AH! Njoe Tjoeb 4000 CD player Measurements

Sidebar 3: Measurements

Despite being triple-boxed, the first sample of the AH! Njoe Tjoeb 4000 was a victim of shipping damage. It arrived at my place with only one channel working, and that for only a short while before the player expired. Upscale Audio's Kevin Deal sent me a replacement sample, which worked fine. It was identical with respect to the extras, other than not having the premium AC cable fitted.

As the 6922 tubes, upsampling board, and transport clamp all have to be installed before the player is fired up, this gave me the opportunity to look inside. Although the Njoe Tjoeb 4000 is based on an OEM player badged for Marantz, the tubed audio circuit and its associated board and power supply are all new, and the construction quality appears excellent. I was also impressed by the excellent instructions, complete with step-by-step photographs, that are supplied with the player and duplicated on the distributor's website. I wish more manufacturers were as helpful.

Absolute polarity was correct, and as set for the review, the 4000's maximum output level at 1kHz was almost 3dB higher than the CD standard, at 2.75V into a high 100k ohm load. The player couldn't drive a signal at 0dBFS into impedances lower than 2k ohms without the top half of the waveform squaring off, so even though the Njoe Tjoeb's output impedance was a low 170 ohms across most of the band, a preamplifier with an input impedance higher than 10k ohms is advised. Error correction was excellent, the player coping with gaps in the data spiral of up to 1.25mm in length without audible glitches.

The AH! player's frequency response was, surprisingly, less than flat (fig.1, top pair of traces), with the midbass-to-midrange region shelved up compared with the mid-treble. The boost is just over half a dB, but as it covers five octaves, it will be very audible. "I wouldn't characterize the 4000's sound as bright," wrote Chip Stern; from its measured response, neither would I. The tonal balance was even more skewed to the midrange and bass with a pre-emphasized signal (fig.1, lower traces), but such discs are relatively rare. Channel separation (not shown) was good, at better than 90dB in both directions at 1kHz, but less good at low and high frequencies, due to capacitive coupling in the latter case, and perhaps to increasing power-supply impedance in the former.

Fig.1 AH! Njoe Tjoeb 4000, frequency response at -12dBFS into 100k ohms without (top) and with de-emphasis (right channel dashed, 0.5dB/vertical div.).

However, the 4000's noise floor also slopes up at low frequencies, which can be seen in fig.2, a 1/3-octave spectral analysis of the Njoe Tjoeb's output while it played a dithered 1kHz tone at -90dBFS. Perhaps more significant, a number of peaks can be seen at power-supply-related frequencies, both at 60Hz and at its odd-numbered harmonics (which are probably generated by magnetic coupling from the AC transformers), and at 120Hz and its harmonics (which are due to electrical coupling from the full-wave-rectified supply). Even the highest in level, at 120Hz, only just pokes its head above the -90dB level, so they shouldn't be audible in typical systems. Still, I'd rather they weren't there at all.

Fig.2 AH! Njoe Tjoeb 4000, 1/3-octave spectrum of dithered 1kHz tone at -90dBFS, with noise and spuriae, 16-bit CD data (right channel dashed).

There is a hump around 2kHz in the noise floor visible in fig.2, which can be seen more clearly in a spectral analysis of the player's output while it decoded "digital black" (fig.3). The power-supply components can also be seen in this graph, as can a small peak just below 200kHz. It isn't possible with a 1/3-octave spectrum to distinguish between pure tones and true noise, so I repeated the analysis using an FFT technique (fig.4). This graph shows the phenomenon consists of a hump in the noise rather than idle tones, meaning that it is probably benign. However, I haven't seen something like this before.

Fig.3 AH! Njoe Tjoeb 4000, 1/3-octave spectrum of digital black, with noise and spuriae, 16-bit CD data (right channel dashed).

Fig.4 AH! Njoe Tjoeb 4000, FFT-derived spectrum of digital black, DC-4kHz, with noise and spuriae, 16-bit CD data (linear frequency scale).

The relatively high level of analog noise in the AH! Njoe Tjoeb 4000's output means that its linearity plot (fig.5) is marred by what appears to be an increasing amount of positive error below -100dBFS. But above that level the actual amplitude error is very low, suggesting good DAC linearity. However, the low-frequency noise obscures the waveform of an undithered 1kHz tone at -90.31dBFS (fig.6).

Fig.5 AH! Njoe Tjoeb 4000, left-channel departure from linearity, 16-bit CD data (2dB/vertical div.).

Fig.6 AH! Njoe Tjoeb 4000, waveform of undithered 1kHz sinewave at -90.31dBFS, 16-bit CD data.

The Njoe Tjoeb player's distortion signature comprised low-order harmonics at moderately low levels (fig.7), the third harmonic rising above the second as the load impedance drops. The subjectively benign second harmonic was the highest in level, at around -66dB (0.05%) with the player driving the 100k ohm lab load, which again suggests that the Njoe Tjoeb will be happiest driving preamps of high input impedance. Intermodulation distortion was moderately low in level (fig.8), and not particularly affected by the load impedance (at least while it remained above 10k ohms). But there is a reasonably strong component at 24.1kHz in this graph, which suggests that the player's reconstruction filter has been optimized for time-domain performance rather than for maximum stop-band rejection.

Fig.7 AH! Njoe Tjoeb 4000, spectrum of 1kHz sinewave, DC-10kHz, at 0dBFS into 8k ohms (linear frequency scale).

Fig.8 AH! Njoe Tjoeb 4000, HF intermodulation spectrum, DC-25kHz, 19+20kHz at 0dBFS into 8k ohms (linear frequency scale).

The Njoe Tjoeb 4000 did very well on my usual jitter-rejection test, producing just 180 picoseconds of wordclock-related spuriae (fig.9). Most of these were data-related (red numeric markers), though some low-frequency sideband pairs could be seen (purple markers). Note the very low noise floor in this graph, which covers the 7.7-14.7kHz range. The noise introduced by the player's use of tubes appears to be almost all low-frequency in nature.

Fig.9 AH! Njoe Tjoeb 4000, high-resolution jitter spectrum of analog output signal (11.025kHz at -6dBFS sampled at 44.1kHz with LSB toggled at 229Hz). Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz.

The AH! Njoe Tjoeb 4000's measured performance is generally good, with some areas of excellence. Only its unflat response and the presence of supply-related spuriae caused my eyebrows to rise.—John Atkinson