Direct Digital Amplifiers

Taken from the tact s2150 owners manual:

The digital input is taken to the central processor where it is reformatted into apulse width modulated signal of extreme precision. The pulse rate is measured atprecisely 384,000 pulses per second. Each pulse can have 256 different widths,with the narrowest pulse being a mere 10 nanoseconds wide. The clock frequencytherefore is 98 MHz. The central processor uses proprietary patented algorithms(Equibit) to arrive at exactly the right combination of pulse widths produce a highlyaccurate waveform. This is the most fundamental departure from conventional am-plifiers. TacT defines the waveform mathematically - we are not trying to follow oremulate a waveform by using feedback or feed-forward.Once the decision of the duration of the pulse is made the central processor con-trols FET-switches at the output with extreme precision. Voltage and current aredrawn from the power supply and fed to the speakers.The level of playback is controlled by adjusting the voltage of the power supply. Asthis voltage is switched directly to the speakers, it is of paramount importance thatthe power supply be totally free of ripple and noise. For TacT digital amplifiers, aswitch mode power supply of extreme precision with ripple rejection of more than135 dB has been developed. At full volume (voltage) the TacT M/S2150 delivers 58volts, equivalent to 150 Watts into 8 ohms. To reduce the volume the voltage of thepower supply is reduced. This means that the volume control is no longer part of anactive circuit.

Drubin, please keep us posted with your thoughts as you get to know the C390DD. I'm very interested in that piece! Thank you.

Cheers,

Scott

Bombaywalla, Adjusted analog quantity is duty cycle. For instance, when you adjust oscillator's frequency by turning knob it doesn't matter if output signal is square, triangle or sinewave - your adjustment is still analog. The fact that voltage jumps between two level's doesn't make it digital if adjusted quantity is duty cycle. Also, amplifier doesn't make any use of these levels other than converting to output voltage by taking average value (filtering).

what you wrote about osc freq adjustment being analog is correct but.........
when it comes to a class-D amplifier, it is well-known & well-documented that this amplifier is a discrete-time system. You cannot analyze a class-D amplifier using pure analog techniques. The switcher portion of the class-D amplifier is treated as a digital/discrete-time system & discrete-time + analog analysis techniques are jointly used to analyze the entire system.
Pick up any book on SMPS & you will see what I mean. Basics of DSP that you would learn from a text like Rabiner & Gold are used in the analysis of class-D/SMPS systems.
Even tho' the switcher's main function is PWM (which might be a analog quantity of duty cycle variation), this part of the class-D amplifier IS digital. Make no mistake about it. What I'm gathering from your comments is that I do not think that you have ever analyzed or designed a SMPS or class-D amplifier hence you are so vehement with your comments.
Open up a SMPS text book & read a bit of the analysis of the system. You'll quickly find out that it's a discrete-time system - part digital & part analog.

Distorted electric guitar (square wave) dosn't make it digital. FM radio is not digital in spite of voltage moving between two levels.

these were bad examples. Of course, a distorted electric guitar wave is not digital 'cuz there are stable operating points at the flat tops, flat bottoms & all points connecting the flat top to the flat bottom. This is way different from the class-D switcher output that is stable at only 2 points: logic 1 or logic 0 & nothing in between.
And, FM is phase modulation - a very analog concept. Today it is digitized in HD radio for more noise immune over-the-air transmission. So, while the over-the-air transmission is digital or discrete-time, FM radio is still analog.
we understand this.......

It is not discrete time system since discrete time system has countable number of states by definition. The states here are percents of duty cycle (quantity of interest) that changes with infinite number of states in similar fashion to change of frequency in analog Frequency Modulation. Yes I do design and use SMPS in my work. I'm surprised that you're confused by two levels in class D amp but not in FM radio signal. Again, for system to be "digital" at any point it has to have limited resolution (countable number of states) of quantity of interest - duty cycle, which class D amplifier isn't, being completely analog with unlimited resolution. In similar way FM radio signal has two states of voltage, but quantity of interest - frequency has unlimited number of states.

Very early primitive class D amps had digital modulator but these days almost everything is purely analog with unlimited resolution (Icepower, Hypex, NuForce etc.) Analog input voltage turns into analog duty cycle to become again analog output voltage by means of taking average value (filtering) of square wave (50% duty cycle representing 0V). Whole thing is continuous time system and has unlimited resolution unless, of course, amplifier has digital input which imposes resolution limit by definition.

I just read JA's review of the Devialet D-Premier. It's a different architecture and not "direct digital," but in a somewhat similar vein. I must hear it.