Jump to content
IGNORED

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


Message added by The Computer Audiophile

Important and useful information about this thread

Posting guidelines

History and index of useful posts

Most important: please realize this thread is about bleeding edge experimentation and discovery. No one has The Answer™. If you are not into tweaking, just know that you can have a musically satisfying system without doing any of the nutty things we do here.

Recommended Posts

Hello Hols,

 

That was a while back. I had forgotten about the two old SR9 power supplies I custom designed and built for the Trinity and 10M. I just recently allotted the designation SR9 to a design under discussion with Roy (Romaz), without realising I had already used it once before for your custom power supplies. Woops!

 

For clarity, these two SR9 power supplies were totally different to the current SR5 and SR7 type power supplies I currently provide. They used one of my older shunt regulator topologies with a back biasing current source that reduced the standing current after the 10M start-up surge. I only made two of these SR9 and they are in no way comparable to the SR5 and SR7 power supplies or my current shunt regulator circuit topologies regarding performance.

 

The following is a general statement regarding equipment comparisons and not aimed at you, Hols. For the record, over the years, I have noticed references to power supply comparisons using my power supplies. They usually state “the Paul Hynes Power supply”. Well which one? I have a very wide range of power supply options, so a vague comparison like “the Paul Hynes Power supply” is absolutely meaningless as there is no specific power supply reference. Also, many of my power supplies have had multiple performance improvements as time goes by, so older Paul Hynes power supply versions compared to current competing power supplies are once again rather meaningless. The marketing boys often use this vague statement presentation to give the impression their products are better than their competitors. If you are going to do a comparison with my power supplies (or other manufacturers products whether or not they are power supply related) make it current and specific like Roy has recently done in this thread.

 

Regards

Paul

Paul Hynes Design

Design and manufacture of high performance power supplies

Link to comment
1 hour ago, hols said:

I think my experience here is just another testimony to Roy's very accurate predication of the effect of the SOtM clock as well as the importance of giving it a really good power supply. Actually I have gone through a path quite similar to Roy's(Of course I don't have his knowledge and instincts in picking the most significant piece) and I have tried the various power supplies he mentioned, Teradak, Kenneth Lau, JS-2, Paul Pang power supplies(years ago) and Paul Hynes SR9(tailored made for my Antelope Trinity and 10M ) and I agree that the Ultracap is probably the best of them all if not limited by its small amperage. And here I think it is a must have for the SOtM USBUltra. 

 

Just one more comment for those familiar with the Holo Spring and Singxer SU-1 I2S output. It is often said that the I2S input to Holo Spring is superior to USB input because the onboard clock(used by USB but not by I2S) is only mediocre. Now with this SOtM tx-USBUltra in place there is not much difference in SQ between the USB input and I2S input in my system. This is probably just another illustration to the importance of putting a good clock close to the DAC.

 

 

1 hour ago, tboooe said:

Ted, can you pls explain why you think doing so would make no sense? I intend to put the IsoRegen in front of my SU-1.

 

I'm hoping you SU-1 owners can help me.  Background: I'm currently using a server with bridged NICs to an sMS-200 (also have a microRendu) then to my DAC via USB.  The DAC also has both I2S and AES/EBU inputs.  The I2S happens to be a 5 pin DIN.

 

I'm not looking for my server to be in a different location than the DAC so using the sMS or mR as an endpoint isn't necessary.  I'm using them for perceived sound improvement.

 

My questions are, would I be better off using the SU-1 in place of the endpoint and come out of the server with USB to the SU-1? Then could the SU-1 go to the DAC via I2S or AES/EBU?

 

Paul is building me an SR7 to power my server and endpoint.  I'm glad he just posted here because his input on powering the SU-1 if I go that route will be helpful.

Link to comment
1 hour ago, ted_b said:

Great post and feedback info, hols. 

 

To your last comment, I guess this says good things about the SU-1 as it seems to be holding its own against a 3x priced "competitor".  I assume putting the Ultra on the USB path the to Singxer would make no sense, but have you tried it anyway?

Hi Ted,

 

I did try using the Ultra on the path before it goes to the Singxer the improvement is not great. Previosuly I2S has a very accurate feel and dynamics is much better compared to USB alone. Now the end result of USB using Ultra direct to Holo is similar to I2S with Ultra added to USB path before going to Singxer and then Holo.  Soundstage, dynamics, musicality all comparable.

 My interpretation is that the improvement of Ultra is so big that the previous big advantage of the I2S route now appears not significant. On the one hand I agree with you that the Singxer did a good job with 1/3 of the price yet viewed from another angle it seems that there is no more need to further explore the I2S route?

