A novel way to massively improve the SQ of computer audio streaming

[mikicasellas]

16 hours ago, bobfa said:

A Tale of Three Roon Servers

 

 

To haul this thread a bit more back on track, here is a little story about SQ in my systems.

 

 

I have been using Allo USBridge Sig endpoints to a couple of different music server software systems.  I have tried UPnP, Audirvana, and Roon.  I have narrowed down the software in the Allo to either MoOde or GentooPlayer.  While both sound good, MoOde does not seem to work right with Roon!  GentooPlayer has held up well for me.  The Allo USBridge Sig is the best sounding endpoint I have used to-date in my systems.   

 

While the Allo does an excellent job of isolating the noise from the DAC, it is not perfect.   

 

Two NUCs running ROCK and a Xeon running Euphony

I wanted to find out if the server mattered any more!  I started with two different 7i7 NUC devices: The first one is in the Intel tall chassis with fan and the external switching supply.  The second one is my fan-less build with an HDPLEX 200 power supply.  I installed Roon ROCK on both systems so I could swap back and forth.  There are other differences in the hardware, RAM, and SSD vs. Optane, so the differences are exaggerated.

 

On both Allo endpoints, the difference was instantly observable.  The Intel chassis model sounded harsh, and the dynamics are constrained.  It felt like I stepped back ten years, yuk!  I hope my HD-800 cans were not insulted!

 

The next test is to bring in my Xeon server with an HDPLEX 400 LPS.  I added back in the Sonore Opticalmodule, bridged the network to the Allo on my Kii Three speaker system.   The music feels excellent; there is additional detail, bass clarity, and midrange strength with none of the high-end glare! 

 

Now the fan-less NUC running Roon ROCK is pretty darn good, but the Xeon box is way better.  And not a Paul Hines SR power supply in sight!

 

There is so much more to research.    I can take out the Opticalmodule and use a generic FMC or wired ethernet.  So much more with software.  Different kernels and settings.  I am not even loading endpoints or server OS into RAM yet.  

 

What is incredible is that after several months of reviewing and testing, I got my Roon back!  For right now, this is an excellent place to pause and Enjoy the Music.

Bob

 

Reference Threads:

 

https://audiophilestyle.com/forums/topic/55383-from-0-to-nucalram-in-2-hours/?tab=comments#comment-917302

 

https://audiophilestyle.com/forums/topic/55681-from-0-to-xeonalroon-server-in-2-days/?tab=comments#comment-928961

 

https://audiophilestyle.com/ca/reviews/allo-usbridge-sig-review-r858/

 

https://audiophilestyle.com/forums/topic/55235-gentooplayer/#comments

 

 

What Power Supply are you using for the Allo USBridge Sig ?

 

 

[bobfa]

2 minutes ago, mikicasellas said:

 

What Power Supply are you using for the Allo USBridge Sig ?

 

 

I have the Shanti supplies on both of them.  I have not experimented with others.  Too many items in the matrix to test.  I also wanted the Allo testing to be on a common ground and price scale....

 

 

[ted_b]

1 hour ago, bobfa said:

When I say endpoint I mean. The renderer, D to D. Typically Ethernet to USB or other DAC input. 
 

 

Sorry if I didn't explain my point.  I realize that an endpoint can be a renderer  But it can also be a rather simplified FIFO buffer, as in Jussi's NAA architecture.  I am asking whether you (or Romaz, in this example) believe these different functions would have different cpu, power, noise profiles....or is low-power-low-noise no longer the perfect endpoint box in ANY circumstance?

[bobfa]

1 minute ago, ted_b said:

Sorry if I didn't explain my point.  I realize that an endpoint can be a renderer  But it can also be a rather simplified FIFO buffer, as in Jussi's NAA architecture.  I am asking whether you (or Romaz, in this example) believe these different functions would have different cpu, power, noise profiles....or is low-power-low-noise no longer the perfect endpoint box in ANY circumstance?

I have no idea.  I think the modalities here are not well understood, at least by me!

 

What I have found is that every software package and OS makes changes to how things sound!  So the Roon Bridge sounds different that Squeezelite.  

