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

[marce]

Some interesting cables from a recent email from the signal integrity journal, they have transmitted up to 112GBs in PAM4, covers all the usual impedances (the 85R twinax could be used for a USB Cable if thats your thing, only 5R off the standard). We are doing some high speed LVDS stuff and have to go off board so the 100R looks perfect for us, as we are feeding 24bit DAC's, but only at low MHzs. May be useful for those that want to tortue themselves and make their own clock cables.

The demo in  question.

https://blog.samtec.com/post/112-gbps-pam4-micro-twinax-cable-demonstration/

woops the cables:

http://suddendocs.samtec.com/productcharts/drcable_specs.pdf

[rickca]

1 hour ago, BigGuy said:

None of the links took me to sites in English.

Use  google translate.

[the_doc735]

1 hour ago, seeteeyou said:

 

Finding the right HDMI cable should be the key for an I2S connection, pretty much the same deal as clock cables because we've gotta get something that's short and sweet

 

https://www.amazon.com/StarTech-com-0-3m-Short-Speed-Cable/dp/B00K3HF276

https://www.amazon.com/Wireowrld-Island-Cable-Meter-Length/dp/B00C6IOQ7W

https://www.amazon.com/WIREWORLD-Chroma-Audio-18Gbps-Cable/dp/B00BUMO350

https://www.amazon.com/WIREWORLD-Ultraviolet-Audio-Video-Cable/dp/B00C6GTI72

https://www.amazon.com/Wireworld-Silver-Starlight-Cable-Meters/dp/B00C6FSQ5I

https://www.amazon.com/Wireworld-Platinum-Starlight-Cable-Meter/dp/B00C6FDG8U

 

Besides, let's hope that Gustard U16 is compatible with the I2S pinout of PS Audio

 

https://www.head-fi.org/threads/denafrips-terminator-the-king-of-r2r-dac.851085/page-4#post-14199895

https://docs.google.com/spreadsheets/d/1h5PMUBkldkpt1rCnAR4ZHYGZNeCe-vwIFyKWYMZWsX0

 

 

OMG, are you implying that you'd like to support something that's even more unthinkable than sweatshop operations?

 

I guess that 15 bucks could cover (part of) the tariffs, depending on what the percentage is gonna be.

I've got the QED HDMI 2.0 Reference cable.

[Confused]

The Gustard looks interesting.  From what anyone can glean, is there any reason to think it could match or beat established products like the MC3+USB?  A genuine question, personally I have no idea.  Also, is that a display on the front of the case, or just a label with Gustard on it, I’m not sure from the pics?

 

Its interesting to to see new kit with a 10 MHz input though, irrespective of the above.

[Confused]

3 hours ago, austinpop said:

BTW - is that ¥ referring to yuan (CNY) or yen (JPY)?

 

I'm assuming yuan as, if that were yen, the cost would be about $15!

It’s very cheap irrespective, £160?  You will of course need a Euro 800 Habst cable to make the thing sing.  Crazy?  Maybe, maybe not.

[austinpop]

The Gustard has I2S output which the MC-3+ USB does not.

 

What is not clear is whether the frequency synthesizer that is disciplined by the 10MHz reference clock also synthesizers the USB clock. The verbiage only talks about the sample rate clocks in the 44.1 and 48kHz families.

 

To be fair, I’m not sure the MC-3+ improves the USB clock with its reference clock input either.

 

This is my concern with all the DDCs in general.

[the_doc735]

4 hours ago, austinpop said:

I've been thinking about it more, and the other possibility is simply that the (very slight) difference I'm hearing between network playback on the SE through the Ethernet interface and local playback from the SSD is just inherent differences in the particular adapters chosen in the SE.

 

So I could just be chasing a phantom problem upstream.

 

Have others with the Zenith SE experimented and tested the SQ between local SSD and network streams? Your experiences, please!  

doesn't the Innuos ZENith SE have switching technology though? (on its mainboard):

 

it also has a pico and built in toroidal transformer & optical drive & standard SATA cables! Yet this is supposed to be high end gear? Wonder what else is not so special? Maybe you could put JCAT SATA cable in there? Use external LPSU? Does it have clocks that can be improved or are you using external clock anyway? Is that their own m/board design or is it an off the peg thin-mini ITX?  Have Fun!

