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

[TopQuark]

1 hour ago, seeteeyou said:

Now I see what you're looking for, let's say we're already owners of the original sMS-200 and then we're paying for that 990 + 210 = $1,200 tX-USBultra. It's like killing two birds with one stone.

 

I asked this question before to SOtM.  What I got was that modding sMS-200 will not be as good as having the ultras because the external clock connection in the mod is not as good as the internal clock connection with the ultras.

[seeteeyou]

More good news regarding Allo DigiOne, a nice option for those of us who would like to avoid everything related to USB signal integrity etc.

 

http://forum-hifi.fr/thread-4520.html

Allo DigiOne board

https://community.roonlabs.com/t/allo-digione-board/18564

 

DigiOne Coaxial/BNC output for RPI

https://volumio.org/forum/digione-coaxial-bnc-output-for-rpi-t6219.html

 

HiFIBerry Digi+ Pro and Allo DigiOne Volumio special Version : Unofficial Portings

https://volumio.org/forum/hifiberry-digi-pro-and-allo-digione-volumio-special-version-t6849.html

 

10% off

https://www.allo.com/sparky/digione.html

 

 

Quote

10 LDOs, 12 LPF, DC/DC convertors, galvanic isolator, WM8805, flip flops, buffers, NDK oscillators.

 

Those clocks could be accessed easily and we should know what to do about them.

[seeteeyou]

We could find some router boards and they really don't cost that much to begin with

 

https://mqmaker.com/product/witi-board/

https://item.taobao.com/item.htm?id=521875114158

http://www.cnx-software.com/2015/08/12/witi-openwrt-router-board-features-6-ethernet-ports-dual-band-wifi-sata-and-usb-3-0-ports-crowdfunding/

http://www.cnx-software.com/2015/12/14/witi-router-board-powered-by-mediatek-mt7621a-soc-now-sells-for-79/

 

The performance just sucked pretty bad, though

 

https://www.reddit.com/r/linux_devices/comments/51g93o/qmaker_witi_board_with_2xsata_6xgigabit_ethernet/

 

Maybe it's a better bet to go for something like APU3 ($107 to $127) with x86 processors? As usual the clocks could be replaced by SOtM with sCLK-EX

 

https://www.pcengines.ch/apu3b2.htm

http://blog.3mdeb.com/2017/05/12/openwrt_on_apu3/

https://www.linuxserver.io/2016/12/17/review-pcengines-apu2-c4-prebuilt-by-linitx/

https://www.lteforum.at/mobilfunk/pc-engines-apu2-open-source-lte-router-project-mc7455.6852/

[String]

3 hours ago, String said:

to

Yes but it's beginning with the Router!

 So if it's not okay, well you have that problem with you all the way to your computer or other streaming devices!

 You now the chains is never stronger than the weakest! Or?

[seeteeyou]

https://wiki.mikrotik.com/wiki/Manual:RouterOS_FAQ

https://ark.intel.com/products/76308/Intel-Core-i5-4300U-Processor-3M-Cache-up-to-2_90-GHz

 

We could even install both RouterOS and Roon Server on Jetway NF542 since i5-4300U should be powerful enough

 

https://centralcomputers.com/p-554322-jetway-hbjc512f3dp-1900-b-fanless-embedded-4-x-10100-lan-with-poe-up-to-25w-per-port-2-x-intel-gbe-lproductproduct.aspx
https://centralcomputers.com/p-554321-jetway-hbjc38af542xa-4300u-b-fanless-embedded-with-intel-haswell-i5-class-vpro-soc-6-x-intel-gbe-lanproductproduct.aspx

http://www.jetwaycomputer.com/JBC38AF542AA.html

http://www.jetwaycomputer.com/NF542.html

 

And then it's also fanless to boot, expandability should be unmatched for a relatively small size @ 110mm x 167mm.

• 1 SATA3 6Gb/s Cable-Free Connector
• 1 Shared mSATA/mini-PCI Express Combo slot (full-size)
• 1 Mini-PCI Express slot (full-size)
• 1 PCI Express 2.0 x1 side-slots, Supports Jetway GEN3 Daughter Boards
• 1 PCI Express 2.0 x4 side-slots, Supports Jetway GEN3 Daughter Boards

The only question is whether those PCIe slots are connected to the processor directly or going through PCH.

[String]

Thanks ?.

A lot of alternative to choose between.

Do you have any knowledge about the company?

Who makes them?

[austinpop]

Just FYI to everyone - 

 

I've taken the first step to creating an index to this thread, which is to get edit access to the first thread. I've created a skeleton placeholder where the index will go. 