Further answers to other questions is my Singxer is not modified with LPS . I have an Uptone regen too and when compared to Ultra seems more like a 'toy'(don't mean to deprecate its value). Even after adding the Intona the difference is still huge. I have ordered Iso regen too. 

 

And answer to Elviacaprice's question is mine is iPower 9v2A as marked clearly on the power supply not 1.5A. But I don't know why it actually cannot charge the Ultracap fast enough.

Link to comment
1 minute ago, Johnseye said:

 

My questions are, would I be better off using the SU-1 in place of the endpoint and come out of the server with USB to the SU-1? Then could the SU-1 go to the DAC via I2S or AES/EBU?

 

GREAT question here!  I have not yet tried removing my NAA from my system.  Recently I tried using a microrendu as my NAA before the SU-1 and it improved the sound quality so I think even with the Su-1 having an NAA or something like an ISOREGEN before the SU-1 could be beneficial.  When I get my ISOREGEN I will try removing my NAA and see if I can tell the difference.

12TB NAS >> i7-6700 Server/Control PC >> i3-5015u NAA >> Singxer SU-1 DDC (modded) >> Holo Spring L3 DAC >> Accustic Arts Power 1 int amp >> Sonus Faber Guaneri Evolution speakers + REL T/5i sub (x2)

 

Other components:

UpTone Audio LPS1.2/IsoRegen, Fiber Switch and FMC, Windows Server 2016 OS, Audiophile Optimizer 3.0, Fidelizer Pro 6, HQ Player, Roonserver, PS Audio P3 AC regenerator, HDPlex 400W ATX & 200W Linear PSU, Light Harmonic Lightspeed Split USB cable, Synergistic Research Tungsten AC power cords, Tara Labs The One speaker cables, Tara Labs The Two Extended with HFX Station IC, Oyaide R1 outlets, Stillpoints Ultra Mini footers, Hi-Fi Tuning fuses, Vicoustic/RealTraps/GIK room treatments

Link to comment

Thanks Johnseye and tboooe for the interesting discussion.

 

I'm trying to decide whether to just get a dX-USB Ultra to replace my Singxer SU-1 or to mod my SU-1 and get a sMS-200 Ultra or just get a tX-USB Ultra in front of my Singxer SU-1 with my microrendu.

Link to comment
21 minutes ago, hols said:

Hi Ted,

 

I did try using the Ultra on the path before it goes to the Singxer the improvement is not great. Previosuly I2S has a very accurate feel and dynamics is much better compared to USB alone. Now the end result of USB using Ultra direct to Holo is similar to I2S with Ultra added to USB path before going to Singxer and then Holo.  Soundstage, dynamics, musicality all comparable.

 My interpretation is that the improvement of Ultra is so big that the previous big advantage of the I2S route now appears not significant. On the one hand I agree with you that the Singxer did a good job with 1/3 of the price yet viewed from another angle it seems that there is no more need to further explore the I2S route?

Further answers to other questions is my Singxer is not modified with LPS . I have an Uptone regen too and when compared to Ultra seems more like a 'toy'(don't mean to deprecate its value). Even after adding the Intona the difference is still huge. I have ordered Iso regen too. 

 

And answer to Elviacaprice's question is mine is iPower 9v2A as marked clearly on the power supply not 1.5A. But I don't know why it actually cannot charge the Ultracap fast enough.

@Hols, I also had similar results in  my system. The tX-USBultra replaced both the Intona and RUR and I never looked back for a second. My tX is the 12V version.

Link to comment
21 minutes ago, hols said:

Hi Ted,

 

I did try using the Ultra on the path before it goes to the Singxer the improvement is not great. Previosuly I2S has a very accurate feel and dynamics is much better compared to USB alone. Now the end result of USB using Ultra direct to Holo is similar to I2S with Ultra added to USB path before going to Singxer and then Holo.  Soundstage, dynamics, musicality all comparable.

 

 

This is surprising.  I would have thought that the additional clocking before the SU-1 by the TX Ultra would have been a boost to SQ.  I wonder if the need for the SU-1 USB input of 5V is taxing your TX Ultra too much with only the LPS-1 at 7V?  Thus the TX Ultra is not as good as with the DAC directly via USB and no 5Vbus required.  (Is this correct for the Holo?)

(JRiver) Jetway barebones NUC (mod 3 sCLK-EX, Cybershaft OP 14)  (PH SR7) => mini pcie adapter to PCIe 1X => tXUSBexp PCIe card (mod sCLK-EX) (PH SR7) => (USPCB) Chord DAVE => Omega Super 8XRS/REL t5i  (All powered thru Topaz Isolation Transformer)

Link to comment
5 minutes ago, limniscate said:

Thanks Johnseye and tboooe for the interesting discussion.