[Confused]

46 minutes ago, austinpop said:

The Gustard has I2S output which the MC-3+ USB does not.

 

What is not clear is whether the frequency synthesizer that is disciplined by the 10MHz reference clock also synthesizers the USB clock. The verbiage only talks about the sample rate clocks in the 44.1 and 48kHz families.

 

To be fair, I’m not sure the MC-3+ improves the USB clock with its reference clock input either.

 

This is my concern with all the DDCs in general.

I’m struggling to rationalise that statement, if the DCC is not outputting USB, then it does not need a USB clock as such, but it does need to correct the clock signal that feeds it?  It’s been a long day here, I’ll think about this one again after a good nights sleep!?

[austinpop]

17 minutes ago, Confused said:

I’m struggling to rationalise that statement, if the DCC is not outputting USB, then it does not need a USB clock as such, but it does need to correct the clock signal that feeds it?  It’s been a long day here, I’ll think about this one again after a good nights sleep!?

 

Get some sleep!

 

It's not a question of need. None of the clock mods reported here lo these many months were needed. But think of it this way. In an sMS-200ultra, you have a clock for input USB, a clock for the system mobo, and a clock for the output ethernet. To improve SQ, all were upgraded with sCLK-EX clock points, then disciplined by an external reference clock. Not just the output clocks.

 

A DDC is a similar story. You have a clock for the input USB, and 2 clocks (one for each sample rate family) for the synchronous outputs (AES, S/PDIF, I2S). I'm not clear if there are any other clocks. If you look back on this thread, you'll see @elan120 modded his Singxer SU-1 to accept sCLK-EX inputs both to the sample rate clocks AND the USB clock. 

[austinpop]

Actually, this post has a really nifty diagram!

 

I added this to the index as well.

[shadowlight]

7 hours ago, shadowlight said:

Nope, but the router in question has triple band support 2.4 and 2 x 5GHz band (one of the Netgear Nighthawk with 6 antennas).  I setup one of the 5Ghz band dedicated for music with it's own SSID.  So I have three separate SSID running.  One dedicated for mobile phones and tablets, one for laptops and one for music.

 

Fios --> Fortinet Firewall --> HP Switch --> Wireless Router (I believe I have it setup in AP mode and not router mode, I will have to confirm later tonight).  HQPe server is connected to HP Switch which is the distribution switch for wired connection in family room, basement and study.  I also have a small switch in the music room connected to HP Switch for devices that do not support wireless connectivity.

Confirmed that NetGear Nighthawk X6 R8000 is running in Access Point mode.

[lmitche]

1 hour ago, the_doc735 said:

doesn't the Innuos ZENith SE have switching technology though? (on its mainboard):

 

it also has a pico and built in toroidal transformer & optical drive & standard SATA cables! Yet this is supposed to be high end gear? Wonder what else is not so special? Maybe you could put JCAT SATA cable in there? Use external LPSU? Does it have clocks that can be improved or are you using external clock anyway? Is that their own m/board design or is it an off the peg thin-mini ITX?  Have Fun!

Yes there is switching technology on the motherboard, but with use of the two power inputs from two of the three lpus the dependency on switched PSU sections is reduced to a level that greatly benefits SQ.

 

The same thing happens with mainstream gaming boards when two power supplies are used to power the cpu separately from the ATX connector.

[lmitche]

NUC update:

 

Today the usb 3.1 enclosure and associated hard disk was moved from direct connection to the NUC to the Netgear r7000. Now music is served wirelessly to an Asus WAP connected to the nuc via an USB 3 nic. This means there are only two devices with a ground connected to the NUC, the WAP and the power supply to the nuc. Tomorrow a WiFi antenna and 10 foot extension arrives. This will eliminate the WAP and it's power supply, making the NUC loopless, assuming that the moat created by the ISO Regen holds.

 

There was a nice increase in SQ created by moving the HDD to the router, with a noticable increase in timbre on classical tracks.

