NAD M2 Direct Digital integrated amplifier Page 2
In its simplest form, the output stage of a class-D amplifier comprises two complementary switches (usually power MOSFETs), one connecting the output terminal to the positive voltage rail, the other to the negative rail. When no input signal is present, the switches alternately open and close at a very high frequency, sending a series of full-scale positive and negative pulses to the outputin the case of the M2, ±50V. The switches are never on at the same time, and as the average voltage at the output is zero, there is no output signal. With an input signal present, the oscillator controlling the switches adjusts its duty cycle so that the full-scale positive pulses last longer when the audio signal is in its positive phase, and the full-scale negative pulses last longer for the negative signal phase. The higher the signal level, the longer the switches stay closed for each pulse, and the higher the average voltage fed to the output for each signal phase. You have an amplifier! For obvious reasons, this operating principle is called Pulse Width Modulation (PWM). And because the switching devices are either fully on or fully off, no power is wasted and the efficiency approaches 100%. However, a hefty low-pass filter needs to be in series with the output in order to prevent the high level of high-frequency switching noise from contaminating the neighborhood, and to reconstruct the analog waveform.
In practice, of course, there are many engineering problems to be solved in the design of a PWM amplifier, and many proprietary solutions are offered. NAD has collaborated with a British semiconductor company, Diodes Zetex Ltd., which developed a novel feedback topology in which the output pulses are continuously compared with a reference to produce an error signal. This error signal is integrated, digitized (at 108MHz), and fed back, with noiseshaping, to the PWM modulator. The signal is also monitored at the output low-pass filter, to give a low output impedance. The Zetex team refers to their topology as a Direct Digital Feedback Amplifier, and the NAD M2 is the first commercial product to feature DDFA.
In the main, a PWM output stage follows a conventional small-signal analog amplification stage. However, if the PWM stage can be fed PCM data directly, there is no need for there to be any analog amplification at all. This is what NAD has done in the M2, which is why they call it a Direct Digital amplifier. From the block diagram in NAD's white paper on the design of the M2, it looks as if the PCM data are first converted to a noise-shaped bitstream that is then applied to the 108MHz PWM modulator, along with the feedback signal.
There have been similar products before, in which a digital input signal is directly fed to an amplifier's output stage. The original Wadia company bet the farm on what they called a PowerDAC, but couldn't bring it to market successfully. The TacT amplifiers did generate some marketplace traction, and the Toccata PCM-to-PWM interface used by TacT was licensed to Texas Instruments in 2001. The Sharp SM-SX100, from the start of the century, was functionally very similar to the M2 but was compromised in terms of dynamic range, and never sold in significant numbers.
The advantage of keeping everything in the digital domain is that, provided the math is done with sufficient precision, the only source of noise and distortion is the PWM output stage itself. The M2 thus has the potential for sounding better than a conventional analog PWM design.
A couple of operational niggles were apparent when I first set up the M2. The shrouded EuroNanny speaker terminals accept spade lugs from one angle only, which makes dressing speaker cables a hit-or-miss affair. The terminal's opening is also too narrow to take the thick lugs now found on some high-end cables, such as AudioQuest's K2. Fortunately, the terminals do accept 4mm plugs, which I fitted to the cables I used. I was also initially puzzled by the M2's sample-rate display, which obstinately told me that the incoming sample rate was 44.1kHz, even when I was feeding the amplifier data at a different sample rate. According to Greg Stidsen, NAD's director of product development, even though the M2 does adjust itself to operate at the incoming sample rate, the display shows the sample rate set by the appropriate status flag in the datastream. Unfortunately, if this flag is left blank, as can happen with some source components, the display will default to "44.1kHz."
I had a problem with the first sample of the M2 (serial no. H99M200085) I received for review. After a couple of weeks of operation, during which time I left it on continuously, I got home one evening to be greeted by the sound of clicking relays and the front-panel message "Overheat." Unfortunately, I had no idea how long the M2 had been in this state. Following the instructions in the manual, I turned the amplifier off at the rear panel, unplugged the AC cord to make sure there would be a hard reboot, and left the M2 powered down for an hour. When I turned it back on, it passed signal for about 5 seconds, then displayed the "Overheat" message againthe amplifier was now stone coldand shut down. NAD shipped me a new sample (serial no. H99M200094), and I continued the review, including the measurements, with that sample, which performed perfectly.
I initially set the M2 up in its Fixed Gain mode, for use as a power amplifier, but after a New York minute's reflection, I realized that that was not going to take advantage of what the NAD could do. So I rethought the system's architecture, feeding to three of the M2's digital inputs the digital outputs of the dCS Puccini (for CDs) and Ayre Acoustics C-5xeMP (for DVD-Audio discs) players, as well as my Mac mini server via a Bel Canto USB-to-S/PDIF converter. For SACD playback, I fed the Puccini's analog output to the M2's balanced inputs, making sure its maximum output level was set to 2V so as not to overload the M2's own A/D converter (see "Measurements" sidebar). I left off the NAD's Soft Clipping feature. I also set the speaker compensation to ">8 ohms," which seemed appropriate for the Aerial Acoustics 20T V2 speakers with which I did most of my auditioning. The Aerial has a low impedance in the midrange, but much higher in the top two octaves; setting the M2 to "4 ohms," as I initially did, resulted in too much high-treble energy. For the PSB Synchrony Ones I used the "6 ohm" setting, to match that speaker's top-octave impedance.