 

I'm trying to decide whether to just got a dX-USB Ultra to replace my Singxer SU-1 or to mod my SU-1 and get a sMS-200 Ultra or just get a tX-USB Ultra in front of my Singxer SU-1 with my microrendu.

The permutations make my brain hurt!  At least for me, the decision to ditch my SU-1 would depend on how good the USB input is in my DAC.  In my case, the clock in the SU-1 is better than the clock in the Spring DAC USB input.  In my opinion, no matter how good the clock or input is in front the DAC, if its USB input is not very good you will never get the best sound.  

 

Are you using the SU-1 I2S output?  What DAC do you have?

12TB NAS >> i7-6700 Server/Control PC >> i3-5015u NAA >> Singxer SU-1 DDC (modded) >> Holo Spring L3 DAC >> Accustic Arts Power 1 int amp >> Sonus Faber Guaneri Evolution speakers + REL T/5i sub (x2)

 

Other components:

UpTone Audio LPS1.2/IsoRegen, Fiber Switch and FMC, Windows Server 2016 OS, Audiophile Optimizer 3.0, Fidelizer Pro 6, HQ Player, Roonserver, PS Audio P3 AC regenerator, HDPlex 400W ATX & 200W Linear PSU, Light Harmonic Lightspeed Split USB cable, Synergistic Research Tungsten AC power cords, Tara Labs The One speaker cables, Tara Labs The Two Extended with HFX Station IC, Oyaide R1 outlets, Stillpoints Ultra Mini footers, Hi-Fi Tuning fuses, Vicoustic/RealTraps/GIK room treatments

Link to comment
2 minutes ago, tboooe said:

The permutations make my brain hurt!  At least for me, the decision to ditch my SU-1 would depend on how good the USB input is in my DAC.  In my case, the clock in the SU-1 is better than the clock in the Spring DAC USB input.  In my opinion, no matter how good the clock or input is in front the DAC, if its USB input is not very good you will never get the best sound.  

 

Are you using the SU-1 I2S output?  What DAC do you have?

I'm using USB input on the SU-1 from a microRendu via a Curious USB cable and AES output on the SU-1 to Schiit Yggdrasil.

Link to comment
34 minutes ago, limniscate said:

I'm using USB input on the SU-1 from a microRendu via a Curious USB cable and AES output on the SU-1 to Schiit Yggdrasil.

 

This is one possibility for me. Do you hear an improvement with the SU-1 in the chain vs. mR direct to your Yggdrasil? Any concerns with the AES? 

Link to comment
16 minutes ago, Johnseye said:

 

This is one possibility for me. Do you hear an improvement with the SU-1 in the chain vs. mR direct to your Yggdrasil? Any concerns with the AES? 

I haven't tested just the SU-1 in the chain versus the microRendu and SU-1.  I think the AES sounds the same as the microrendu to Yggdrasil direct with Curious USB cable.  However the SU-1 sounded better than the microrendu to Yggdrasil direct with Oyaide Neo d+ USB cable.

Link to comment
15 hours ago, hols said:

I just reveal my experience here for your reference. 

 

 

Thank you for waking up the thread and for sharing your experience, hols!  The perspectives that you and @mozes have provided are very helpful.  

 

With the upcoming ISO Regen as well as with your experience with the tX-USB Ultra, it may turn out that a superior USB endpoint may be just as effective as an Ethernet endpoint (NAA).  This is a question I would like to answer for myself.  

 

With analog, it is generally accepted that the fewer components in the chain, the greater the transparency.  With digital, it could possibly go both ways.  My theory remains (based on my observations) that the more damage caused by bad clocks and noisy components upstream, the more reparative devices you need (like the tX-USB Ultra) downstream to restore signal integrity.  As I have added a reclocking switch, an sMS-200 Ultra and a dX-USB HD Ultra (4 clocks total) in series within this "direct connection," as long as each successive clock is at least equal to or better than the previous clock and as long as each device is powered to an equivalent high standard with a low noise and low impedance PSU, as many of you know, I have found that SQ improves and the cumulative improvement can be dramatic.  

 

While there is more to SQ than just the clock and quality power, the "before" and "after" where only the clock has been replaced has been very surprising to me.  I have never heard this kind of a delta before based on a clock change alone and so it suggests that SOtM's new clock is either really that good or that certain components, depending on their position within the chain, are simply more capable of manifesting the benefits of a good clock.  