[seeteeyou]

For that Supermicro X10SBA-L inside Innuos ZENith SE, some switching regulators could still be bypassed according to Innuos' director Nuno Vitorino

 

https://www.computeraudiophile.com/forums/topic/30376-a-novel-way-to-massively-improve-the-sq-of-computer-audio-streaming/?page=313&tab=comments#comment-813712

On 4/26/2018 at 4:25 PM, nvitorino said:

If you connect the ATX supply, the board will not use it's own switching regulators to convert to 5V and 3.3V from 12V. The interesting part is that it still uses the 12V rail on both the ATX and the 12V input separately if the 12V input is available.

 

Basically we could order four pieces of LT3045 from Alexey ($150 plus shipping) since the current draw of X10SBA-L is up to 1.5A. Two with 12V output, one for 5V output, and one for 3.3V output respectively

 

http://www.ldovr.com/category-s/123.htm

On 4/26/2018 at 4:25 PM, nvitorino said:

The result is the same, adding both ATX and 12V results is clearly superior sound with this board. If you don't want / can't get separate rails for ATX and 12V, getting at least another 12V supply and connecting a Pico-PSU will still yield benefits compared to feeding the 12V alone. If you are using a separate power supply for the SSD, 1.5A on ATX and 1.5A on 12V is enough for this board.

 

Further details were already posted on this thread back in April 2018

 

https://www.computeraudiophile.com/forums/topic/30376-a-novel-way-to-massively-improve-the-sq-of-computer-audio-streaming/?page=304&tab=comments#comment-806606

 

https://www.computeraudiophile.com/forums/topic/30376-a-novel-way-to-massively-improve-the-sq-of-computer-audio-streaming/?page=312&tab=comments#comment-813489

 

Other discussions

 

https://www.computeraudiophile.com/forums/topic/46284-lpsu-choices/?tab=comments#comment-848591

http://jplay.eu/forum/index.php?/topic/1979-rebuild-and-improved-power-supply-for-a-singlepc/?p=33317

http://www.tirnahifi.org/forum/viewtopic.php?t=2832

 

We could also modify the motherboard to order to feed both RAM modules directly with 1.2V, one more switching regulator is out of the way

 

http://www.tirnahifi.org/forum/viewtopic.php?t=3267

 

Pearce shared his success story with LT3045 here

 

https://www.diyaudio.com/forums/vendor-s-bazaar/310402-fs-ultra-low-noise-power-supply-lt3045-based-pcb-6.html#post5353416

 

I posted a list of SO-DIMM DDR3L RAM that won't draw more than 1.5A

 

https://www.computeraudiophile.com/forums/topic/30376-a-novel-way-to-massively-improve-the-sq-of-computer-audio-streaming/?page=225&tab=comments#comment-762083

 

Here's a link to (possibly?) the only USB 3.0 to Gigabit Ethernet adapter with external DC power input just in case you're looking for something that's potentially better than those on-board Ethernet ports

 

https://www.amazon.com/IO-Crest-SY-HUB24047-Gigabit-Network/dp/B01N16C75R

 

USB to Gigabit Fiber Ethernet for 60 bucks, just get your own GBIC off eBay for cheap

 

https://www.dx.com/p/winyao-usb1000f-ethernet-lan-realtek-rtl8153-usb-3-0-gigabit-fiber-network-card-white-342903

https://www.dx.com/p/winyao-usb1000f-lx-usb-3-0-1000mbps-fiber-optical-network-card-w-sfp-optical-module-black-352476

 

Another brand from China that's bloody expensive, though you could pick your own USB 3.0 cable

 

https://item.taobao.com/item.htm?id=566593291519

https://item.taobao.com/item.htm?id=566621762771

 

How to properly install RTL8153 driver on Linux

https://www.pcsuggest.com/install-rtl8153-driver-linux/

http://www.realtek.com.tw/downloads/downloadsView.aspx?PFid=56&Level=5&Conn=4&DownTypeID=3

[mourip]

7 hours ago, seeteeyou said:

Here's a link to (possibly?) the only USB 3.0 to Gigabit Ethernet adapter with external DC power input just in case you're looking for something that's potentially better than those on-board Ethernet ports

 

https://www.amazon.com/IO-Crest-SY-HUB24047-Gigabit-Network/dp/B01N16C75R

 

 

Thanks. Good scouting. Unfortunately the IO-Crest has an AC power adapter so it probably has in internal SMPS. We would probably need to mod it in order to use an LPS.

