Columns Retired Columns & Blogs |
People love to bash Stereophile for whatever reason they may have, but this article is an example of truly superior professional writing. Entertaining and engaging while remaining incisive and informative.
After Michael Fremer reviewed the darTZeel NHB-18NS full-function preamplifier in the June 2006 issue of Stereophile, he bought the review sample. He subsequently replaced it with the NHB-18NS Mk.2, which he wrote about in June 2017. It has been his reference preamplifier ever since.
Although I performed a complete set of measurements on the original NHB-18NS, I never did so on the Mk.2. As darTZeel had sent a new sample of the Mk.2 along with the NHB-108 Model Two power amplifier that Jason Victor Serinus will be reviewing in the March 2022 issue, I took advantage of this opportunity to do those measurements.
I examined the darTZeel NHB-18NS Mk.2's measured performance with my Audio Precision 2722 system, repeating some of the tests with the magazine's higher-performance Audio Precision APx500. As I do not have test gear with the 50 ohm impedances necessary to test darTZeel's proprietary impedance-matched Zeel input and darT output connections, I tested the preamplifier's conventional balanced and unbalanced line inputs and outputs. The preamp's behavior using the proprietary connections can be inferred from the performance of its conventional, unbalanced operation.
The NHB-18NS Mk.2 uses a battery-based power supply. I ensured that the battery was fully charged before starting the measurements. Looking first at the line inputs, the maximum gain without the switchable 6dB input attenuation was 11.63dB for balanced inputs measured at the balanced outputs. From the unbalanced inputs, the gain was 5.4dB at the balanced outputs and 10.65dB at the unbalanced outputs. The "unity gain" setting of the volume control gave a gain of 0.27dB, left channel, and 0.1dB, right channel for balanced operation and 0.56dB and 0.75dB, respectively, for unbalanced operation. (These differences were too small to be compensated for with the balance control.)
The major difference between the original NHB-18NS and the Mk.2 was the replacement of the transformer-coupled balanced inputs and outputs with a direct-coupled topology. With the balanced input impedance set to normal mode, the input impedance was 18.8k ohms across the audioband. In "Pro" mode, which was the only balanced input option with the original NHB-18NS, it was an extremely low 640 ohms. The unbalanced input impedance was 22k ohms at 20Hz and 1kHz, dropping to a still-high 13k ohms at 20kHz. The balanced output impedance was usefully much lower than the original version's, at 199 ohms from 20Hz to 20kHz. The unbalanced output impedance was a low 153 ohms, again from 120Hz to 20kHz. The output impedance at the Rec Out jacks was even lower, at 79 ohms.
The frequency response from the unbalanced inputs to the unbalanced outputs was perfectly flat within the audioband, with the ultrasonic output down 1.5dB at 200kHz (fig.1). The frequency response measured at the balanced outputs with the balanced inputs was also flat in the audioband but was now down by just 0.1dB at 200kHz (fig.2). Both the frequency response and the channel matching with both types of inputs and outputs were preserved at all settings of the volume control. Both types of input and output preserved absolute polarity, which indicates that the XLR jacks are wired with pin 2 hot.
Channel separation (not shown) was superb below 2kHz at >110dB for balanced operation, >100dB for unbalanced operation, and a still excellent 92dB and 80dB, respectively, at 20kHz. The wideband, unweighted S/N ratio, ref. 1V output and taken with the unbalanced input shorted and the volume control set to its maximum, was a good 72.2dB (average of both channels). Switching in an A-weighting filter gave a slight improvement, to 80dB. The low-frequency spectrum of the darTZeel's balanced noisefloor with the volume control set to its maximum was free from power supplyrelated spuriae (fig.3), though the level of the random noise components is around 20dB higher than that of the Topping Pre90 preamplifier that is reviewed elsewhere in this issue.