 

Here's a different example of the impact of clock placement.  As many know, the sMS-200 contains 2 clocks -- one clock for the Ethernet input and another clock for USB.  Based on the experience of another CA member who replaced the clocks on his standard sMS-200 one at a time, he found that replacing the USB clock made a significantly larger difference than replacing the Ethernet clock.  This would make sense as I would expect the clock for the output stage to always have the greater impact.  It also makes sense from the standpoint that your clock at the output stage will always be that much closer to your DAC (or to the next component in the chain).

 

This brings up another point.  Since I have found that multiple reclockings can result in additive improvements, does the sMS-200 Ultra (with its 2 clocks) sound better than the tX-USB Ultra (with its one clock)?  

 

To add to the mystery, here's one more observation.  Over the weekend, a good friend lent me a pair of his SOtM dCBL-CAT7 ethernet cables ($500 each).  According to SOtM, combining a pair of these cables along with their iSO-CAT6 LAN isolator ($350) can lead to dramatic improvements.  As I already own (and have currently been using) SOtM's dCBL-CAT6 combined with their iSO-CAT 6 LAN isolator in my direct connection (which has resulted in decent gains in SQ over my BJC CAT6A by itself), I thought there would be no way their CAT7 Ethernet cable that incorporates a "filter" would do that much more.  Boy, was I wrong.  I elected to place SOtM's recommended "ultimate" combination of dCBL-CAT7 + iSO-CAT6 LAN isolator + dCBL-CAT7 between my Mac Mini and my Trend Net reclocking switch within the "direct connection" pathway and compared this against my dCBL-CAT6 by itself, a Supra CAT8 cable, and a BJC CAT6A cable and the improvement was nothing short of stunning in the form of a larger and more holographic soundstage and improved dynamics.  It's as if those other cables were suffocating my music.  The midrange sounded a bit more recessed than I would prefer but overall, the SQ improvement was tremendous and of similar impact as the reclocking switch itself!  In fact, it literally doubled the impact of the switch and nearly matched the improvement of the sMS-200 Ultra in terms of "air" and dynamics.  

 

Was it because my previous short run of Ethernet cable (1m in length) was causing that much harm or were these cables infusing their own proprietary goodness?  Because Ethernet assures bit-perfect delivery through its error correcting protocol, I have always assumed that the best an Ethernet cable can do is filter noise even though Ethernet is already partially galvanically isolated but is there really that much noise within this "direct connection" pathway that is already devoid of the direct impact of the router?  Of course, this led me to wonder if these filters were just removing RF or doing more.  I decided to directly compare this combination against my FMCs again but I'm not sure this answered my question.  Certainly, fiber should block RF at least as well (if not better) than these cables but just like before and similar to the experience of others, even with my receiving FMC powered by an LPS-1, the FMCs by themselves in the direct pathway didn't add much that was good and compared against the pair of dCBL-CAT7s + iSO CAT6 LAN isolator, it was no contest.  With the FMCs, the soundstage sounded flat and there was definitely a slight edge or grain that was undesirable.  

 

I opened up one of my FMCs (as I have done many times in the past) and I believe the problem lies within the FMC circuitry which contains a bad clock and numerous switching regulators among other components that have their own sonic signature, I'm sure:

 

0502172302a_HDR_resized.thumb.jpg.0ccba7a434e3e83d053b2b195f2b2e85.jpg

 

I then moved this combination of dCBL-CAT7 / iSO-CAT6 / dCBL-CAT 7 out of the direct connection pathway and between my router and music server.  It definitely brought about some of the same qualities but only a very small fraction compared to what I had heard when this combination was in the direct connection pathway, certainly not enough to want to pay $1,350.

 

This has led me to conclude that Ethernet may not necessarily be better than other pathways such as USB as it seems just as susceptible to cables, at least within this highly revealing direct connection pathway.  As the weeks go by, it will be interesting to hear from others who buy the tX-USB Ultra or ISO Regen regarding whether they believe either of these USB endpoints by themselves can better their NAA counterparts like the mR or sMS-200 (Ultra or otherwise).  

Link to comment
15 hours ago, paulhynes said:

Regards

Paul

Paul Hynes Design

 

 

Welcome, Paul.  Thank you for your input.  My own perspective is that the pair of LPS-1s that I own are incredible and compared against most of my other power supplies, are in a class of their own with respect to not just dynamics but also this natural ease and flow that my music takes on.   I believe I am being consistent, however, when I say that my SR7 takes all the qualities of my LPS-1s even further.  Not to take anything away from my LPS-1s but I have yet to experience anything that does what my SR7 does.  

Link to comment
15 hours ago, hols said:

I did try using the Ultra on the path before it goes to the Singxer the improvement is not great.