 

Package Contents

 

USB 3.0 to 4 Port Gigabit Ethernet Network Adapter 1 x USB type A to Type B cable 1 x AC power adapter

[seeteeyou]

Nah, it's taking 5V/2A according to both official sites

 

http://www.iocrest.com/en/product_details571.html

http://www.mmui.com.cn/en/product_details942.html

Quote

1x DC 5V/2A Power adapter

 

Same deal when we're looking at their store in China

 

https://detail.tmall.com/item.htm?id=543750880680

 

https://img.alicdn.com/imgextra/i3/1045640990/TB2MFsbdNBmpuFjSZFDXXXD8pXa_!!1045640990.jpg

[mourip]

By AC power adapter they probably just means that it gets power from the wall. Thanks for the correction!

[Confused]

23 hours ago, austinpop said:

 

Get some sleep!

 

It's not a question of need. None of the clock mods reported here lo these many months were needed. But think of it this way. In an sMS-200ultra, you have a clock for input USB, a clock for the system mobo, and a clock for the output ethernet. To improve SQ, all were upgraded with sCLK-EX clock points, then disciplined by an external reference clock. Not just the output clocks.

 

A DDC is a similar story. You have a clock for the input USB, and 2 clocks (one for each sample rate family) for the synchronous outputs (AES, S/PDIF, I2S). I'm not clear if there are any other clocks. If you look back on this thread, you'll see @elan120 modded his Singxer SU-1 to accept sCLK-EX inputs both to the sample rate clocks AND the USB clock. 

Ok - that makes sense.  I’m thinking that in the MC3+USB the job of the USB clock is to govern the rate at which the USB data is received.  Over time, this data must be received at exactly the same rate as the output, otherwise the FIFO buffer would empty or overflow.  Considering this, I would assume the MC3+USB’s output and USB clock would both take reference from the reference clock.  I might be wrong, but I think this makes sense.  Subjectivity, the results I am getting with my Devialet amp, which likes an AES/EBU feed, are superb.  So whatever the MC3+USB is doing, I’m pretty happy!

 

If @julian.david happens to check in, maybe he can advise.

[barrows]

2 hours ago, Confused said:

Over time, this data must be received at exactly the same rate as the output, otherwise the FIFO buffer would empty or overflow.

No, this is not an issue with USB audio transmission.  There will never be empty or overflow conditions in the buffer because the rate of the data being sent from point to point is controlled via bi-directional communication, regardless of clock accuracy.

[lmitche]

16 minutes ago, barrows said:

No, this is not an issue with USB audio transmission.  There will never be empty or overflow conditions in the buffer because the rate of the data being sent from point to point is controlled via bi-directional communication, regardless of clock accuracy.

True, but over an awful half duplex physical layer. Where are the USB 3 DACs anyway?

[lmitche]

NUC update:

 

I agree with most of you in believing that the quality of upstream components matters. Nevertheless I have always been troubled by the excellent sound quality of Tidal from Roon.  I can't hear a difference, mastering quality aside, from locally sourced music from highly optimized equipment and Tidal source files. Given this, it shouldn't mattter if local files are served from the dirty side of the LAN, just like Tidal.

 

Yesterday the USB 3 disk containing my music files was moved to the untreated power side of the room. The music disk is served as a share from a SMPS powered Netgear R7000 router. Today the USB NIC and LPS powered WAP were removed and the AC class wireless card in the NUC enabled at 5ghz. This means that the NUC has three cables, a wifi antenna, 19 volts from a linear power supply and the USPCB to the ISO Regen.  Given the moat created by the ISO Regen it is now ground loopless. I also attached a 10 foot antenna cable from the nuc wifi card and placed the antenna inches from the router hosting the music disk.

 

So the music is now being served over the NUC 433mps wifi link from the router(as NAS) on the dirty side.

