The output transformer mod was enough update for me for this year, next year, and the year after… 
1 Like
The specs are here: https://www.psaudio.com/directstream-dac/
My error above, DXD is 24/352 (who can get all these numbers right?).
The specs confirm usb and I2S handle this bit depth and sampling rate.
DXD (32 bit floating point at 352.8kHz), 24/352.8kHz and (hence double rate DoP) have been supported for quite a while. The I2S inputs and USB hardware support it, but not the Bridge, AES3, S/PDIF nor TOSLink hardware.
3 Likes
You guys should all clearly defer to Steven, as he writes more opinions than Ted.
3 Likes
Although I also don’t care about digital formats (different than about masterings), except getting the best version of a release (which would be DSD), I wonder that someone who cares for his streamer supporting a very high PCM sample rate (of which only very few recordings are offered) doesn’t care for having the chance to play a superior format like DSD (of which similarly few recordings are available but which makes a bigger difference than increased PCM sampling rates).
If then the DSD supporting device is also I improving every PCM recording by transforming it to DSD, I have no clue why I should limit myself to plain PCM only (for sound quality and format compatibility reasons), especially because it’s not even more expensive than the more limited alternatives.
1 Like
You are incorrect about a couple things: Most DSD master files are just that…DSD. The SACD era ones were largely done before DXD existed, and after that a common practice was to use DXD to process only the edit points and fades and not the bulk of the audio. For a few years some DSD done using one particular workstation, Pyramix, (not Sonoma or Sadie) was converted to DXD and back. In recent times more recordings are done natively as DXD, and they let you know it.
24/192 and 24/384 do sound different, and you can say that the higher sample rate sounds more lifelike, but they aren’t that different compared to DSD. DXD is like 24/192 on steroids, more air, more percussive, more image height, but it is marketing that equates it with DSD, because they don’t sound the same. I understand that they separate 192 from DXD to establish a certain standard, but DXD isn’t a quantum leap above 192. And yes, source files matter more than upsampling.
1 Like
Thanks for the reply, @tedsmith. I have several 24/352.8 albums purchased from HighResAudio, e-Onkyo and NativeDSD. Roon plays them nicely through this bundle from Sonore which connects to my DS via I2S. They sound great with the Windom firmware.
Some time ago, I bought a few 24/384 albums such as this, this and this. Sadly, those albums aren’t available in 24/352.8. Roon is aware the DS is limited to 24/352.8, so it downsamples 24/384 albums in half to 24/176.4. Also, I would like to buy this album in the format it was recorded in, which is 24/384. It would be great if Roon doesn’t have to downsample them for playback throught the DS.
Since DSD256 via I2S is in the works, I’m wondering if the ability to decode 24/384 is also on the table? 
Besides an extra filter, 24/384 doesn’t divide 44100644 like 705.6kHz does so 384 is harder in more than one way compared to 352.8k
Thanks to nativeDSD, I think you will find DXD is the predominant master file format. Sorry for the long post, but the key point from their site is in bold below.
About a month ago I was looking to buy this album and found it was only available in DSD format, as I was not even aware that my system could play DSD files. Turns out it can receive DSD64.
Not knowing anything about DSD files, I read what nativeDSD had to say and almost everything I’ve said above came directly from their pages, to which I posted the link. I also spoke to the Chief Engineer of Devialet.
With regard to master files being DXD, nativeDSD say:
Most new DSD recordings released today are made with multiple microphones that are digitally recorded, then level balanced and mixed in DXD during post production, yielding a DXD edited master. That’s then followed by converting back to DSD for consumer delivery.
You can find that here, about two-thirds of the way down.
So, unless I have read this incorrectly, the edited master file is DXD, but may be converted to DSD without actually changing the sound. This implies the DSD format is only used for consumer/marketing reasons, because DXD can be handled by most good DACs as it is just high rate PCM.
The DSD DAC can handle DXD (as per their website, I posted the link earlier) and as Ted confirmed, the limiting factor is the cable type.
