Yamaha @PET RP-U100 personal receiver Measurements part 3
Fig.7 Yamaha RP-U100, high-resolution jitter spectrum of analog output signal (11.025kHz at -6dBFS with LSB toggled at 229Hz), driven by PS Audio Lambda transport via 75 ohm S/PDIF cable. Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz. Grayed-out trace is Musical Fidelity X-24K.
As well as a noise floor 12dB higher than that of the identically priced Musical Fidelity X-24K processor (also shown in fig.7), and equivalent to a dynamic range of around 14 bits, there are some high-level sidebands visible, these related both to the 229Hz data frequency (red numeric markers) and to the power-supply frequency and its harmonics (purple markers). The measured amount of jitter was a very high 2.95 nanoseconds, more than 10 times higher than the Musical Fidelity taken under the same conditions. Of course, as a dedicated high-end D/A unit, the X-24K lacks all of the Yamaha's features!
Note that this measurement was taken with the RP-U100 driven by its S/PDIF data input from a conventional CD transport. When the Yamaha is fed music data via the USB connection, its jitter performance will be different, perhaps even better. To test the jitter performance in this mode, I recorded the analytical test signal on the computer's hard drive by feeding it from the CD transport to one of the Yamaha's optical data inputs, from where it was fed by the USB bus to the host computer. Playing back the resultant WAV file by connecting the computer soundcard's S/PDIF data output to the Yamaha's S/PDIF data input gave jitter performance comparable to what I'd obtained before, showing that retrieving the audio data from the hard drive is not any different from doing so from a CD.
Unfortunately, when I selected the Yamaha's USB audio function as the computer's preferred playback device rather than the soundcard, the resultant analog signal was very slightly unstable, in terms of both absolute frequency accuracy and level. As the Miller Analyzer needs to see a rock-steady signal in order to perform the 64 averages necessary for it to obtain the required resolution of the jitter sidebands in the presence of analog noise, it kept jumping out of its acquisition mode. So I used instead a PrismSound DScope, with a dCS 904 24-bit A/D converter as its acquisition front-end. The FFT length was set to 32k to obtain sufficient resolution, and meaningful differences can be seen in fig.8 between the jitter performance using an optical S/PDIF output from the computer's soundcard (red trace) and via the USB port (blue trace).
Fig.8 Yamaha RP-U100, high-resolution jitter spectrum of analog output signal (11.025kHz at -6dBFS with LSB toggled at 229Hz), driven by WAV file on host PC's hard drive via S/PDIF datalink (red) and USB datalink (blue). Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz.
The noise floor was very slightly higher in the USB mode, but, far from the jitter dropping in level, it rose significantly. The power-supply-related sidebands at ±120Hz, for example, measured -92.5dB when the Yamaha was driven by the S/PDIF connection, but 10 times higher at -72.5dB via the USB link, while the data-related jitter at ±229Hz more than doubled, rising from -84dB to -77.2dB. In addition, a pair of high-level sidebands at ±1kHz made their appearance, and the absolute frequency error increased by 182ppm.
While it seems a good idea to drive the Yamaha with digital audio data via the same USB connection used to control its functions, either the Yamaha's implementation or the USB concept seems flawed as far as highest-quality sound is concerned (footnote 1).
Looking at the Yamaha's analog performance, its volume-control steps are sensibly arranged to be a uniform 1dB (±0.1dB) down to a displayed setting of "6," whereupon the steps increase to 3dB. The receiver's output impedance was a moderately low 0.12 ohms in the left channel, 0.19 ohms in the right, both figures maintained across the audioband. This gave just ±0.1dB of response variation into our standard simulated loudspeaker load.
Driving the RP-U100's analog inputs gave a frequency response identical to that in fig.1, and so is not shown. In fact, the analog inputs are brickwall low-pass-filtered above 20kHz because the Yamaha digitizes them in order to apply the DSP. This can also be seen in fig.9, the waveform of a 1kHz squarewave. The "ringing" apparent on the tops and bottoms of the waveform is actually due to the absence of all odd-order harmonics above the 19th at 19kHz. The low-pass filtering results in squarewaves, with frequencies above 6.6kHz being reproduced as sinewaves (not shown).
Fig.9 Yamaha RP-U100, small-signal 1kHz squarewave into 8 ohms, analog input.
Footnote 1: Shannon Dickson has hypothesized that it is actually the USB driver in Windows 98 that is to blame here.