 

Bob

[matthias]

18 minutes ago, ted_b said:

....or is low-power-low-noise no longer the perfect endpoint box in ANY circumstance?

 

Maybe you find the answer here:

 

Matt

[cat6man]

20 minutes ago, ted_b said:

Sorry if I didn't explain my point.  I realize that an endpoint can be a renderer  But it can also be a rather simplified FIFO buffer, as in Jussi's NAA architecture.  I am asking whether you (or Romaz, in this example) believe these different functions would have different cpu, power, noise profiles....or is low-power-low-noise no longer the perfect endpoint box in ANY circumstance?

 

great question.

running hqplayer and having an NAA architecture at the endpoint, does increased processing power still sound better?

 

[cat6man]

3 minutes ago, matthias said:

 

Maybe you find the answer here:

 

Matt

 

if i recall, romaz was not using hqplayer/NAA so i'm not sure his conclusions re: processing power map into that case...........they might, but i don't see evidence yet.

[Miska]

The original thought of NAA is to run it on a low power Intel or ARM device where 5V USB bus power feed also comes straight from the 5V DC PSU feed the board uses. Thus the 5V USB bus power doesn't go through DC-DC conversion on the computer board. And the CPU has low operation current. Thus most Intel based boards I'm using for the purpose are using Atom SoC.

 

[matthias]

23 minutes ago, cat6man said:

 

if i recall, romaz was not using hqplayer/NAA so i'm not sure his conclusions re: processing power map into that case...........they might, but i don't see evidence yet.

Agree,

no evidence related to HQP/NAA.

 

Matt

[seeteeyou]

I dunno man, pairing USBridge Sig with this upcoming DC4 from Dr. Sean Jacobs could be such an interesting combo by saving money on the endpoint while getting one of the best PSUs out there

 

http://www.custom-hifi-cables.co.uk/home/power-supplies/dc4-power-supply

Quote

The DC4 uses a much larger case than the DC3, and can be fitted with 1, 2 or 3 outputs. If you require more than 3 outputs, then you will need to order multiple DC4 units. In some configurations, it may be possible to build hybrid DC3/DC4 PSUs to suit specific cost/performance requirements.

 

That might / might not be able to meet + beat single-regulated SR7 but we'll see, of course the $64,000 question is how double-regulated SR7 rail(s) would stack up.

[elan120]

1 hour ago, ted_b said:

Sorry if I didn't explain my point.  I realize that an endpoint can be a renderer  But it can also be a rather simplified FIFO buffer, as in Jussi's NAA architecture.  I am asking whether you (or Romaz, in this example) believe these different functions would have different cpu, power, noise profiles....or is low-power-low-noise no longer the perfect endpoint box in ANY circumstance?

 

1 hour ago, cat6man said:

 

great question.

running hqplayer and having an NAA architecture at the endpoint, does increased processing power still sound better?

 

I recently done some comparisons between two NUC's with results might be helpful in this question.

 

I ran an Intel NUC7PJYH in an AKASA enclosure as endpoint when I first switched from single PC to a two PC setup late summer last year, it was using AL as OS and NAA streaming ROON/HQPd from my server.  It was working nicely for a while until I changed endpoint to an Intel NUC7i7DNHE couple month later, also in an AKASA fanless case running AL/NAA.  I stayed with this setup since that time, but continue to tweak this NUC such as clocks, internal DC cables, RAMs, external power supplies...etc., it continue to improve the SQ and recently, I built a new server running AL/HQPe that has a sizable step up in SQ, I got curious how much impact endpoint can still contribute with this new server, so I put NUC7PJYH back in using the same AL/NAA setup as well as the same power supply and RAM, the result still clearly show NUC7i7DNHE is a step up in SQ.

 

Although there are other things I did on NUC7i7DNHE that can't be transferred to NUC7PJYH, but one obvious difference is the CPU power.  There are clearly more things would be nice to find out such as will even higher power CPU be helpful in raising the SQ?  What is the best balance?  I am not sure the answers to these, but in my system, there is no one "perfect" endpoint setup.

