Yes, We Use Class D Amps. No, You Shouldn’t Freak Out About That. Here’s Why.

Class D amplifiers are kind of like DSP in that once upon a time they weren’t very good, which caused a whole bunch of people suffering from Dunning Kruger effect (ahem) whom take to the internet with strongly held opinions to excoriate anyone who dared to suggest they might have gotten better in the past forty years. As it turns out however, the best Class D amplifiers in 2020 are exceptionally good, so much so that the benefits they have always presented in terms of efficiency, longevity, thermal management and weight savings no longer come at the cost of any real sacrifice in audio quality. To explain why this is so, let’s begin with a brief explanation of the difference between a Class D amplifier and Class A, Class B or Class AB amplifier, and how each works.

Let’s start with the older designs – Class A, Class B and Class AB. The distinction between these primarily refers to the topology of the output stage.  A class A amplifier’s active output device is always “on” – whether it’s a FET, a tube or a transistor, it conducts the entire waveform continuously. Conversely, a Class B amplifier uses two active devices that trade off between each half of the waveform, with one device handling the first 180 degrees and the other handling the second. This arrangement is far more efficient than a traditional Class A amplifier, which generally wastes about 3 out of every 4 watts of power it produces as heat. The tradeoff for a Class B amplifier is that it’s very difficult to get the exact handoff right in terms of the timing for when one device switches off and the other switches on, which can cause severe distortion. The most traditional compromise between these two is the Class AB amp, which is essentially a class B amp with overlap. Each side conducts a small amount on the other half of the waveform, which allows the amplifier to behave as a Class A amplifier for quieter, more sensitive passages and helps mask the crossing distortion when greater output is necessary. While less efficient than Class B, it remains much more efficient than Class A, and is generally considered a good compromise between the two amongst more traditional amplifier topologies.

Class D amplifiers use an entirely different approach called pulse-width modulation. Class D amps are often mistakenly referred to as digital amplifiers because this approach to decimated waveform reproduction bears a great deal of similarity to the manner in which sampled digital audio is stored and reproduced. Essentially it uses a very fast series of rectangular pulses to represent the amplitude and timing characteristics of the waveform instead simply amplifying the original waveform signal shape intact. This modulation, ideally, occurs at a much higher frequency than is audible, and a low pass filter is employed to keep the switching noise of this modulation out of the audible band. The primary benefit to this method is efficiency – 85-90% operating efficiencies are common. And because higher efficiency means less thermal loss, simpler and smaller heatsinks can be employed, which can dramatically cut down on weight. It’s also much easier to design a low pass filter that can suppress the high frequency switching noise (which in good modern Class D designs is usually higher than 100khz) than it is to design a competent high pass filter for a Class A or AB power supply to suppress PSU related 60hz harmonics to an equivalent level.

Class D amplifiers, however, are complicated to design well. The switching speed must be both precise and very fast (sometimes as fast as in the nanosecond range) to reproduce complex and especially high frequency waveforms accurately, and the switching noise must be properly isolated and suppressed into the ultra-high frequency range so as not to interfere with the audible band. Early class D designs didn’t do this very well: high degrees of particularly harsh intermodulation and odd order distortion were common, switching artifacts often remained in the audible band, and phase/magnitude linearity often began to suffer greatly in the upper octaves. Simply put, these early designs did, in fact kinda suck. But 21st century science has made the degree of precision necessary to make a truly high fidelity Class D design a reality, which brings us to Hypex Ncore, which we use across our range.

We primarily employ Hypex’s NC252MP OEM module, which you can see the detailed data for here. While there are many relevant performance metrics that typify a good amplifier design, total harmonic distortion plus noise (THD+N on most data sheets) relative to frequency and output power, intermodulation distortion specifically, frequency/phase linearity and output impedance or damping factor are the big ones. While, distortion, noise and linearity numbers all indicate in some regard the ability of the amplifier to reproduce the signal without coloration or alteration, output impedance or damping factor numbers indicate the ability of the amplifier to control the excursion of the driver, especially at low frequencies. Lower output impedances equate to higher damping factor – the difference is damping factor is a representation of both the output impedance of the amplifier and the impedance of the driver. The lower the impedance of the driver, the more critical it is for the output impedance of the amplifier to be negligible.

So let’s take a look at some of these numbers at a glance. Frequency and phase response are dead flat across the 20hz to 10khz passband, and are no more than -1.5db down and 30 degrees off at the frequency extremes. These variances are pittance relative to even the natural variations of our drivers, and easily corrected by our DSP. THD+N graphs are stellar – at half rated power the average is around .0008%, a frankly obscene number for an amplifier, especially one of this power. Most converters and line drivers would struggle to match that performance. As importantly, the IMD portion of the THD spectrum is also exceptionally low, with the worst peaks at around -110. In no uncertain terms, this degree of distortion is completely inaudible in even the best loudspeaker system. The world record for transducer distortion, currently held by a pair of electrostatic headphones, is nearly an order of magnitude higher than this number. Lastly, let’s take a look at the output impedance curve, which stays flat around .002 ohms for essentially the entire band, rising only to .004 ohms by 20khz. Not only is this flat an impedance curve rare and exceptional, as a point of reference, .002 ohms is roughly the equivalent resistance of a single foot of 12AWG speaker wire. Quite literally, the hookup wire to our drivers represents a higher portion of the total output impedance than our amplifiers do. This allows us to maintain an exceptionally high damping factor, which is especially important for the response of our subwoofers. As a whole, these numbers would all be considered stellar for a low wattage Class A amplifier, let alone an equivalently high output Class AB amp. In short, Class D is all grown up ladies and gents, and Hypex Ncore is proof of it. The weight, heat and inefficiency of a Class A or AB amp is simply no longer necessary.  

header.page-header h1 {font-size:40px !important;}