As I said I didn’t expect you aim for such very concrete sound related aspects during development, but more measurable and then things turn out to improve widely. I did expect DAC developers in the past didn’t get timing good, at least that’s what I experienced as faulty.
Of the following terms I’d say the first section got very good already in earlier high end DAC’s while the second further improved around 2000-2010.
But the last section to me really first reached a more than satisfying level with your DAC for those who knew high end vinyl playback.
Detail retrieval
Resolution
Overall tonality
Dynamics
Removal of flatness, recessed or thin/artificial treble
Natural tonal colors
Soundstaging, transparency
Coherence of music presentation
Proper ambiance, decay, reverb
Pace/timing
Natural sounding top end
Top end airyness
Just from reading some comparisons here in the forum of the DS with Chord or some other DAC’s, I guess I’d categorize the others several years back for what people found as “advantages”.
Knowing of experienced listeners who usually buy in a price range of multiple the DS and stick to your DAC, it seems you’re especially good in “reaching a price point without too many compromises”. I think I won’t listen to the TSS until I could buy it I’m still thinking the DS is enough DAC for a long time.
No dither in the PCM upsampler, I don’t narrow the word width after each stage of processing, I let it grow a little to keep all of the input accuracy. But I do it smartly - I know that I have a unity gain filter and audio input that didn’t clip (much) from the source so I don’t need more than one extra bit on the top even tho an analysis of the filter might indicate a growth of, say, 3.5. with the worst case input. On the bottom end I know that I don’t need (many) more bits than the accuracy of the input, I keep three more least significant bits than the previous releases.
I use linear phase FIR filters, I keep the input pass band as wide (and level) as I can and still be down to at least -144dB at the Nyquist frequency - with Redcloud, Hurron and perhaps Torreys the pass band was flat to about 21k. With the next release I’ll go just a little higher to 21.7k The filters I use are flat to about a millionth of a dB until just before the cutoff. (The previous releases kept them flat to a billionth of a dB, but I could save some FPGA resources by relaxing that a little.)
As I mentioned a few posts back with this release I also search for the best rounding of each coefficient to the size I use (approx. 30 bits.) This search looks at the filter performance for rounding a specific coefficient up and for rounding it down and chooses the one that achieves the best overall performance. It does this iteratively since previous decisions can affect new ones. As I mentioned this takes a while (many hours.)
The new filters are definitely a part of the new sound that Paul mentioned, but also by writing them myself I can better control the timing of doing the work. That also makes an audible difference.
Ted, once again a big thank you! for answering all of our questions. I find the inner workings fascinating and greatly appreciate your patience and thoroughness in sharing the process.
The last day of your posts have been well worth the admission price alone.
Well said. I find Ted’s limitless knowledge and insightfulness truly fascinating. I have read them all, saved his posts to my favourites and I am almost ready to start building my own DAC…
Cool. I look forward to trying the new release. I will certainly turn off all the HQPlayer upsampling, filters, and noise shaping to hear what you have done in this area. Thanks!
Well this is just tremendous news. Thanks so much ted for your hard work and efforts. The reason I chose this dac was for moments like this. I highly recommend ps audio on other forums for this reason. We all should.
Thanks for your valuable insight and suggestion…now if you could only pause and comprehend what was written you’d realize I’m not the person winding up any issue here.
Is most of the improvement in the new release in the PCM upsampling or will there be improvements for both DSD and for those that upsample to max PCM rate?
The code that’s explicitly changed will affect upsampling of PCM audio at 192k or lower. 352.8k PCM and DSD will benefit as well, I don’t know exactly how much, but the FPGA itself should be a little quieter and that helps everything.
Just wondering. Seems like you’ve been having an issue(s) for quite some time. Did you ever send the unit to PSA for repair or replacement. Would think that would be a pretty quick and easy resolution.
This has been discussed multiple times in the past. Unfortunately, the discussion led to unpleasant controversy. Let’s not go through this again. If you are curious, a couple of searches should lead you to the various discussions.
Perhaps I was led by my expectations, but I heard about what I was expecting: a quieter background and a more solid soundstage, giving more detail which all in all made the music a little more real and led to better separation of, say, lead singer from backup singers or let you listen to a particular instrument in an orchestra better. Perhaps a little unexpected was a little less interaction with my tinnitus, a little like moving out from a corner in a room which leads to a little less strain in the ears.
My wife liked the difference immediately, but I didn’t lead Paul or Darren, I wanted their opinions without my expectations. They confirmed my predictions.
I’m still thinking the DS is enough DAC for a long time.