[rickca]

2 hours ago, seeteeyou said:

this upcoming DC4 from Dr. Sean Jacobs

Can someone please explain why this unit would have a 400VA transformer if it can only do 5A continuous current?  Even at 25V, isn't that way overspec?  Clearly I don't understand enough about power supply design. 

[sandyk]

1 hour ago, rickca said:

Can someone please explain why this unit would have a 400VA transformer if it can only do 5A continuous current?  Even at 25V, isn't that way overspec?  Clearly I don't understand enough about power supply design. 

" The DC4 uses a much larger case than the DC3, and can be fitted with 1, 2 or 3 outputs"

 I suspect that this refers to a 3 separate outputs version.

When using a typical Full Wave Bridge , Capacitor input load for example , the DC current out is only about 62% of the
Secondary A.C. current . So in your single output example, you would need to use something larger than a 160VA transformer, with the next in the range typically 300VA.

Perhaps Sean has standardised on a 400VA transformer to his own specifications ?

[Miska]

12 hours ago, elan120 said:

I got curious how much impact endpoint can still contribute with this new server, so I put NUC7PJYH back in using the same AL/NAA setup as well as the same power supply and RAM, the result still clearly show NUC7i7DNHE is a step up in SQ.

 

Have you tried going with something low power? Like the UpBoard or RasPi4 running from 5 V LPS?

 

Preferably with the bootable NAA image so that you get same OS consistently, even on different hardware.

 

[elan120]

3 hours ago, Miska said:

 

Have you tried going with something low power? Like the UpBoard or RasPi4 running from 5 V LPS?

 

Preferably with the bootable NAA image so that you get same OS consistently, even on different hardware.

 

I ran RPi3 with LPS-1 for a while before moving to single PC setup and now back to two PC.  Never tried UpBoard or RPi4, but I might revisit and try using the RPi3 I already have at some point down the road to see what happen now with the new PC.

 

When I did the comparison earlier, it was using the same OS with same configuration, so the consistency should be present.

[Miska]

7 hours ago, elan120 said:

I ran RPi3 with LPS-1 for a while before moving to single PC setup and now back to two PC.  Never tried UpBoard or RPi4, but I might revisit and try using the RPi3 I already have at some point down the road to see what happen now with the new PC.

 

RPi3, as the earlier models too have the problem of overloaded USB bus, so you get clicks. So it is only good with cards that hook to it's I2S, like HifiBerry, etc.

 

My issue with NUCs are their built-in PSUs which I don't like...

 

7 hours ago, elan120 said:

When I did the comparison earlier, it was using the same OS with same configuration, so the consistency should be present.

 

Same OS and same configuration on RPi and PC?

 

[elan120]

12 minutes ago, Miska said:

Same OS and same configuration on RPi and PC?

Not this time, only compared the two NUCs mentioned, which were using the same OS and configuration.

 

18 minutes ago, Miska said:

My issue with NUCs are their built-in PSUs which I don't like.

What built-in PSUs are these?  Are you referring to the included AC adapters like below picture?  If yes, I changed them to a 19V LPS. 

 

[Miska]

3 minutes ago, elan120 said:

What built-in PSUs are these?  Are you referring to the included AC adapters like below picture?  If yes, I changed them to a 19V LPS.

 

DC-DC converters / switching regulators to convert the variable input voltage (they accept fairly wide range) to the various needed voltages, including USB bus power.

 

5 minutes ago, elan120 said:

Not this time, only compared the two NUCs mentioned, which were using the same OS and configuration.

 

OK, I think it is just important factor when making comparisons.

 

[iCarus]

On 12/20/2019 at 1:39 PM, seeteeyou said:

 

Quite expensive for its specs, though it's a tried and true motherboard that's also found inside Innuos ZENith SE

 

https://www.amazon.com/Supermicro-X10SBA-L-B-Celeron-Mini-ITX-Motherboard/dp/B00F0YJC28

Many thanks for the suggestion. In the end I decided to go for the Intel DN2800MT. Many of them were available as used computers on eBay for less than 50GBP.  And they're used in the following thread:

 

https://www.stereo.net.au/forums/topic/82802-building-the-idealish-music-server/

[Confused]

On 1/8/2020 at 9:24 PM, cat6man said:

 

great question.

running hqplayer and having an NAA architecture at the endpoint, does increased processing power still sound better?

