About the Hugo 2's output stage: It's stated as being Class A. I just wanted to clarify if it was Class A all the way through to maximum power output or if it goes into Class AB at some point.
If you had DSD256 or DSD128 files and decided to convert them into PCM, do you have any suggestions as to what software and what settings to use for doing so? I'm asking as while I'm reasonably good at comprehending information about digital music, helped quite often by your posts, many aspects of DSD don't make sense to me, such as how it is possible to have varying levels of volume in music which is only 1 bit.
Yes it's Class A with a 300 ohm load at 5v RMS; but for lower impedances it will drop into class AB mode - not that that adds significant extra distortion as the benefit the Hugo 2 has (in line with TT2 and Dave) is the second order analogue noise shaper. A simple way of thinking how that works is to see an extra high gain bandwidth amp whose sole purpose is to linearise the output stage, so that when it changes to Class AB mode we get no increase in high frequency distortion.
I'm afraid I don't have any experience of software DSD conversion.
As to converting the gain of a DSD stage. So how do we attenuate regular PCM? It gets done by (assume 24 bit volume coefficient and 24 bit PCM):
Output [47:0] = PCM[23:0] * Volume coefficient [23:0]
So when we multiply two 24 bit numbers together, we get a 48 bit output. If we wish to get it back to 24 bits, then we need to truncate; and we can do that by truncating the Output after adding appropriate dither; or we can noise shape it down to 24 bits.
DSD is exactly the same process, as DSD is actually 1 bit PCM, but in this case the Output is 25 bits:
Output [24:0] = DSD[0] * Volume coefficient [23:0]
But in this case we can't use dither, we have to use another noise shaper (or calling it another way another DSD modulator) to get you back to 1 bit.
Indeed, in the early days, they promised that EQ and volume could be done in this fashion - but it was quickly discovered, that cascading DSD noise shapers together created so much distortion and noise, it wasn't viable. This isn't an issue with PCM of course - you just run at an improved bit depth internally, doing everything you need to do, then truncate/dither or noise shape at the end for the final output.
So in short DSD is simply 1 bit PCM.