 

SQ is fabulous, once again more transparent, with a beautiful timbre from instruments. Harshness is gone. I had to lower volume almost 4db on many tracks. My tinnitus is much reduced.

 

Roon core on Audiolinux is running on the NUC.

 

Tomorrow a second NUC arrives. It cost $125. Keeping it in the dirty zone, I'll move the hard disk there, enable Roon core there, and run Roonbridge on the clean, ramboot, Dac connected NUC.

 

It was a fascinating, and surprisingly rewarding day. I feel like I'm almost done with this silly hobby.

 

P.S. I know, I know, I've recreated the ethernet to USB streamer here, just like the microRendu or SOTM SMS ultra . . .  nevertheless I've auditioned both and had nothing like the SQ heard here today.  I don't know where the magic is, Audiolinux perhaps?

 

 

 

 

[the_doc735]

On 4/26/2018 at 9:25 AM, nvitorino said:

 

Hi everyone,

 

I can provide a bit of feedback on this based on our experience with the Supermicro X10SBA-L. On this board there is a significant difference in sound quality when powering both the ATX and the 12V Input compared to 12V alone. If you connect the ATX supply, the board will not use it's own switching regulators to convert to 5V and 3.3V from 12V. The interesting part is that it still uses the 12V rail on both the ATX and the 12V input separately if the 12V input is available. We've tested 3 configurations:

 

1. ATX only (using a Pico-PSU with a clean 12V input)

2. 12V only

3. ATX + 12V

 

1. and 2. sound similar in our tests. 3 sounds significantly better. At first, we were expecting 2. to be the best, as it removed the Pico-PSU from the equation. But seems the board itself will do the same job as the Pico and doesn't seem to do it any better than that.

We've fthen further tested with separate rails using the ATX only and compared to using separate rails for ATX + 12V input. The result is the same, adding both ATX and 12V results is clearly superior sound with this board. If you don't want / can't get separate rails for ATX and 12V, getting at least another 12V supply and connecting a Pico-PSU will still yield benefits compared to feeding the 12V alone. If you are using a separate power supply for the SSD, 1.5A on ATX and 1.5A on 12V is enough for this board.

 

Nuno
 

can someone please show pictures with text to better explain the 3 configurations on the board, together with their PSU's? Thanks!

[seeteeyou]

3 hours ago, lmitche said:

True, but over an awful half duplex physical layer. Where are the USB 3 DACs anyway?

 

There are quite a few DACs plus another DDC that are designed for audiophiles

 

https://ifi-audio.com/products/pro-idsd/

https://ifi-audio.com/products/nano-ione/

https://ifi-audio.com/products/micro-idac2/

http://matrix-digi.com/en/products/158/index.html

 

The others are meant for the pro audio market

 

http://motu.com/products/avb/8a

http://motu.com/products/avb/624

https://tascam.com/us/product/us-20x20/top

https://www.presonus.com/products/Studio-192

https://www.presonus.com/products/Studio-192-Mobile

http://www.rme-audio.de/en/products/madiface_xt.php

http://www.rme-audio.de/en/products/digiface_avb.php

http://www.rme-audio.de/en/products/digiface_dante.php

http://www.rme-audio.de/en/products/fireface_ufx-plus.php

https://www.zoom-na.com/products/production-recording/audio-interfaces/zoom-uac-2-usb-30-audio-converter

https://www.zoom-na.com/products/production-recording/audio-interfaces/zoom-uac-8-usb-30-audio-converter

 

BTW, I've been reading quite a bit about how these audiophiles were improving SQ with something called U3TT

 

http://www.my-hiend.com/psycho/U3TT.jpg

http://www.my-hiend.com/vbb/showthread.php?5404-走向超值而極緻的-HI-FI電腦訊源（連載）&p=228538#post228538

https://www.ptt.cc/bbs/Headphone/M.1511162787.A.786.html

 

Basically they're cutting off specific pins for USB 2.0 in order to maintain that "USB 3.0 only" mode

 

 

Since most USB DACs are only good for 2.0 speed, they're adding this Adnaco-SU1 adapter for that final conversion from 3.0 to 2.0 accordingly

