I use HQPlayer to upsample in Roon instead of Roon itself. I find improvements of upsampling PCM to DSD vary by recording. Some times I prefer the upsampled DSD version, at others the PCM is more dynamic while the DSD version is more homogenized. I have over 400 DSD albums, many of which are the best sounding in my collection. Its certainly worth starting to collect them.
Just an update. It’s landed in Australia! Yay! It flew into Sydney today (Tuesday) so it’ll still be a few more days to come down south to Melbourne. Will let you know
I would love to hear this difference.
On my system, I can’t tell the difference.
Update. Picked it up from the post office this morning. New-toy excitement. Installing this afternoon. Report back soon.
OK, got it working in a rudimentary fashion. Currently running from a USB3 port on a win10 gaming PC (my standard home PC) using a commodity USB cable. Matrix out is a short, non-glass optical cable to the DS Sr. Still have normal USB from PC to DS and Bridge II connected. My reference material is a recently released 24/96 DG download from Presto Classical, “Anne-Sophie Mutter / John Williams: Across the Stars”.
Even with this less than optimal setup, I think I’m hearing improvements, certainly over the simple USB but also over the Bridge! Seems to be a tad more luscious blackness. That is, silence between the notes, if you get my drift. Anyway, more to do…
For best results, disconnect the USB and Ethernet cables from the DAC. Also, I recommend trying an HDMI (I2S) cable from the Matrix to the DAC (disconnecting the TOSlink cable from the DAC when you do).
I second @bootzilla recommendations. Also, drive the Matrix with a quality LPS for best effect.
Agree. Both top priorities on the todo list.
Based on what some of you have done or recommended, I’ve ordered an UpTone UltraCap LPS-1.2 and Wireworld Silver Starlight USB and HDMI cables.
In the meantime, I’m learning how to operate the Matrix. Again based on what some of you have done, I’ve installed HQPlayer and set it up to upsample non-DSD input to DSD128. The light on the Matrix is white for everything from HQPlayer. It falls back to green for other sources like standard PC sounds and from other apps. Upsampling to DSD128 seems to be the highest that the I2S input on the DS will take. Is that what you’ve found?
Yes, DSD 128 is the highest for DS DACs. . When I upsample PCM to DSD in HQPlayer my Matrix light is always blue. Not sure why or how yours is white, not that it matters, just curious. What are your HQPlayer DoP settings?
I do prefer upsampling PCM to 352k in HQPlayer instead of DSD. I find upsampling PCM to DSD in HQPlayer creates a less dynamic sound than allowing the DAC to do the DSD upsampling.
Here’s some screen shots.

I’ve opted to output straight DSD instead of DoP. I figure removing the extra processing step of unpacking the DSD from the PCM packet, albeit small, to help matters. Of course, I’m not sure if that relieves any processing in the DS Sr. I’m experimenting I may end up doing what you do anyway.
That explains the white light which is for native DSD, the blue is for DoP. Not sure why, but I couldn’t get native DSD to ever work out of HQP. As I recall, the DS converts DSD to DoP anyway, so there’s no escape…
OK, no joy there. Perhaps @tedsmith could clarify for us please?
@tedsmith has stated, to the best of my knowledge, that the DS/DSJ converts everything to DoP, though Ted is a rather enigmatic when it comes to quotations
Yes, he has said that
Indeed the DS (and DS Jr) convert any incoming native DSD to DoP. But since it’s a lossless conversion it’s nothing to be concerned about.
The thing is that the FPGA architecture in the DS is always doing everything. There is no concept of saving the DS any work. It’s always processing all inputs, converting the inputs to DoP, measuring their sample rates, etc. It’s always upsampling even if the input is DSD. It’s always deemphasizing CD’s even if playing DSD. The trick is that it only pays attention to the answers it cares about. If you are a programmer think of it doing an if-then-else by evaluating both the then clause and the else clause and then picking which to pay attention to with the if clause. There aren’t any instructions like a standard CPU, everything is a state machine: all possible new states are generated and then the next state’s outputs are selected.
The best way to make decisions about what processing to do externally is simply to try them and if you like them use them. If you like the DS’s version better use it. But don’t try to figure it out with assumptions about how the DS works and logic.
Thank you, Ted. Great description and explanation.
Probably not practical but this (Ted’s response) would be a great candidate for “pinning” here somewhere.
Would a PS Audio DirectStream DAC FAQ thread made sense? I am torn because it would be nice if only factual answers were placed in such a thread and wonder if an open DS FAQ thread could run the risk of disseminating incorrect information. On the other hand, these things tend to self-correct over time.
Just thinking out loud…
SEE
Certainly start such a thread if you would like.
The hard parts are going through the forum, harvesting the useful bits, and keeping the useless and off-topic posts out.
Given how this group likes to topically wander, my guess is such a thread would quickly self-destruct.
In 3…2…1.