I guess most of my ramblings about new releases are buried deep in other threads.
Some of the improvements from Pike’s Peak to Yale were directly driven by feedback about the change between Pike’s Peak and 1.2.1 - I knew exactly which changes were explicitly made and could correlate them with most of the feedback. Over the same time period I found some ways of better controlling noise and jitter in the FPGA. Doing the first steps of that jitter/noise related work facilitated combining the best of 1.2.1 and Pike’ Peak. So I knew that the next release (Yale) would be better in some technical ways and I hoped that the jitter/noise work would also be evident.
The FPGA code roughly comprises misc control (e.g. talking to the display processor), digital input decoding/processing, PCM upsampling to 176.4k or 192k, upsampling DSD and PCM to 28.224MHz and the then down converting to single bit at 5.6448MHz with a Sigma Delta Modulator. The misc. control and initial PCM upsampling work are both relatively simple compared to the final big upsampler and sigma delta processing so I first worked on those simpler parts of the FPGA code to avoid wasting a lot of time if I was barking up the wrong tree. Yale showed that the noise/jitter work was productive and on the right track.
The next release of the DirectStream OS will certainly continue the jitter/noise reduction work by containing a reworking of the final 28.224MHz upsampler and a similar reworking of the sigma delta modulator. Until it’s done we won’t really know how big of difference in sound quality it will be, but it should be noticeable.
As in Yale, once again that work will enable some further refinements of the overall design. In particular it will allow the incorporation of some ideas I had last summer. One of them should lower the low level low frequency distortion continuing some of the work for the 1.2.1 release. The other should lower some jitter induced noise across the board. These two new work items should each also allow a measureable decrease in the ultrasonic noise which will be noticeable in some (most?) systems. I expect that their effects will somewhat overlap so I’m not making any specific predictions for the final measurable reduction in HF noise.
There are also a few bugs that need fixing and tho Yale makes most DS customers happy, there are a few who prefer an older release and I’d like to learn from that as well. I know that no software release will be a step forward for all customers and I’m pleased at the general reception that Yale got, but I can still try to do better and please more customers.
Some of the expected changes in sound quality with the new software will echo some of the changes from 1.2.1 and Pike’s Peak to Yale: A blacker background with less noise and jitter will add more definition/separation of the instruments/voices and deliver more detail without being analytical. Less jitter always seems to add bass definition and also a more accurate portrayal of the performance’s/recording’s timing.
Not picking on anyone and just trying to be honest, given the architecture of the DS I don’t completely understand how one would make changes to DSD sound production without making the same changes for PCM inputs. I.e., I think the differences that people report for DSD vs. PCM presentation in the DS are actually differences in the recordings, not a reflection of either inherent PCM vs. DSD differences nor any particular features of the DS. For example the recordings that I have that were mastered for DVD-A have a different sound character than the recordings that were mastered for SACD. The DVD-A recordings in general have stronger bass, more drive and slam. But I have no reason to believe that those differences comes from being distributed on a DVD-A vs. being delivered on an SACD. I think it reflects the styles of the recording and mastering engineers involved - they pick the editing and delivery system that best reflects their taste/style/preferences/preconceptions/etc. and this becomes a self perpetuating paradigm. (Don’t get me wrong, I do think there are some significant technical and sound quality differences between PCM and DSD, but anything you think is a feature of the DS’s PCM is obviously possible in DSD.)
In an ideal world the DS is getting better at being faithful to the recording, without being brutal about it. That’s a hard edge to walk but so far I believe I’ve done it by being as technically correct as I can be at each point in time and I have to believe that this is also the correct path for the future.