darTZeel NHB-108 Model One power amplifier Measurements
As mentioned by Wes Phillips, its generous output stage bias causes the NHB-108 to run hot. After the preconditioning period of running at one-third power into 8 ohms, its chassis and heatsink were very hot. (A thermal cutout shuts the amplifier down if the chassis temperature exceeds 70ºC.)
The amplifier is very fussy about how it is used. Turning it on without a load attached blows the 6.3A fast-blo fuses, as does unplugging an input jack. Several times during the testing I inadvertently blew fuses, and got quite proficient in replacing them, even though the fuseholders are hard to reach and manipulate, given their placement inside the chassis adjacent to each power transformer.
Unusually, there were DC offsets present on the review sample's output terminals. With the NHB-108's input short-circuited, I measured 98.5mV on the left channel, 115mV on the right. The manufacturer hand-trims each amplifier before shipping; as this sample had seen a lot of travel over the past two years, it's possible that its DC operating points had gone out of alignment. (Though it's fair to point out that the DC present on the amp outputs would not affect WP's auditioning—the woofers of his speakers would have been moved very slightly away from their rest positions—it might have explained why the amp was so quick to trigger its protection during testing.)
I did a complete set of measurements with the NHB-108 set to its Hi condition, with some follow-ups set to Lo. The voltage gain at the Hi and Lo settings was 26.3dB into 8 ohms, absolutely normal, and the amplifier was noninverting; ie, it preserved absolute polarity. The XLR jacks on this sample input were simply connected in parallel with the unbalanced RCA jacks; current production, I'm told, has true transformer-coupled balanced inputs in addition to the unbalanced and 50 ohm Zeel inputs. The RCA jacks offered a higher-than-specified input impedance, at 220k ohms at 1kHz, dropping to 110k ohms at 20kHz. (I didn't check the Zeel inputs.) As a result, the NHB-108 will not load down preamplifiers.
The NHB-108 lacks overall negative feedback. According to the designer, the only feedback in the input stage is degenerative (an unbypassed resistor in the emitter circuit of each device), while the output stage is "full open-loop, with no feedback at all, no emitter resistor, no serial inductance whatsoever." A small, local, symmetrical feedback is wrapped around the middle, second stage, but there is no feedback path from the output to the input. Amplifiers lacking overall loop negative feedback can have high output impedances. However, the NHB-108 measured quite well in this regard, the source impedance remaining around 0.34 ohm across the audioband. As a result, the modification of the amplifier's frequency response by the Ohm's Law interaction between its source impedance and the impedance of our simulated loudspeaker load (fig.1, top trace at 2kHz) stayed within respectable ±0.3dB limits.
Fig.1 darTZeel NHB-108, frequency response at 2.83V into (from top to bottom at 2kHz): simulated loudspeaker load, 8 ohms, 4 ohms (1dB/vertical div., right channel dashed).
Into resistive loads, the measured small-signal response (fig.1, lower pairs of traces) revealed a very wide bandwidth, with the amplifier's output down by just 1dB at 148kHz. (This behavior was the same in both the Hi and Lo conditions.) As a result, the NHB-108's reproduction of a 10kHz squarewave was superb, with small risetimes and an absence of overshoot and ringing (fig.2).
Fig.2 darTZeel NHB-108, small-signal 10kHz squarewave into 8 ohms.
One thing I noted during the measurements was that after an abrupt change in load impedance, the amplifier's output level continued changing very slightly, settling asymptotically on its new steady-state value. I also noticed this phenomenon with Ayre's AX-7 amplifier (reviewed in October 2003), which resembles the darTZeel in not having an overall feedback loop.
The darTZeel's channel separation was excellent, at 95dB L–R and 100dB R–L. But this is still a higher level of crosstalk than I was expecting from the amplifier's physically dual-mono construction. (The input of the undriven channel was physically short-circuited with a grounding plug for these measurements.) Perhaps the need with this early-production sample to connect the two channels' speaker ground terminals introduces some small degree of crosstalk. One reason I could see this low level of crosstalk on my oscilloscope screen was because the NHB-108's noise floor was superbly low. Referenced to an output level of 1W into 8 ohms, the A-weighted signal/noise ratio was 91.4dB. Even with a wideband, unweighted measurement, the S/N was 76.5dB.