 

One unfortunate consequence of the way edit authority is granted with Chris's site software, it causes the OP name to change as well. Believe me, this is not what I wanted, as all credit is due to @romaz. So I've done the next best thing, which is to give him the credit in the heading of the original post. 

 

Take a look. I cannot provide an ETA on the index. Soon™.

 

And I still need volunteers for some sections.

[BigAlMc]

43 minutes ago, austinpop said:

I've taken the first step to creating an index to this thread,

 

Hi @austinpopmany thanks for doing this. This is one of the best threads out there but at 100+ pages it's very hard to find stuff. 

 

I think this and a few similar threads could do with some sort of "How To" guides. But an Index will definitely help.

 

I'm assuming that you are looking for more expert volunteers than myself but PM me if there's a relatively simple section I can help with.

 

Cheers,

Alan

[rickca]

52 minutes ago, austinpop said:

I've taken the first step to creating an index to this thread

Maybe we should fork this thread if we're going to delve further into subjects like master clock input and Paul Hynes power supplies so that index maintenance doesn't continue to get more challenging.

[austinpop]

1 hour ago, BigAlMc said:

I'm assuming that you are looking for more expert volunteers than myself but PM me if there's a relatively simple section I can help with.

Alan,

 

All help is welcome! Once I get started, I'll know better what I need. Stay tuned.

[tapatrick]

4 hours ago, austinpop said:

Just FYI to everyone - 

 

I've taken the first step to creating an index to this thread, which is to get edit access to the first thread. I've created a skeleton placeholder where the index will go. 

 

Thanks Austinpop for making this massive thread manageable. I will be keeping an eye on how this all unfolds.

[austinpop]

Attention, propeller-heads! Mad science experiment alert!

 

I am stoked. Barring any last-minute conflicts, @limniscate and I are getting together next week for an SOtM Ultra-vaganza! 

 

There are far more experiments and potentially interesting questions that we will be able address in a reasonable session time - although, we may decide on followup Ultra-vaganzae as warranted.

 

In the spirit of community, I wanted to open the floor and entertain any experiments you all might have high on your list, but we haven't covered below. I make absolutely no guarantees about whether we will be able to perform these experiments, due to time and other unforeseen constraints.

 

This is the aggregate BoM (bill of materials) we will have on hand:

• 1 x Zyxel switch (sCLK modded)
• 1 x D-Link switch (sCLK modded)
• 1 x sMS-200 (sCLK modded) 6-9V
• 1 x sMS-200ultra 12V
• 1 x tX-USBultra 6-9V
• 1 x dX-USBultra 12V
• 1 x Singxer SU-1 stock PSU
• 4 x LPS-1 supplies
• HDPlex 100W LPS, supplying 2 x 12V rails
• Assorted dCBL-CAT7 and iSO-CAT6 cables and isolators
• Hopefully, I'll be able to use my newfound DIY DC cable skills this weekend to whip up a Canare 4S6 serial cable, to connect 2 x LPS-1s in series to provide 12v.

Note that with this gear, we now have the ability to have up to 6, count 'em, 6 sCLK-EX reclockings in the chain. Do these many reclockings still provide an incremental boost in SQ? More on this below. The baseline chain we'll use is Eric's:

• Roon Core W10 bridged > D-Link modded switch > sMS-200ultra > (USB) > dX-USBultra > (AES) > Yggy DAC

Here are some burning questions we want to answer:

1. What is the improvement due to 2 x LPS-1 over HDPlex on both sMS-200ultra and dX-USBultra?
2. Is sMS-200ultra 12V (fed by 2 x LPS-1) SQ better than modded sMS-200 6-9V (fed by LPS-1 7V)?
3. Add a 5th reclocking in the chain:
   • D-Link modded switch > sMS-200ultra > tX-USBultra > dX-USBultra
4. Add a 6th reclocking in the chain:
   • Zyxel modded switch > D-Link modded switch > sMS-200ultra > tX-USBultra > dX-USBultra
5. Compare modded Zyxel and D-Link switches
6. dX via AES to yggy vs. tX via USB to Yggy
7. Singxer SU-1 vs dX-USBultra
8. Effect of dCBL-CAT7 and iSO-CAT6. Too many experiments here to enumerate, but this may have to wait for the Ultra-vaganza redux.

The floor is open.

[tapatrick]

3 minutes ago, austinpop said:

Attention, propeller-heads! Mad science experiment alert!

 

.......the floor is open.

WOW!

[tboooe]

Live stream!!!!!

[austinpop]

A couple clarifications.

 

7 minutes ago, austinpop said:

HDPlex 100W LPS, supplying 2 x 12V rails

Eventually, this will be replaced by the Paul Hynes SR7MR2DRXL that Eric has on order. That will be another Ultra-polooza for sure.

