The recent discussion about chipsets and what constitutes what prompted us to bring in part 4 of the Software Design Notes. (also the AudioStream interview is quite heavy going as Thorsten does like the details).
Software Design Notes (4)
DSD – No ordinary DSD
There is DSD playback and DSD by iFi which we believe is different.
We are really pleased with our DSD implementation - it really is quite one of the most original DSD implementations out there. At the core, no data conversion and manipulation is done INSIDE this chip, so the DSD data is preserved Bit-Perfectly. An analogue filter is applied to the DSD data before it is sent to the output stage.
For us, "doing digital" means keeping the signal path original as much as possible; ideally from the ADC being DSD kept through to DAC being DSD (the same of course goes for PCm which we will cover later).
But in the case of the nano iDSD and micro iDSD, we have this:
Therefore, when playing back an originally-recorded DSD file through this Burr-Brown chipset on the iDSD, the listener has ensured that the WHOLE ADC > DAC path has remained native” in its DSD format. Like this:
The DSD chip we use is under full software control in order to gain access to many “undocumented features” (yes, they all have them). What are those “undocumented features”, we will announce them (a few should be a world’s first) closer to when we officially launch the micro iDSD.
Compared to the lion’s share of DSD-capable chips out there, most of them perform data conversion and manipulation of the DSD data and hence are no longer Bit-Perfect. This maybe the designer's choice, which is very much their prerogative but for us, we just about make it our mission to avoid such conversion processes.
Why? - conversion from one format to another is lossy. Hence to us, this is best avoided as much as possible.
What is the giveaway? If they do digital volume control on the DSD stream, this is very likely to convert the DSD data into NON-DSD data (does not necessary mean that they are converting into PCM, but converting into something non-DSD at the very least).
Even Pro Audio Studios use DAWs that convert
Pyramix do not even pretend to use DSD, in their DAW any processing means the DSD signal is converted to DXD (24Bit/352.8KHz) and processed as this, then converted back to DSD.
And Sonoma converts to DSD wide (aka PCM narrow), then applies a 24Bit scaling factor, ending up with 32PCM which is then converted back to DSD (remodulated from PCM).
It means that any processing of DSD in either Sonoma or Pyramix converts DSD to a form of PCM. Only "Tape Splice" (this refers to physically cutting the old magnetic tape and splicing two different tapes together) style edits can be done while retaining DSD.
If these VERY expensive DAW (Digital Audio Workstation) music recording/production packages cannot fade/volume control DSD without turning it to PCM, this begs the question of how come mainstream and far less costly commodity DAC's chipsets can?
This comes full circle back to why we selected the Burr-Brown chipset AND central to this why we use an analogue volume control (which yes, also has its issues but to us, these are far less sonically-damaging).
Edited by iFi audio - 6/17/14 at 10:13am