Musical Fidelity V-DAC D/A processor Measurements
The V-DAC offers bare-bones construction. The power supply is a separate wall wart. A plain, black-painted box of extruded aluminum carries a single printed-circuit board, with the input jacks (TosLink, coaxial, USB) at one end and a single pair of RCA output jacks at the other. Other than three surface-mount LSI chips, the circuitry is based on traditional through-hole components, and local voltage regulation appears to be performed with the usual ±5V chips. The USB input feeds the ubiquitous Burr-Brown PCM2706 receiver chip, which is limited to 16-bit data and sample rates of up to 48kHz. The USB receiver operates in "adaptive" mode, where control of the data flow is subcontracted to the PC; it feeds the recovered audio data to a Burr-Brown SRC4392 sample-rate-converter chip, which also handles data up to 24-bit resolution and sample rates up to 96kHz from the TosLink and coaxial S/PDIF ports. Using this chip to upsample incoming data to 192kHz reduces the effect of datastream jitter.
The upsampled data are decoded to analog using yet another Burr-Brown chip, a DSD1792, which also does the necessary digital filtering. A high-speed quadop-amp chip, a Motorola MCC33079, does the current-to-voltage conversion. This is followed by the output stage, based on a JRC 5532 dualop-amp chip. While the 553x family of op-amps is now long in the tooth, Musical Fidelity uses them for the output stage of many of its products, due to their ability to drive low-impedance loads with very low distortion.
I examined the measured behavior of the Musical Fidelity V-DAC1 using the Audio Precision SYS2722 system (see www.ap.com and "As We See It" in the January 2008 issue, as well as, for some tests, my Audio Precision System One Dual Domain and the Miller Audio Research Jitter Analyzer. Test data were sent to the V-DAC via TosLink from the AP systems or from the USB 2.0 output of my Intel MacBook running OS10.4.11 and playing back WAV files using Bias Peak 6.2. Unlike iTunes, Peak takes control of the OSX CoreAudio engine to ensure that audio at the correct sample rate is sent out through the computer's USB and FireWire ports. To avoid problems of noise contamination, I ran the MacBook on battery power for the testing, and used a premium USB cable from Belkin.
The V-DAC's maximum output was 2.08V and it preserved absolute polarity; ie, was non-inverting. The output impedance was a low 42 ohms at high and mid frequencies, rising slightly but inconsequentially to 78 ohms in the low bass. The S/PDIF inputs successfully locked to data with sample rates up to 96kHz. The USB input identified itself to the host computer as "USB Audio DAC" and was limited to sample rates at or below 48kHz. The audioband frequency response was the same at all sample rates, with a gentle droop evident above 10kHz (fig.1). Increasing the sample rate increased the frequency at which the ultrasonic brickwall filter cut in: with 44.1kHz data (fig.1, cyan and magenta traces), the response was 0.25dB at 19kHz; with 96kHz data (blue and red traces), the rolloff continued to reach 1.1dB at 40kHz. Channel separation (not shown) was a superb 110dB in both directions in the midrange, decreasing to a still-excellent 100dB at 20kHz (due to capacitive coupling between the channels) and to 100dB in the bass (presumably due to increasing power-supply impedance).
Fig.1 Musical Fidelity V-DAC, frequency response at 12dBFS into 100k ohms with 44.1kHz data (left channel cyan, right magenta) and data at 96kHz (left channel blue, right red; 0.25dB/vertical div.).
For consistency with my two decades' worth of previously published tests of digital components, I first examine resolution by sweeping a 1/3-wide bandpass filter from 20kHz to 20Hz while the device under test decodes dithered data representing a 1kHz tone at 90dBFS. The top pair of traces in fig.2 show the result for the V-DAC decoding 16-bit datathe trace peaks at exactly 90dBFS, suggesting minimal linearity error, while the noise floor is free from harmonic- or power-supplyrelated spuriae. In fact, all the traces show is the spectrum of the dither used to encode the data, the V-DAC's own noise being much lower in level. Increasing the word length to 24 bits (using the S/PDIF input) gives the middle pair of traces in fig.2. The noise floor has dropped by 20dB, suggesting that the V-DAC has better than 19-bit resolution, which is competitive with D/A processors costing many times its price.
Fig.2 Musical Fidelity V-DAC, 1/3-octave spectrum with noise and spuriae of dithered 1kHz tone at 90dBFS with 16-bit data (top), 24-bit data (middle at 2kHz), and of dithered 1kHz tone at 120dBFS with 24-bit data (bottom at 1kHz). (Right channel dashed.)