 

https://www.amazon.cn/dp/B00K5NADIQ

http://www.adnaco.com/doc/Adnaco-SU1-DS.pdf

https://www.adnaco.biz/blogs/adnaco-for-medical/13654565-adnaco-su1-usb-3-0-fiber-optic-extension-system

 

Lindy jacked up the price after it's discontinued by Adnaco

 

https://www.amazon.com/USB-2-0-Adapter-3-0-Cables/dp/B014S7HVI4

https://www.lindy-international.com/USB-2-0-Adapter-for-USB-3-0-AOC-Cables.htm?websale8=ld0101.ld020102&pi=42682

 

It's still available from China and Taiwan at the moment for just under $60

 

https://item.taobao.com/item.htm?id=535913193820

https://goods.ruten.com.tw/item/show?21524750112022

http://www.everprotech.com/UploadFiles/Files//admin/201503/U3D02AR User Manual.pdf

 

"USB 3.0 only" cables

https://goods.ruten.com.tw/item/show?21627020094867

https://goods.ruten.com.tw/item/show?21627020099660

https://goods.ruten.com.tw/item/show?21627020106445

 

"USB 3.0 only" adapters

https://goods.ruten.com.tw/item/show?21534462330340

https://goods.ruten.com.tw/item/show?21534462335648

https://goods.ruten.com.tw/item/show?21742234983117

https://goods.ruten.com.tw/item/show?21742234985186

 

Obviously Ghent could read Chinese and therefore most likely he'll make something like that.

[austinpop]

6 hours ago, Confused said:

Ok - that makes sense.  I’m thinking that in the MC3+USB the job of the USB clock is to govern the rate at which the USB data is received.  Over time, this data must be received at exactly the same rate as the output, otherwise the FIFO buffer would empty or overflow.  Considering this, I would assume the MC3+USB’s output and USB clock would both take reference from the reference clock.  I might be wrong, but I think this makes sense.  Subjectivity, the results I am getting with my Devialet amp, which likes an AES/EBU feed, are superb.  So whatever the MC3+USB is doing, I’m pretty happy!

 

If @julian.david happens to check in, maybe he can advise.

 

Yes, it would be good to confirm whether the MC-3+ USB:

1. has the same quality (i.e. phase noise) clock for USB as it has for the sample rate clock(s), and
2. if the USB clock is also disciplined by the reference clock input?

I wouldn't be surprised if the answer is no. The idea of improving "system" clocks (USB, Ethernet, mobo) is really something that gained traction with this thread (courtesy Romaz), although, to be fair, there were vendors dabbling in it - like Paul Pang, SOtM, SGM, among others.

 

After the discussion on the Chinese Gustard DDC that CTU posted, it got me thinking - imagine how great a DDC the Mutec MC-3+ USB would be if:

1. it had an input for an external PSU, so powered with an SR-4 or SR-7
2. it had an I2S output
3. it's USB clock was of the same quality as the sample rate clocks, and could be disciplined by the input reference clock.

Something like that from Mutec would be pretty special.

[the_doc735]

56 minutes ago, austinpop said:

 

Yes, it would be good to confirm whether the MC-3+ USB:

1. has the same quality (i.e. phase noise) clock for USB as it has for the sample rate clock(s), and
2. if the USB clock is also disciplined by the reference clock input?

I wouldn't be surprised if the answer is no. The idea of improving "system" clocks (USB, Ethernet, mobo) is really something that gained traction with this thread (courtesy Romaz), although, to be fair, there were vendors dabbling in it - like Paul Pang, SOtM, SGM, among others.

 

After the discussion on the Chinese Gustard DDC that CTU posted, it got me thinking - imagine how great a DDC the Mutec MC-3+ USB would be if:

1. it had an input for an external PSU, so powered with an SR-4 or SR-7
2. it had an I2S output
3. it's USB clock was of the same quality as the sample rate clocks, and could be disciplined by the input reference clock.

Something like that from Mutec would be pretty special.

have you  got any links for improved mobo clocks please? I thought that had been abandoned because it creates too many problems i.e. paul pang?