Again due to its lack of loop negative feedback, the NHB-108 is not a low-distortion design. Fig.3 plots the level of THD and noise present in its output (set to Hi) as the output power rises from 100mW to clipping (1% THD+N) into 8, 4, and 2 ohms. The distortion level at low powers is relatively high and almost doubles with each halving of the load impedance. Even so, the amplifier's linearity remains reasonably constant over a wide range of output power, meaning that the circuit's operating conditions will not be modulated by the signal. Clipping occurs quite abruptly, but the amplifier easily exceeds its specified 100Wpc/20dBW output power. I measured 160W into 8 ohms (22dBW), 230W into 4 ohms (20.6dBW), but 162W into 2 ohms (16.1dBW). The manufacturer strongly recommends using the Lo setting if the amplifier is to be used with low impedances. However, the amplifier then delivers significantly lower power at clipping—43W into 8 ohms (16.3dBW), 68W into 4 ohms (15.3dBW), and 95W into 2 ohms (13.75dBW)—with no reduction in low-power THD.
Fig.3 darTZeel NHB-108, Hi setting, distortion (%)vs 1kHz continuous output power with one channel driven into (from bottom to top at 10W): 8 ohms, 4 ohms, 2 ohms.
The distortion, though relatively high, also remained constant with frequency, as can be seen in fig.4. Of more subjective importance than the absolute level of distortion present in an amplifier's output is its harmonic spectrum, and here the NHB-108 did well. As can be seen in fig.5, the distortion is predominantly low-order harmonics. At low powers, the subjectively benign second harmonic was dominant; at higher powers, the third became dominant (fig.6).
Fig.4 darTZeel NHB-108, Hi setting, THD+N (%)vs frequency at 2.83V into (from bottom to top): 8 ohms, 4 ohms, 2 ohms (right channel dashed).
Fig.5 darTZeel NHB-108, Hi setting, 1kHz waveform at 1W into 8 ohms (top), 0.165% THD+N; distortion and noise waveform with fundamental notched out (bottom, not to scale).
Fig.6 darTZeel NHB-108, Hi setting, spectrum of 50Hz sinewave, DC–1kHz, at 55W into 8 ohms (linear frequency scale).
At that point I still had one more measurement to perform, the high-frequency intermodulation test, in which I drive the amplifier being tested with an equal mix of 19kHz and 20kHz tones and turn up the signal until the amplifier just clips. I then reduce the level by a couple of dB and perform a spectral analysis of the amplifier's output signal, to see what tones are present other than the original 19kHz and 20kHz components under these very demanding conditions.
Unfortunately, when I approached clipping with this signal, a click emanated from the NHB-108, which I assume was the crowbar protection circuit cutting in. Before I could reach over and tug the AC cord loose, there was a loud bang from the left-channel output devices and my test lab filled with blue smoke. Once my ears had stopped ringing, it turned out that, when I'd last replaced the fuses, I had inadvertently used a slo-blo type for the left channel. When the crowbar circuit operated, the right-channel fast-blo fuse had blown as required, protecting the output stage. Unfortunately, the left channel's devices blew up to protect the fuse!
The NHB-108's measured performance indicates that, with the exception of its slightly bent transfer function, designer Hervé Delétraz has avoided the problem most commonly associated with the absence of the usual overall negative-feedback loop. But that highish DC level and the hair-trigger crowbar protection make it difficult for me to recommend on a general basis. The darTZeel NHB-108 is like a Shelby Cobra—a handbuilt, high-performance design that makes demands on its always gratified owner, rather than a plug'n'play, never-very-gratifying Ford Taurus. But given that caveat, I was impressed by the NHB-108, and my own auditioning echoes that of both Wes Phillips and John Marks: I found it a very sweet-sounding amplifier.—John Atkinson