Power Amplifiers

Spectron Musician III Mk2 Class D Stereo Power Amplifier


The Design

We have all heard about the use and misuse of negative feedback in amplifiers. This process feeds signal from the output back to the input, in the case of global negative feedback. The electrical phase of the feedback is inverted, and the difference between the input and the output is cancelled - not totally, but large enough to improve the quality of the output, because the difference between the input and output, besides the volume, is the presence of distortion. The circuit for negative feedback in a conventional amplifier usually consists of a resistor and capacitor. The resistor determines how much voltage (dBv) is fed back, and the capacitor determines the slope of the feedback, i.e., the relative amount of feedback in the low frequencies vs. the high frequencies. This is an analog process, and the time that it takes for the feedback to take place is about 2,000 nanoseconds, or 2 µs. Now, just think about this for a minute. A 20 kHz sine wave has a duration of 50 µs. So, because the negative feedback takes 2 µs to feed through to the output, the corrective effects of the feedback are 2 µs late, and extend 2 µs beyond the signal it is supposed to correct. The result is that we trade one type of distortion for another. While low order distortion is reduced, we end up with more higher order distortion. The bottom line with analog negative feedback is that it is a balancing act. What are you willing to give up in order to gain something else?

Shown below in the first diagram is a generic amplifier example illustrating conventional global analog negative feedback. I used a tube schematic because it is simpler to illustrate. By "Global" it is meant that the feedback goes from the output all the way back to the input stage. Local feedback refers to feedback between two stages or within a single stage. Notice how simple the feedback circuit is. Just a resistor and a capacitor in parallel (at the bottom of the circuit diagram). Electrons may travel at the speed of light, but only in a vacuum. When moving through an electronic circuit, they get delayed. Thus, the problem mentioned above.

The second diagram shows how Spectron utilizes negative feedback. It is accomplished by converting the audio to a PWM signal and using a digital amplifier (called the Forward Amplifier in the diagram). It performs its duties ten times faster (200 nanoseconds, or 0.2 µs) than conventional analog negative feedback circuits. Secondly, the Spectron includes a set of speaker cables that has a feedback circuit built in, so that the amplifier can sense the voltage actually being delivered to the speaker compared to what is being sent to the speaker, and correct for any error. Because the Spectron's digital handling of feedback is so fast, there is also less group delay. This means that each frequency in the music arrives at the speaker at the same time, so deep bass is in sync with violins, for example.


Because Class D is so efficient, it can deliver current into low impedances, so the Spectron can easily handle the low impedances found in planar speakers, such as electrostatics, where the impedance can dip to 0.1 ohm. There is also a large amount of headroom, again, due to its efficiency. The power supply is not wasting energy dissipating heat, so that energy can be called upon during extreme transients. There are 100 capacitors in the Spectron power supply, amounting to 33,000 µF of capacitance, and because there are so many capacitors, the output impedance of the power supply is extremely low.

Shown below is a photo of the rear panel.

Besides the XLR and RCA input jacks (with selector), as well as the gold-plated speaker binding posts, there are several toggle switches. These allow for phase inversion. This can be used if, for example, your preamplifier inverts the phase. The phase inversion switch on the Spectron would be used, in that case, for re-inverting the phase so that the output is properly phased with respect to the music. However, the phase inversion toggle has another function. By using the included Y connector, you input the signal from one channel of your preamplifier's output into both inputs on the Spectron, and toggle the phase inversion on one channel. You connect the speaker cable to the + binding post on the left and right outputs of the Spectron. What you end up with is a 1,500 watt (at 8 ohms), fully balanced monoblock power amplifier.

Above the speaker binding posts is a special connector for the voltage sensing speaker cables (these are an optional purchase).

The inside of the chassis shows that it is almost entirely occupied by the power supply. Look at all those capacitors. There are only two output transistors for each channel, because that is all that is required. Note that this amplifier uses a conventional power supply, and it is the output stage that is switched.


The way Class D amplification works is that the input signal is modulated onto a carrier signal, and in the case of the Spectron, the carrier frequency is 500 kHz. The modulation is in the form of 1's and 0's, and the duration of the 1 or 0 determines the voltage of the music signal being modulated. This is called Pulse Width Modulation. This differs from the CD itself, which also has 1's and 0's, but the duration is the same for all of them. In fact, if there is any error in passing the 1's and 0's along the bitstream from a CD, this results in something we have all heard of, called "jitter".

The signal on the carrier in the Class D circuit is demodulated, resulting in the higher voltage (higher than the preamplifier) output that goes to the speaker. But, you still have that 500 kHz carrier in the amplifier circuit to deal with. You don't want it going out to the speakers too, so a low-pass filter is applied to remove it.