Direct Stream Dac can support DSD 256 one day soon?

I didn’t mean to be ambiguous - The DS hardware almost certainly supports 4x DSD.

Which external hardware will work isn’t in PS Audio’s control.

I don’t speak for PS Audio re bridges/Octave and I don’t know what hardware the newer bridges/Octave, etc. will be using but PS doesn’t like the current output rate limits of the Bridges. Speaking of newer bridges, etc. - once again the limits might be the hardware / software of the sources, switches, etc.

Would 4x native DSD or DoP be easier to implement over the I2S input?

Hi Ted, and thanks very much for the response. I believe my Melco N1Z can output 4xDSD. The Melco is feeding a Matrix X-SPDIF 2 via USB, and the Matrix converts the digital signal to I2S. Matrix claims the X-SPDIF 2 can output 4xDSD via I2S. It would be awesome if the DS could receive/decode 4xDSD someday. Thanks again.

Can somebody enlighten me on 4xDSD. I assume that is a downloadable file you can buy from HD Tracks or wherever, with the intent of playing it via the front USB port on DS? As opposed to a physical disc that you will spin in the DS.

@brodricj Right,we are discussing playback of digital DSD files as opposed to physical optical discs. Here is an example of a recording that can be purchased for download in various formats up to and including 4xDSD:

I was thinking of 4xDSD playback through the DS’s I2S inputs (not USB) when I posted at #16, above.

My Esoteric does DSD512 however I haven’t been able to find any content published at that resolution.

Some people like to upsample to their DAC’s highest playback rate using computer software, i.e., upsampling RBCD to DSD512 for playback on an Esoteric DAC that tops out at DSD512. One program that provides this functionality is HQPlayer. As the DS does its own internal conversion to 20xDSD, I choose not to upsample ahead of the DS (although I have MinimServer set to convert 16-bit and 24-bit FLAC to 24-bit WAV in my Melco).

Minimalist said

Would 4x native DSD or DoP be easier to implement over the I2S input?

For the DS side they are the same. For the source, it has to be able to generate a clean clock that’s 50% higher for DoP then raw DSD. For some systems the multiply by three / two for DoP might be a little work.

My Sonore Signature Series Rendu supposedly supports DSD256 via its I2S output. With my limited experience testing DSD through it, I notice it always outputs DSD via DoP. Given your remarks saying it’s a little harder to do DS256 via DoP, I wonder if it really supports it. I guess I won’t know until (if) the Directstream supports DSD265 and then I can test.

I don’t mean to imply that it’s hard. It’s just that when many of the I2S devices (especially those based on XMOS chips for USB) were designed these high rates weren’t expected. For example there are some manufacturers examples for I2S (to and from ADC’s and DAC’s) that use master crystals that are too low for these higher rates: if copied naïvely they would limit the maximum rates. As another example I know of some I2S conversion boards that can do double rate raw DSD but not double rate DoP. On the other hand there are many pretty good clock converters/generators that allow very accurate synthesis of any output clock rate and/or conversion from any rate to any other rate. They’ve lowered their jitter in these devices quite a lot over the years and they are perfectly good for these I2S devices.

It’s hard to see in this photo, but the clocks say:

Crystek CCHD-957-25

Crystek CCHD-957-25


Would that be fast enough? The molex flex cable you see going into that board is the I2S output from the Mr MOD NMR UPnP module, which is underneath the i2s output board, that the Sonore SSR uses to decode/render the audio. According to the data sheet for the Mr MOD, the I2S clock information when using the I2S output is slave to external clocks, which would be the ones listed above on that I2S board. That begs the question, are these clocks fast enough.

Mr Mod NMR info if interested:

Those are fine clocks.

I’m sure they’re fine clocks. I guess what I meant to ask is whether or not you think these clocks would be able to do the high rate DoP signals based on those clocks. I have no idea how (or if I can) change the output from DoP to raw DSD. So I may be forced to use DoP if I want DSD256 out of the Sonore SSR. I suppose I should contact Sonore and ask if this is possible.

Sorry, I meant that they should easily do native DSD at those rates and probably DoP with 32 bit words (so they don’t have to deal with 24 bit words and multiply or divide by 3 somewhere.)

Hi Ted. I may be in a similar boat to Seegs108, as my Melco N1Z outputs DSD as DoP. I have the Melco connected to a Matrix X-SPDIF 2 via USB, and I set the Matrix to convert DoP to native DSD, which the Matrix then outputs to the DS via I2S. Matrix states that the two clocks in the X-SPIDF 2 are “Accusilicon high-quality femtosecond clocks as [a] reference clock source, which work at the state[s] of 44.1kHz, 48kHz and their multiple sample rate[s]. The [Xilinx Spartan XC3S50AN] FPGA automatically control[s] the frequency assignment and switching, and the femtosecond clock[s] ha[ve an] independent ultra-low noise LDO power supply.”

Do you think the Matrix’s Accusilicon clocks, which appear in this Matrix picture to be two variants of the “Accusilicon AS318-B Series Professional Audiophile Crystal Oscillator,” are up to the tasks of processing 4xDoP and 4xDSD? Thanks.

Very probably - those are some of the type of oscillators I mentioned above - they also claim to be designed for our audio quality concerns which is good too.

Hi Ted,

I purchase a lot of DSD files from NativeDSD, now almost all new DSD title available at DSD256, and charging at same price as DSD128, so go ahead for DSD256 purchase, getting DSD128 and DSD64 of same title for free.

The question is here, I’m using Roon for playback, the DS DAC profile at DSD128 Max, will downsample DSD256 to DSD128 before sending to DS, as I have more than one DAC that capable at DSD258 and even DSD512, I only want to keep one copy of same title in Server, playing DSD256 Native recording downsampled by ROON to DSD128 will have better sound, or playing DSD128 copy supplied by NativeDSD converted from same DSD256 master by Merging Technology Workstation?

I’m sorry, I don’t know how good ROON’s DSD converter is. They are smart people and I suspect they do a good job. If you can’t hear a difference I wouldn’t worry about it.

1 Like

One other crazy thing you might consider - downsampling quad rate DSD to 24/352.8k PCM - doing a quality DSD -> PCM conversion is much easier (or at least many more people have the chops) than doing a quality xxx -> DSD conversion. There’s a chance that converting to higher rate PCM and then letting the DS do the conversion to DSD might better than an on the fly conversion to DSD upstream. Anyway it’s probably worth a listen.


A clever idea, Ted.

Plus, most DSD recordings have already spent some time as DXD, at least at their edit points.