Dithering: Digital Volume Control (To-Be) Made Right

The following articles show how a well implemented dithered volume control could let us forget about expensive preamplifiers, and use the PC or other digital sources to control volume without losses.
As the articles below will explain, dithered volume control works the best the higher the quantization goes (24 bit and over)

Digital Level Control – done the right way by Daniel Weiss
Dither on Wikipedia – Well made explanation
Dither Explained – Detailed Article for the Sound Engineer

Further reflections:

One of the most appealing uses of dithering would be to let us get rid of expensive preamplifiers (taking away one step from the chain) and, when the DAC’s make it possible, connect it directly to power amps. Most DAC’s, such as Idat-44 or Bidat, have as output an opamp, and driving a power amplifier load (100 kOhm, for a generic example) would be easy for them. Most preamps are just an op-amp too, so, impedance matching wouldn’t make much difference.

The best candidates for digital volume control are 24 (or more) bits DACs. Thus, in case of a 16 bit audio stream, the first 16 digits would contain the music, while the remaining 8 bits would be zeroes. Lowering the volume digitally, in such conditions, would take advantage of all 24 bits (not just 16).

As an example:
When writing the bits from MSB to LSB, a 16 bit sample might look something like:
011111111111111100000000
After 12dB volume control, it might look like this:
000111111111111111000000
Since the lowest 8 bits were zeros, none of the data is lost/truncated. This is because a 24 bit DAC receives the 16 CD audio bits as bits 8..23 with 0..7 set to zero

The above advantage wouldn’t be applicable to DACs capable only of 16-bit.

Generally, for a 16-bit DAC, the only clever way left, in order to exploit digital volume control could be to use some inline attenuators (cables with resistors inside) between the DAC and power amplifier, so to provide a starting attenuation, and have to attenuate less digitally: the less digital attenuation, the less digital degradation (this is especially true for lower resolution music).