It’s not a DSD world …. yet – says Paul McGowan


In my opinion the future of all high resolution capable DACS is DSD.  As mentioned earlier, most modern DACS are already converting PCM to DSD internally anyway, even many low cost USB only DACS that handle 192kHz/24 bits are Delta Sigma converters (at their hearts) and thus the architecture used is essentially that of DSD.

So one might ask why not start with DSD in the first place?  Why convert from PCM to DSD when a pure DSD stream would make more sense?  There are a number of good reasons why this hasn’t yet happened more.  Certainly there’s an increase in DSD downloads available; for which you need a DSD capable DAC to play them.  There are also a few original PCM recordings that have been converted to DSD – a practice I find loathsome at best, certainly not on the up and up at the least.  An original DSD recording is one thing, converted analog to DSD is a great practice, converting original PCM to DSD and calling it “DSD” is not kosher IMHO.

Now let’s understand DSD isn’t all that friendly to recording engineers and studios.  For one thing, you cannot edit DSD.  You cannot digitally EQ DSD.  When a recording studio wants to make edits or EQ the raw data they must first convert it to PCM, edit it, then convert it back to DSD.  So that’s a problem – but it turns out this is ok if you start with DSD, go to PCM and then back again.

Modern recording equipment that can handle DSD, like the Sonoma workstation used by our friends Gus Skinas and Cookie Marenco, can go from DSD to PCM and back again with nary a hitch in the get along.  There doesn’t seem to be a way to go from an original PCM recording to DSD and gain any of the advantages of DSD (which kind of makes sense), but going from DSD to PCM and back again is apparently ok.  But it’s not recordings or media I am really focusing on.  No, it is DAC architecture.

Modern DACS, for the foreseeable future, will have to include a PCM to DSD converter.  The trick to building superior DACS, I believe, will be to base the newest generation of DACS around DSD centric engines with PCM converters tacked on, as opposed to PCM centric architectures using a DSD Delta Sigma converter tacked onto the back.  In fact, if you remember our post Lost my bits, I posted a block diagram of the Wolfson DAC chip we (and many others) use in the PWD.  There you can see a PCM centric DAC with the capability of a DSD input.  This describes 99% of the available chip sets out there for audio designers today.  No effort has been put into manipulating the volume in DSD, or anything else other than just bypassing the PCM converter and giving designers a direct shot into the Delta Sigma modulator section.

So the current crop of the very best DAC chipsets available are what I would think of as generational hybrids based on the reality of a PCM based audio/music industry.  There’s big changes ahead in that field – of this I can speak on a personal experience basis.  The future’s bright indeed.

Let’s attempt tomorrow to explain how a 1-bit DAC works.

This entry was posted in WRITERS and REVIEWERS and tagged on by .

1 Comment

  1. “…but it turns out this is ok if you start with DSD, go to PCM and then back again”

    To me, this seems like real angels-on-the-head-of-a-pin stuff. The way I see it, the sigma delta modulation technology was invented as a means to get closer to the ideal mathematical sampling theory with real world hardware and PCM storage. So yes, a PCM recording made with modern ADCs and played back with modern DACs both utilising SDM technology would be better in terms of ‘artefacts’ than early digital recordings. I can see how it must have looked superficially obvious to some bright spark that if the output of the DAC could be driven directly by the stream generated by the input stage of the ADC that it might make some sort of sense to bypass the intervening PCM conversion. But about five seconds later it should have dawned on them that *any* manipulation of the sampled audio (one of the major benefits of digital audio) would negate this idea. Five seconds later they probably said “But the punters won’t realise that!”, and the SACD campaign swung into action with the same vigour and attention to technical validity as the earlier MiniDisc campaign.

    Converting the SDM stream to PCM and vice versa are mathematical operations (which may or may not be lossless depending on the relationship between source and target sample rates, bit depths etc.) and it matters not where in the chain they occur, whether at the ADC or at the mastering stage, or at the player. It seems to me that if you are going to convert to PCM *anywhere* in the chain, you are better off keeping it as PCM for as long as possible thereafter, which allows the end user the option of performing room correction, active crossover filtering etc. without another SDM->PCM->SDM conversion. So if you are digitising a pure analogue source and can guarantee that no subsequent digital manipulation is going to occur (a tiny proportion of recordings), the DSD system may have some minuscule level of theoretical advantage, but that conversion to PCM anywhere in the chain erases that tiny theoretical advantage.

    However, the placebo effect is probably a million times more powerful than any of the technical issues, so the real DSD advantage is down to the marketing campaign, rather than the almost-irrelevant, flimsy technical foundations!

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

Follow us

Follow Howard Popeck on TwitterFollow Howard Popeck on Facebook


  • NEWS

© Copyright Our Listeners Club 2006-2013. All rights reserved.