 

7 minutes ago, austinpop said:

Here are some burning questions we want to answer:

? Hmm, hope that wasn't a Freudian slip? ?

 

[AmusedToD]

32 minutes ago, austinpop said:

Attention, propeller-heads! Mad science experiment alert!

 

I am stoked. Barring any last-minute conflicts, @limniscate and I are getting together next week for an SOtM Ultra-vaganza! 

 

There are far more experiments and potentially interesting questions that we will be able address in a reasonable session time - although, we may decide on followup Ultra-vaganzae as warranted.

 

In the spirit of community, I wanted to open the floor and entertain any experiments you all might have high on your list, but we haven't covered below. I make absolutely no guarantees about whether we will be able to perform these experiments, due to time and other unforeseen constraints.

 

This is the aggregate BoM (bill of materials) we will have on hand:

• 1 x Zyxel switch (sCLK modded)
• 1 x D-Link switch (sCLK modded)
• 1 x sMS-200 (sCLK modded) 6-9V
• 1 x sMS-200ultra 12V
• 1 x tX-USBultra 6-9V
• 1 x dX-USBultra 12V
• 1 x Singxer SU-1 stock PSU
• 4 x LPS-1 supplies
• HDPlex 100W LPS, supplying 2 x 12V rails
• Assorted dCBL-CAT7 and iSO-CAT6 cables and isolators
• Hopefully, I'll be able to use my newfound DIY DC cable skills this weekend to whip up a Canare 4S6 serial cable, to connect 2 x LPS-1s in series to provide 12v.

Note that with this gear, we now have the ability to have up to 6, count 'em, 6 sCLK-EX reclockings in the chain. Do these many reclockings still provide an incremental boost in SQ? More on this below. The baseline chain we'll use is Eric's:

• Roon Core W10 bridged > D-Link modded switch > sMS-200ultra > (USB) > dX-USBultra > (AES) > Yggy DAC

Here are some burning questions we want to answer:

1. What is the improvement due to 2 x LPS-1 over HDPlex on both sMS-200ultra and dX-USBultra?
2. Is sMS-200ultra 12V (fed by 2 x LPS-1) SQ better than modded sMS-200 6-9V (fed by LPS-1 7V)?
3. Add a 5th reclocking in the chain:
   • D-Link modded switch > sMS-200ultra > tX-USBultra > dX-USBultra
4. Add a 6th reclocking in the chain:
   • Zyxel modded switch > D-Link modded switch > sMS-200ultra > tX-USBultra > dX-USBultra
5. Compare modded Zyxel and D-Link switches
6. dX via AES to yggy vs. tX via USB to Yggy
7. Singxer SU-1 vs dX-USBultra
8. Effect of dCBL-CAT7 and iSO-CAT6. Too many experiments here to enumerate, but this may have to wait for the Ultra-vaganza redux.

The floor is open.

 

Crazy stuff! This is like 5-6k worth of reclocking equipment 

[ElviaCaprice]

10 minutes ago, AmusedToD said:

 

Crazy stuff! This is like 5-6k worth of reclocking equipment 

Which only begs the question.  If reclocking is so important then why not reclock several times in the same component?  I suspect that its not so much about the actual clock itself (to a certain degree of quality, that being your DAC), but the quality of the power that clock receives.  Too bad you guys don't have an ISO Regen to throw into the mix.

[mozes]

1 hour ago, austinpop said:

Attention, propeller-heads! Mad science experiment alert!

 

I am stoked. Barring any last-minute conflicts, @limniscate and I are getting together next week for an SOtM Ultra-vaganza! 

 

There are far more experiments and potentially interesting questions that we will be able address in a reasonable session time - although, we may decide on followup Ultra-vaganzae as warranted.

 

In the spirit of community, I wanted to open the floor and entertain any experiments you all might have high on your list, but we haven't covered below. I make absolutely no guarantees about whether we will be able to perform these experiments, due to time and other unforeseen constraints.

 

This is the aggregate BoM (bill of materials) we will have on hand:

• 1 x Zyxel switch (sCLK modded)
• 1 x D-Link switch (sCLK modded)
• 1 x sMS-200 (sCLK modded) 6-9V
• 1 x sMS-200ultra 12V
• 1 x tX-USBultra 6-9V
• 1 x dX-USBultra 12V
• 1 x Singxer SU-1 stock PSU
• 4 x LPS-1 supplies
• HDPlex 100W LPS, supplying 2 x 12V rails
• Assorted dCBL-CAT7 and iSO-CAT6 cables and isolators
• Hopefully, I'll be able to use my newfound DIY DC cable skills this weekend to whip up a Canare 4S6 serial cable, to connect 2 x LPS-1s in series to provide 12v.