 

Late last year, I needed to replace my old PC.  So I swapped my early HP i5 machine for a new Dell.  The new machine is an Intel i9 9900k, has liquid cooled PSU and processor, lots of fast memory, and other goodies.  The old HP was an "all in one" design", so little more than laptop spec components with a nice screen.  For a while I had both machines on the same network, and I could perform a direct comparison running HQPlayer 4 with identical settings on both machines.

 

So I loaded a short playlist of identical music files into HQPlayer on both machines to try an A/B test.  This was not an ideal test, when streaming the PC needs to have the web based Eunhasu software running and "talking" the sMS-200ultra Neo.  When swapping from one PC to the other, it was necessary to fully shut down one PC (or pull out the ethernet cable) to get Eunhasu on the other machine working.  So not an ideal A/B test, with much faffing around when swapping from PC to PC.  After much swapping from one machine to the other, I ultimately came to the conclusion that everything sounded pretty much the same, irrespective of which PC was delivering the files.  Not the most exciting of conclusions I know, but perhaps interesting to note that in my subjective experience, making significant hardware changes upstream of the sMS-200ultra appears to make little difference.

Moving out of the subjective domain, with the old HP "all in one", I always found Roon to be a little clunky in operation, just a little slow and cumbersome.  With the i9 9900k, Roon is super slick, so that's at least something!

[guiltyboxswapper]

@Confused can you describe your network setup including which brand of switches etc used?  Lately I've discovered switching to Startech FMCs for both ends made quite a noticeable difference, significant infact.  I'm also a HQPlayer + NAA user.

[Confused]

3 hours ago, guiltyboxswapper said:

@Confused can you describe your network setup including which brand of switches etc used?  Lately I've discovered switching to Startech FMCs for both ends made quite a noticeable difference, significant infact.  I'm also a HQPlayer + NAA user.

At the time I performed the test per post #15520 I was running hard wired Ethernet, PC to Router, hard wired Ethernet to a SOtM modified D-Link switch, then hardwired Ethernet to the sMS-200Ultra.  The SOtM modified Ethernet switch takes its clock signal from the clock board in the sMS-200Ultra, itself fed from the Mutec REF10.

 

I am now running the EtherRegen in preference to my old SOtM switch.

 

I have never experimented with FMC's or optical, I am tempted to give it a try though.

[seeteeyou]

FYI - here's a comparison among Cybershaft Limited2 OP20 / Mutec REF10 SE-120 / Cybershaft OP21A

 

https://www.whatsbestforum.com/threads/reality-is-cruel-cybershaft-new-ultimate-ocxo-10m-clocks-shootout-op20-vs-op17.26610/page-9#post-621445

[mourip]

I would like to try some different memory in my Roon server. It is using an ASRock H87M-ITX board which uses DDR3 1600 un-buffered non-ECC UDIMM memory. I have read through as many posts here as possible and scoured the web for sources and information but have come up dry. It seems like folks like the Apacer high temp DDR4 memory but I have not been able to find a comparable DDR3 UDIMM.

 

So two questions. In searching for better sounding memory what parameters should I look for? The two that I have seen discussed are temperature and that 1R8 chip placement might be a positive factor. Any thing else?

 

The second question is: does anyone have a good source for DDR3 memory that would meet audio needs. I was able to get some Apacer DDR4 SODIMMs from a fellow ASer and this worked well in my NUC but I would still like to try some in my server and my other endpoint, both of which run the same ASRock system board.

 

Thanks for any help or suggestions!

[seeteeyou]

https://www.soselectronic.com/articles/apacer/the-most-reliable-memory-for-industries-2177

Quote

Series 75 uses chips rated to temperatures from -40 to +95°C, 1% resistors, 125°C rated capacitors and 30um gold fingers plating.

 

75.B93ET.G000C
https://ipc2u.com/catalog/75_b93et_g000c

 

75.B93ET.G010C
https://ipc2u.com/catalog/75_b93et_g010c

 

Not that many choices for ASRock H87M-ITX are available to begin with, just send them an e-mail and find out what the lead time is.