Ben Bradley wrote:
The Class D amps are certainly making many inroads in many consumer
products, but how fast it replaces analog power amplifiers remains to
be seen. For stereos sold at Wal-Mart, the transition may already be
100 percent, but I doubt you could find a 'Class D' amp in high-end
audio stores, or if so, that it sounds as good as any analog amplifier
in the store. Analog amps in that area (admittedly not a large
consumer market) will surely be around for a long time, unless Class D
amps show substantially more improvement over current state of the
art.
It's still an art to develop a digital amp. There is a company Tripath
http://www.tripath.com/ which has specialized in high quality digital amps.
They use a delta sigma approach with feedback directly from the output of
the switching stage. The results are really good sounding amps, satisfying
even high-end requirements.
It's debatable whether class "D" is really digital. While it's true
that internally, the signal is switched, it isn't a digital signal.
It's pulse width modulation. The pulse width is continuously
variable, making it an analog quantity.
Pulse width modulation is digital. It fulfills the basic requirements of
digital, which means quantized and time discreet. The output stage has only
2 states and the repetition of that switching follows a fixed time pattern.
If the width is continously variable or in steps doesn't make it analog. The
advantage of a digital output stage is the high efficiency and subsequently
much higher power you can achieve with the output devices. So the main
fields of application are in very high power amps or very small lowish
(0.5-2W) output stages for cellphones and PDAs. But I'm sure in future most
if not all amps will be working like this.
That (punse-width modulation) is the traditional Class D approach.
Some "Class D" amps work just like sigma-delta modulators in A/D
converters. The output is either a 1 or a 0, and the time is also
quantized: the output only changes states at specific times, at the
sample rate, which is very much higher than the audio range, usually
around 2 MHz. These have advantages over straight PWM, but with an odd
disadvantage that there can be significant delays (as in a few
milliseconds) between input and output.
The delays ocurr only with long digital FIR filters, not necessarily. The
output needs to be filtered as well, usually a L/C passive filter, which
adds a delay according to its minimum phase characteristic. With a second
order filter this delay is usually only a few(2-5) microseconds.
The amplifier you referenced is a pulse width modulator. The input is
analog, and the output is analog. Noise and distortion are not
particularly outstanding, either. It's primary claim to fame is
extremely high efficiency, approaching 90%. It might be acceptable
for battery-powered equipment or automotive audio. Conventional
class A-B linear amplifiers are still preferable for high fidelity.
That was maybe 20yrs ago. It is still cutting edge technology, but since
Sony has introduced SACD with a fixed 2.8224MHz modulation, this argument is
totally bogus. It outperforms by far any CD quality. Sony offers all-digital
amps as well, which are optically coupled to the SACD-player, no analog in.
Sure there must be an analog output voltage because our ears work analog,
but this is done with that passive L/C lowpass mentioned before.
I'm currently designing a digital amp. It has analog inputs, and an analog
delta-sigma modulator, but then the signal is digitized and processed
digitally. Since this is not a commercial product, but for my own delight,
I'm willing to share my approach with anybody interested and can post the
schematics on ABSE when done. You can also drop me a mail to have a look at
the current state.
--
ciao Ban
Bordighera, Italy