Class-D or switching amps, any opinions on??

Sam, repeat after me 'Class d is NOT digital'. repeat as necessary. read links I provided in earlier post.

Actually, most or all of the TACT products Sam referred to accept digital inputs, and provide a d/a conversion function in conjunction with a Class D power amplification function. So I think that would legitimize the fact that Tact headlines them as "digital power amplifiers."

Unfortunately, though, that probably contributes to the widespread misconception that "Class D amp" = "digital amp," which it certainly does not in the case of a more conventional Class D amplifier that has only analog inputs.

Regards,
-- Al

Magfan: The new NAD M2 ... operates in the digital mode, which I expect to be the real definition of 'digital amp'.

Agreed 100%. My previous post was not worded as precisely as it should have been, to the extent that it implied anything different than your statement quoted above.

The signal path through a digital Class D amp will consist entirely of either digital signals or switched waveforms, from the amp's input to just before the low-pass reconstruction filter which reconstructs an analog waveform at the amp's output.

However, the literature at the Tact site appears to indicate that their amplifiers do exactly that. For instance, see the section on page 8 of the 2150X Manual entitled "How Does It Work."

I know of no other Class D amplifiers besides the Tact's and the NAD which you mentioned which can properly be called digital amplifiers, although there certainly may be a few isolated exceptions that I am not aware of.

Best regards,
-- Al

Al - I'm not sure what they do to obtain decent resolution. In order to get 16 bit resolution and 20kHz bandwidth with traditional PWM clock has to be 65536x20e3=1.3GHz. It is way to fast and no room to filter out carrier. They might play games with modulating power supply at the same time or creating more states (more Mosfets and more voltages). It is getting very complicated to get real digital amp. Pure digital might be not bad - just look at HDTV.

That's an excellent question, Kijanki. I'm not particularly knowledgeable in that area, but I believe that what is done to provide adequate resolution is oversampling + noise shaping, conceptually similar to what is done in the so-called 1-bit cd players. I found the following paragraph here:

The PWM resolution is typically much more coarse (than that of the incoming pcm), in order to keep switching frequency above maximum audio sample rates of 192 kHz, whilst limiting MCK frequency. Typical ratios are between 32 (5 bits) and 256 (8 bits), giving a best dynamic range of 50 dB. Clearly, this means that other techniques are required to provide sufficient dynamic range, and the answer is oversampling with noise shaping. This technique essentially allows a time averaging of the PWM pulse widths to be applied, to deliver fractional resolution in the amplitude of the output signal.

The theory behind that is esoteric and also somewhat counter-intuitive, and I'm not sufficiently familiar with it to be able to explain it well, but I believe that is basically the answer. Note that the Tact manual I referenced indicates an 8-bit pwm resolution, within each pulse period (corresponding approximately to the 50db dynamic range mentioned in the quote above); a pulse rate of 384kHz; and a master clock rate of 98MHz.

Best regards,
-- Al

Here is a pretty good Wikipedia article on noise shaping. Very basically, oversampling + noise shaping allows the quantization noise resulting from the limited resolution to be mostly shifted up to frequencies that are higher than the audio information, allowing it to be digitally filtered without significantly affecting the audio.

Regards,
-- Al

My problem is with resolution. As you said TACT stated resolution from PWM is 8 bit. Where do they get required extra resolution?

Kijanki, though it may seem counter-intuitive, reducing the quantization noise that results from limited resolution is mathematically the SAME as increasing the resolution, apart from any possible side-effects of the digital filtering processes. So they get the required extra resolution by the oversampling + noise shaping + dither processes that I referred to.

Keep in mind that limitation of resolution to a finite number of bits is mathematically identical to summing a noise component ("quantization noise") into the signal. If I recall correctly, assuming random error distribution (which is pretty much assured by applying proper dither), the rms quantization noise amplitude is equal to the lsb increment (in volts) divided by the square root of 12.

The oversampling allows most of the quantization noise to be shifted to higher frequencies than the audio information, where it can be filtered out with minimal impact on the audio. Dither randomizes the process to eliminate "deterministic errors," as the article indicates.

I believe that the other pwm applications you mentioned, delta-sigma converters, sacd, and dds, do the same thing.

Regards,
-- Al