Fig.4 plots the percentage of THD+noise in the NHB-18NS Mk.2's balanced output against output voltage into 100k ohms. The trace's upward slope with decreasing frequency below 2V indicates that the distortion components are buried beneath the noisefloor. The true distortion begins to rise gently above 3V but remains low up to the actual clipping point, which is just above 10V into this high impedance and still 7.5V into 600 ohms (fig.5). The unbalanced output clipped at 5.5V into 100k ohms (fig.6) and 3.4V into 600 ohms (not shown). These levels are well above what will be required to drive a power amplifier into clipping.
The balanced output's THD+N percentage against frequency at 2V into 100k ohms (fig.7, blue and red traces) was consistent across the audioband. However, into 600 ohms (cyan and magenta traces), there was a peculiar rise in the THD+N centered at 3.5kHz, which was very similar to what I found with the original version's balanced output. This rise was smaller in unbalanced operation (fig.8), but I doubt that it would have any subjective consequences in either mode.
As with the original preamplifier, the distortion is predominantly the benign second harmonic. This peak lay at 86dB (0.005%) for balanced operation (fig.9) and 80dB for unbalanced operation (not shown). When I repeated the spectral analysis at a lower output level using the APx500 analyzer, the distortion harmonics were even lower in level (fig.10). However, as with the 2006 measurements, there was a peculiar rise in the noisefloor between 5kHz and 8kHz. I have no idea what causes this behavior or what subjective consequences it might have, but it can also be seen in the spectrum of the darTZeel's output with an equal mix of 19kHz and 20kHz tones (fig.11). The actual intermodulation products, though, are well down in level: The difference component at 1kHz lies at 86dB (0.005%), and the higher-order products are even lower in level.
Toggle switches next to the phono input jacks allow the gain to be varied from 57dB to 72dB and the input impedance to be varied from 43 ohms to 300 ohms. As supplied for review, the phono input had a maximum gain of 71.3dB measured at both the preamp's main output and the unbalanced Rec Out jack. Set to 150 ohms, the input impedance was 150 ohms across the audioband. Both the gain and impedance ranges are appropriate for low-output moving-coil cartridges.
The rear-panel switches also allow for the activation of a subsonic filter and a choice of conventional RIAA equalization or so-called Neumann 4th pole equalization. The latter choice features an additional filter time constant of 3.18µs, which prevents the RIAA rolloff from continuing indefinitely with rising frequency. (The rationale behind this is to compensate for the fact that the gain of the record-cutting head doesn't rise indefinitely but starts to flatten out beyond 50kHz.) In the original measurements sidebar, I wrote that I wished that this was a switchable option, for reasons I discussed. The Mk.2 preamplifier grants my wish.
Without the Neumann EQ (fig.12, blue and red traces above 80Hz), the Mk.2's RIAA error is minimal and much lower than that of the original NHB-18NS. With the Neumann EQ (green and gray traces), the response rises above the audioband, reaching +7dB at 100kHz. The subsonic filter (blue and red traces below 80Hz) rolls off the low bass, reaching 6dB at 10Hz.
The phono input's channel separation was very good, any crosstalk lying below the noisefloor, and even with the high gain, this input was very quiet. The wideband, unweighted signal/noise ratio (ref. 500µV input at 1kHz and measured at the Tape Out jacks) measured a superb 74.3dB. This improved to 75dB when the measurement was restricted to the audioband, and to 83.6dB when A-weighted.
The NHB-18NS Mk.2's phono-stage overload margin was good, at 14dB across the audioband (ref. 500µV input at 1kHz), but will increase as the gain is reduced. Harmonic distortion was very low, with the second harmonic the highest in level at 83dB (0.007%) and the third harmonic a little lower in level (fig.13). With an equal mix of 19kHz and 20kHz tones at a level 7.5dB below overload, the second-order difference component lay at 50dB, or 0.3%.
In almost all respects, the NHB-18NS Mk.2's measured performance was better than that of its predecessor, though the line inputs still have a higher level of random noise than I was expecting.John Atkinson