View Full Version : Volume on SB2 - is fixed for digital better?

2005-04-09, 03:30
According to the UPS tracking website my new SB2 is just a few hours away so I am getting ready for the installation and testing. One question which has always puzzled me slightly is whether to fix or not fix the volume level on the digital out. I run my current SB1 into a DAC and have always selected 'fixed' for the digital output due to the slightly (to me) unclear statement on the web UI about better analog quality.

I am very keen to ensure that I extract maximum audio quality from the SB, which is one of the reasons I am looking forward to the SB2 with its promise of design features for the audiophile. Can anybody comment on what, if any, will be the difference for the SB2 with fixed or non-fixed digital output? I would prefer to have it variable since my amp has no remote volume control and being able to change volume from the couch is defintely convenient. However if it compromises the audio quality in any way I will stick to fixed.


2005-04-09, 05:02
I don't know how the digital volume control is implemented in the SB2, but generally the best quality is obtained from a digital device at full output. If the volume control is implemented purely digitally then as you lower the level you reduce the number of bits the audio can occupy, and this may make quantisation audible if there is high gain in the analogue side following D/A conversion.

In addition the digital reduction of volume generally requires requantization, any instance of which requires redithering. If dither is sub-optimal (ie not TPDF in most cases) then you will also lose quality here.

If it /is/ dithered properly though, the dither will smooth out quantisation errors. The purpose of dither is to convert quantization errors into de-correlated, benign noise, and in a perfectly-dithered system, haha, the number of bits determines only the noise floor and not the resolution (honest).

An analogue volume control or digitally-controlled attenuator in the analogue path after the D/A converter will also deal with this. If you have a digital signal processor (eg a preamp or surround controller with digital inputs) it may well use a combination analogue/DSP approach which will also avoid the problem.

The real test however is whether or not you can hear the degradation when you turn the volume down. If you're turning the volume down to listen to it quieter (rather than to compensate for excessive gain later on) you probably won't notice. And if you are not listening to losslessly-compressed formats (WAV, FLAC, Apple Lossless, etc), the artefacts of the lossy encoding may well dwarf any problems with digital level reduction unless it's turned down a long way.

--Richard E