT Music Streamer Pro, spectrum of 1kHz sinewave, DC10kHz, at 0dBFS into 100k ohms (left channel blue, right red; linear frequency scale).
Fig.7 HRT Music Streamer Pro, spectrum of 1kHz sinewave, DC10kHz, at 10dBFS into 100k ohms (left channel blue, right red; linear frequency scale).
Fig.8 HRT Music Streamer Pro, HF intermodulation spectrum, DC24kHz, 19+20kHz at 0dBFS into 100k ohms (linear frequency scale).
HRT correctly points out that asynchronous USB operation should eliminate data-related jitter, but there are other mechanisms that might have an effect. There doesn't appear to be a crystal oscillator on either of the Pro's circuit boards, so the quality of its master clock circuit will be critical. The spectrum of the Music Streamer Pro's output while it processed 16-bit J-Test data (fig.9) indicated that the odd-order harmonics of the low-frequency, LSB-level squarewave were at the residual level and were therefore not being accentuated. There were some other sideband pairs visible, however, so I repeated the test using a 24-bit version of the J-Test signal. The results are shown in fig.10; what appear to be pairs of sidebands are visible at ±1, ±2, and ±3kHz, as well as at ±229.5Hz in the left channel (blue trace). However, other than the last pair of sidebands, these spectral spikes are not symmetrical around the spike that represents the high-level 11.025kHz tone, so they can't be jitter-related. I have no idea where these stem from, therefore. Commendably, the central spike is narrow, implying the absence of random low-frequency jitter.
Fig.9 HRT Music Streamer Pro, high-resolution jitter spectrum of analog output signal, 11.025kHz at 6dBFS, sampled at 44.1kHz with LSB toggled at 229Hz, 16-bit data via USB from MacBook (left channel blue, right red). Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz.
Fig.10 HRT Music Streamer Pro, high-resolution jitter spectrum of analog output signal, 11.025kHz at 6dBFS, sampled at 44.1kHz with LSB toggled at 229Hz, 24-bit data via USB from MacBook (left channel blue, right red). Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz.
Other than the idiosyncratic nature of its left channel with very low-level 24-bit data and the relatively poor high-level linearity of its output stage, HRT's Music Streamer Pro offers well-balanced measured performance at an affordable price.John Atkinson