Proceed CD player Measurements
Sidebar 2: Measurements
The Proceed showed no surprises on the bench. Frequency response was extremely flat, with a very slight rolloff (0.1dB right, 0.2dB left) at 20kHz, though there was a minor deemphasis error of +0.2dB at 16kHz. Spectral analysis of a 90dB, dithered 1kHz signal, shown in fig.1, revealed a low level of HF spuriae, though some linearity error can be seen by the 90dB recorded signal rising above the 90dB horizontal division. Somewhat surprisingly, this implied positive level error was not apparent on a plot of departure from linearity (fig.2). Checking with a 1/3-octave spectrum analyzer suggested that it is fig.2 that is correct. not fig.1, with a 2dB error apparent between 70 and 90dB.
The "fade to noise with dither" track on the CBS test disc resulted in a fairly straight line (fig.3), indicating good differential linearity. It also sounded exceptionally pure as it decayed, with no harmonics audible, just rather hashy-sounding noise.
In addition, linearity was nearly perfectly matched between channels. Looking at the low-level waveforms with a storage 'scope indicated that audio-band noise was a little higher than with some other machines, though the absolute noise floor was still below 90dB. Fig.4 shows the 80.77dB undithered 1kHz tone from the CBS test CD: what should be a stepped waveform with seven distinct levels is obscured by the noise present. Similarly with the undithered 90.31dB tone shown in fig.5. This waveform consists of just three levels, +1, 1, and digital zero, but this can be seen to be overlaid with around half of an LSB's worth of noise. The same is true of the monotonicity trace (fig.6), which should feature 10 distinct "steps" in the waveform, but is somewhat obscured by noise. The straight slopes, however, suggest good differential linearity.
The high-level performance of the Proceed's digital filter is revealed by fig.7, which shows how the player reproduces a 0dB, 1kHz squarewave. The flat tops to the waveform suggest that the digital filter actually clips with this signal, while a slight asymmetry is also apparent, the negative "tops" appearing to have a slight HF roll-off compared with the positive "tops." This is a severe test signal, however, and one that is hardly likely to be met in real-life music. Nevertheless, the Proceed might be expected to sound a little hard with extremely hot CDs, such as the Bainbridge Stephen Kates Rachmaninoff recording.
The Proceed's error-correction capability was tested with the second disc of the Pierre Verany set (PV.788032). It proved to have some problems with the larger dropouts, anything over a 1.5mm gap in the data leading to repeated blips in the sound. Finally, its maximum output is 0.5dB higher than the standard 2V, which will be audible on an A/B comparison. Take care when auditioning to match levels carefully.Robert Harley