Note that with this gear, we now have the ability to have up to 6, count 'em, 6 sCLK-EX reclockings in the chain. Do these many reclockings still provide an incremental boost in SQ? More on this below. The baseline chain we'll use is Eric's:

• Roon Core W10 bridged > D-Link modded switch > sMS-200ultra > (USB) > dX-USBultra > (AES) > Yggy DAC

Here are some burning questions we want to answer:

1. What is the improvement due to 2 x LPS-1 over HDPlex on both sMS-200ultra and dX-USBultra?
2. Is sMS-200ultra 12V (fed by 2 x LPS-1) SQ better than modded sMS-200 6-9V (fed by LPS-1 7V)?
3. Add a 5th reclocking in the chain:
   • D-Link modded switch > sMS-200ultra > tX-USBultra > dX-USBultra
4. Add a 6th reclocking in the chain:
   • Zyxel modded switch > D-Link modded switch > sMS-200ultra > tX-USBultra > dX-USBultra
5. Compare modded Zyxel and D-Link switches
6. dX via AES to yggy vs. tX via USB to Yggy
7. Singxer SU-1 vs dX-USBultra
8. Effect of dCBL-CAT7 and iSO-CAT6. Too many experiments here to enumerate, but this may have to wait for the Ultra-vaganza redux.

The floor is open.

@austinpopJust to let you know, I will be doing these tests next week when my VR will have arrived:

1- tX-USBultra vs. Iso-Regen

2- Nimitra>IR>tX>Brooklyn vs. Nimitra>tX>IR>Brooklyn

I think this way we will almost have all the answers that we want so far.

[BigAlMc]

1 hour ago, austinpop said:

Attention, propeller-heads! Mad science experiment alert!

 

I am stoked. Barring any last-minute conflicts, @limniscate and I are getting together next week for an SOtM Ultra-vaganza! 

 

Fantastic! Can't wait for the results posts. Great work guys and loads of fun too I'm sure. 

[rickca]

Eric, Rajiv and Moussa I'm certainly looking forward to the results of your experiments.

 

I'm just wondering ... why does it make sense that multiple reclockings with the same quality clock is better than one?  I've read the input from @romaz that the modded switch significantly improves the sound quality vs using only the sMS-200ultra, but that switch mod also includes replacing the regulators and capacitors on the switch.

 

Wouldn't one top quality reclocking be better than multiple reclockings with something less good?

 

I'm not looking for a scientific explanation at this point, just something plausible.  Even without this, if it sounds better, it sounds better!  So carry on.

[austinpop]

1 hour ago, ElviaCaprice said:

Which only begs the question.  If reclocking is so important then why not reclock several times in the same component?  I suspect that its not so much about the actual clock itself (to a certain degree of quality, that being your DAC), but the quality of the power that clock receives.  Too bad you guys don't have an ISO Regen to throw into the mix.

 

Right now, we are going on purely empirical findings, motivated by @romaz's research. John Swenson has indicated this whole area is one he is actively researching, but is not ready to discuss his theories at this time. Over on the Mutec side, there are similar reports of people cascading MC-3+ USB reclockers to good effect.

 

We are at the dawn of understanding this stuff. Once harnessed, I am sure manufacturers will find clever ways to maximize performance - even for example, like you speculate, multiple times in a component.

 

Until then, I approach this with a spirit of fun, adventure, and the joy of learning.

 

It's a great time to be ...

[austinpop]

32 minutes ago, mozes said:

@austinpopJust to let you know, I will be doing these tests next week when my VR will have arrived:

1- tX-USBultra vs. Iso-Regen

2- Nimitra>IR>tX>Brooklyn vs. Nimitra>tX>IR>Brooklyn

I think this way we will almost have all the answers that we want so far.

 

Awesome, @mozes! Really looking forward to your findings.

 

I am also expecting @Bamber's kind loaner IR/LPS1 when it ships - hopefully in mid-July.

[austinpop]

24 minutes ago, BigAlMc said:

 

... loads of fun too ...

 

That is the operative phrase!

[austinpop]

15 minutes ago, rickca said:

Eric, Rajiv and Moussa I'm certainly looking forward to the results of your experiments.

 

I'm just wondering ... why does it make sense that multiple reclockings with the same quality clock is better than one? 

 

Wouldn't one top quality reclocking be better than multiple reclockings with something less good?

 

I'm not looking for a scientific explanation at this point, just something plausible.  Even without this, if it sounds better, it sounds better!  So carry on.

 

Some day soon, this will be better understood by the best minds in computer audio. Until then, we only have the empirical data to go on.

[austinpop]

Woot!

 

My DIY serial DC cable - while messy looking - works as designed. We are a GO for 2 x LPS-1s to power Eric's 12V Ultra gear.