The need for a preamp (volume control discussion)

[tensor9]

The methods of volume control are numberous these days. There are digital volume controls of different types, some which are bit-perfect, some not. Even for the ones that are (like on my PS Audio DirectStream DAC), lowering the volume brings the signal closer to the analog noise floor. There are stepped attenuators, and arrays of resistors of all sorts, and pots and so on. Then there's the gain...

So, what is optimal? Is a simple array of resistors the best?

For me, in my setup, I just don't need gain (my amp has plenty). I need to attenuate the signal as cleanly as possible. Currently I'm using the DS DAC directly into my amps and using the DAC's built in 20 dB attenuator with my DAC volume in the 80s for normal listening.

Does the average user NEED pre-gain these days?

What are your thoughts? Preamp, no preamp? What's the best method for attenuation?

 

[Daverz]

Some discussion of digital volume control here:

http://archimago.blogspot.com/2019/06/guest-post-why-we-should-use-software.html

While it's true that lowering digital volume brings the signal peaks closer to the analog noise floor of the DAC, for competent DAC the noise floor is usually way below the noise floor of the rest of the signal chain.

See also the discussion of the digital volume control in the RME ADI-2 starting on page 63 of their manual (PDF):

https://www.rme-audio.de/download/adi2dac_e.pdf

Quote:

"In fact an analog volume control has a (theoretical) advantage in only one point, namely the maximum signal to noise ratio at a higher level reduction. In reality, current circuitry overturns the theory, and the SNR at the output of such a device is no better than that of a digitally controlled one. This is even more true the better the DA converter works and the less noise it has [...]

The most often cited issue of a digital volume control is an alleged loss of resolution at higher attenuation. An example: 117 dB dynamic roughly equals 19 bit resolution. A volume attenuation of 48 dB (8 bit) leaves 11 bit of resolution. Such a simple, but important details omitting argumentation, usually ends with: the music must sound distorted in quieter parts, and the signal to noise ratio is down to a useless 69 dB.

The former is simply wrong, the latter irrelevant in practice. Indeed there is a reduced signal to noise ratio, but it doesn't matter, as the noise was not audible before (below the hearing threshold), and is still not audible after lowering the level. And the reduced SNR also applies to devices with potentiometers, since the potentiometer is never placed at the output, but in the middle of the circuit, followed by further electronics which also add some basic noise."

(My bolding.)