I referred to Ted’s comment in 2017 about the limitations of the Bridge, which is here:
I was aware of the limitations because they were the reason I did not buy it.
The good news is I found a 16/44 PCM version of the desired music here. It is excellent.
I have no idea how SACD was produced when it started as I never used it. I only became aware of DSD when I bought a Linn system in 2009 and started buying their downloads. They were using DSD and I must assume they were editing in DXD, which they continued to do until 2015.
24/192 PCM has been their standard since then.
I clicked through a dozen titles on nativeDSD and DXD was the predominant original format.
I think all these different formats give rise to a lot of confusion and the first post I responded to was based on the assumption that more data = better sound. The reality is that more data is likely to result in hardware and cable issues that cannot be solved by software updates and certainly more cost.
At the end of the day, I bought the music at a cost of £9.75, whereas had I bought the highest data format DSD512 it would have cost me £35 and I would have had to set Innuos to reduce it to DSD64 to play it.
Hi Ted, just curious, were the DSD DAC to be updated, would it not be an option to use Dual AES/EBU to facilitate DoP up to DSD128 for those who want to play DSD files, don’t have an I2S source or don’t want to use USB?
Do you have anything against Dual AES/EBU? It is being used by an increasing number of European manufacturers.
The TSS does have dual AES inputs and the prototype for a new DS has dual AES inputs.
I don’t understand the intent of this statement. My original prototypes (at least the first ones that supported PCM) had dual AES. The DS was designed to fit in the PWD chassis so it had the same inputs and outputs that the PWD had.
4 Likes
Thanks Ted. Was not aware TSS had dual AES.
I appreciate the PWD and DSD had the same format as I had the PWD and the option was the kit upgrade.
This is the recording session referred to above. You might think that if you wanted to do a pure DSD recording with a couple of microphones, a solo violin would be the ideal choice. However, it was done in 5-channel DXD. There does not seem to be a lot of equipment involved, the power cable clearly came from a hardware store and they spent at least 2 minutes stacking some chairs to make a bit of space. The result was utterly superb.
1 Like
It does seem like DXD is used a lot now, but the preponderance of DSD recordings that were done by large scale labels were masters from analog, two channel direct DSD recordings, or PCM upsampled. The Sonoma system, which Octave Records uses, does DSD multichannel editing without any PCM conversion.
IMO the best uses for DSD are straight from tape, direct two channel recordings, or multichannel recordings using an analog mixer (Cookie Morenco and also Gus at Octave Records do this). The DXD thing is if you want to mix inside the computer using digital plugins. I don’t think any of these processor plugins sound as good as analog hardware, so the best DSD recordings/mixes are going to involve analog recording, mixing, or mastering in the process.
Btw, when Native DSD or whoever converts DXD to DSD, it definitely changes the sound, in a similar way as when you send DXD to the DSS. They sell these files for people that don’t necessarily have the best upsampling capability, and who prefer the sound signature of DSD. But IMO the upsampled DXD recordings won’t sound as good as real native DSD, although they may be able to accomplish a better recording.
2 Likes
Thanks for the response, @tedsmith. I suppose programming the DS to decode 24/384 is difficult. If it won’t make the next firmware, perhaps the one after?
@stevensegal, I was asking about 24/384.
The 24/352.8 albums I bought from HighResAudio, NativeDSD and e-Onkyo play just fine on my DS DAC.
I thought they all would, usb or I2S.
That’s very much how I understood it. The only DSD download that I bought that I really thought was exceptional was this one:
From a lost tape, a pure analogue to DSD transfer.
I read that the Sonoma system was developed in close association with Sony. I have no idea how many recording studios actually used it.
3 Likes
A wonderful album on vinyl as well. Strongly recommended for fans of Jazz trios and Bill Evans.
1 Like
I just bought this in DSD128. It is a phenomenal recording and should be an example of what Paul should be striving for with Octave Records. The piano sounds so realistic and live. Not at all unrealistic like what we got with “Out of Thin Air”.