Dropping the signal level to 120dBFS gives the bottom pair of traces in fig.2; the tone is easily resolved, though a couple of dB of negative error are evident. With all the 24-bit traces, a very small amount of power-supply hum at 120Hz is unmaskedat 135dB, this won't bother anyoneand a small spectral bump can be seen between 5 and 7kHz. Repeating the analysis using an FFT technique (fig.3), the bump is resolved to two spectral lines just below 6kHz, probably the result of a very slight DC offset being introduced into the data during its mathematical manipulation ahead of the D/A stage. Again, however, the V-DAC's very low noise floor is evident in this graph.
Fig.3 Musical Fidelity V-DAC, FFT-derived spectrum with noise and spuriae of dithered 1kHz tone at 90dBFS with 16-bit CD data (left channel cyan, right magenta) and 24-bit data (left channel blue, right red).
The plot of the Musical Fidelity's linearity error against absolute level with 16-bit data revealed only the effect of the recorded dither noise (fig.4). Fig.5 shows the waveform of an undithered 16-bit/1kHz tone at exactly 90.31dBFS: the three discrete DC voltage levels described by the data are clearly resolved, with excellent waveform symmetry. Increasing the data's depth to 24 bits gives rise to a well-defined sinewave (fig.6).
Fig.4 Musical Fidelity V-DAC, linearity error (16-bit data), dBr vs dBFS.
Fig.5 Musical Fidelity V-DAC, waveform of undithered 1kHz sinewave at 90.31dBFS, CD data (left channel blue, right red).
Fig.6 Musical Fidelity V-DAC, waveform of undithered 1kHz sinewave at 90.31dBFS, 24-bit data (left channel blue, right red).
Only when it came to harmonic distortion did the V-DAC stumble, and then in only a very minor way. Fig.7 shows the spectrum of the DAC's output while it decoded a full-scale 1kHz tone into 600 ohms. (The result into the more benign 100k ohms was basically identical, so I haven't shown it.) A regular series of harmonic spuriae can be seen, as well as the idle tones just below 6kHz, though it is fair to point out that all these lie at or below 96dB in the left channel, 100dB in the right. The left channel (blue trace) has more third and fifth harmonic content than the right (red). Intermodulation distortion was also very low (fig.8), and no aliasing products were visible. Again, the left channel was not quite as linear as the right, and the performance into the punishing 600 ohm load was no worse than into 100k ohms.
Fig.7 Musical Fidelity V-DAC, spectrum of 1kHz sinewave at 0dBFS into 600 ohms, 24-bit data (left channel blue, right red; linear frequency scale).
Fig.8 Musical Fidelity V-DAC, 19+20kHz at 0dBFS peak into 600 ohms, 24-bit data (left channel blue, right red; linear frequency scale).
As with other products using the SRC4392 chip as an S/PDIF receiverthe Music Hall dac25.2 comes to mindthe V-DAC does not reject incoming jitter as well as I would wish, which is presumably why ST heard significant differences between the transports he tried with the V-DAC. The measured jitter level via both the TosLink and coaxial inputs was never high, but it varied considerably according to the source I used. Fed via 15' of TosLink cable from the RME soundcard mounted in one of my test-lab PCs, the measured jitter level was 444 picoseconds peakpeak, with most of the energy in the data-related sidebands at ±229.5Hz and ±689.5Hz. Changing the source to the AP SYS2722's TosLink output and using the same 15' of optical cable, the jitter halved in level, with now just the ±229.5Hz sidebands visible, and then only in the left channel (fig.9).
Fig.9 Musical Fidelity V-DAC, 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 TosLink from AP SYS2722. Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz (left channel blue, right red).
Feeding the V-DAC the same test signal via USB, these sidebands disappeared (fig.10) and the measured jitter level dropped below the resolution limit of the Miller Analyzer. Given that the V-DAC uses the PCM2706 chip in the jitter-prone adaptive mode, this result was surprisinguntil I remembered that the SRC4392 sample-rateconverts the incoming data, which will minimize the jitter. But then I don't understand why its doing so is less effective with S/PDIF data.
Fig.10 Musical Fidelity V-DAC, 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. Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz (left channel blue, right red).
Yes, the D/A chips now coming from the foundries run by companies like Burr-Brown/Texas Instruments are capable of superb linearity and resolution, but the designer of a product like the Musical Fidelity V-DAC still has to be able not to compromise that performance with analog design and circuit layout. The V-DAC's generally superb measured performance indicates that compromise was avoided; it would not disgrace a much-higher-priced product, let alone one that costs just $299half what our family spends on groceries each month.John Atkinson