1

 

I wonder if this is because the clock in the Singxer may not be as good as the clock in the Ultra and that perhaps the Singxer's clock is negating or at least limiting the impact of the Ultra?  This has been my experience any time I have placed a "better clock" before a "merely good clock." Of course, improving the Singxer's PSU would undoubtedly improve things also.  You should try moving your LPS-1 to the Singxer and using the stock PSU on the Ultra and see what happens.

Link to comment

sorry, but it's a bit hard to follow - even though it's really fascinating to read - for a profane like me.

Are we saying now that the ethernet route with the new sMS 200 ultra and dX-USB HD Ultra (or Mutec MC3+ USB for that matter) could be actually inferior than the direct USB route with tX-USB ultra and FMC? or did I misunderstand Roy's post above?

Link to comment

Well, could a single music server with Adnaco-S3B (i.e. occupying PCI-E slot without going through PCH) do the trick by any chance? The clocks and switching regulators on both ends could be replaced accordingly

 

http://www.adnaco.com/doc/Adnaco-S3B-DS.pdf
https://www.adnaco.biz/collections/s3b/products/s3b-01-000-p-e

 

And then we could power Adnaco-R1USB30 with LPS-1 easily (~1.5W to 1.75W) so we're good to go

 

http://www.adnaco.com/doc/Adnaco-R1USB30-DS.pdf

 

Still "Direct USB" in a somewhat indirect way?

Link to comment
39 minutes ago, pam1975 said:

sorry, but it's a bit hard to follow - even though it's really fascinating to read - for a profane like me.

Are we saying now that the ethernet route with the new sMS 200 ultra and dX-USB HD Ultra (or Mutec MC3+ USB for that matter) could be actually inferior than the direct USB route with tX-USB ultra and FMC? or did I misunderstand Roy's post above?

 

Sorry to add confusion to all of this.  My point with hols finding and my recent observations with different Ethernet cables is that it appears a USB endpoint could possibly sound just as good as an Ethernet endpoint as both pathways seem equally susceptible to the same issues.  Personally, it is disheartening to me that a filtering Ethernet cable like the dCBL-CAT7 can make such a large difference suggesting there is plenty of noise in an Ethernet connection.  There are various threads here on CA and on Head-Fi suggesting that AOIP (Audio over IP) or the Ethernet is a superior transmission pathway for audio but now I'm not so sure that it is and when an endpoint is implemented well, it doesn't seem to matter whether it uses pure USB or Ethernet.

 

To be clear, however, I haven't yet heard the tX-USB Ultra (although I have one on order) or the ISO Regen (I missed out on the first run although another CA member who made it on the first run has agreed to loan me their unit) but thus far, my current setup which consists of a Mac Mini "directly connected" to my sMS-200 Ultra and with an intervening reclocking switch is the best setup I have ever heard, bar none (even if I remove the dX-USB HD Ultra from the chain).  I certainly don't want to discourage anyone from going down this Ethernet path.

Link to comment

Bravo, Roy, for another great finding!

 

7 hours ago, romaz said:

 

With analog, it is generally accepted that the fewer components in the chain, the greater the transparency.  With digital, it could possibly go both ways.  My theory remains (based on my observations) that the more damage caused by bad clocks and noisy components upstream, the more reparative devices you need (like the tX-USB Ultra) downstream to restore signal integrity.  As I have added a reclocking switch, an sMS-200 Ultra and a dX-USB HD Ultra (4 clocks total) in series within this "direct connection," as long as each successive clock is at least equal to or better than the previous clock and as long as each device is powered to an equivalent high standard with a low noise and low impedance PSU, as many of you know, I have found that SQ improves and the cumulative improvement can be dramatic.  

 

Serendipitously, I just got my FedEx notification of the arrival of my own Ultra chain: 

  1. tX-USBultra
  2. sMS-200 with sCLK-EX mod
  3. Zyxel switch with sCLK-EX mod

It'll take me a few days to get my head around all this, but once I've verified everything is working, I will be back shortly with a post that describes my planned experiments.

 

7 hours ago, romaz said:

This brings up another point.  Since I have found that multiple reclockings can result in additive improvements, does the sMS-200 Ultra (with its 2 clocks) sound better than the tX-USB Ultra (with its one clock)?  

 

This is one of my planned experiments.

 

7 hours ago, romaz said:

To add to the mystery, here's one more observation.  Over the weekend, a good friend lent me a pair of his SOtM dCBL-CAT7 ethernet cables ($500 each).  According to SOtM, combining a pair of these cables along with their iSO-CAT6 LAN isolator ($350) can lead to dramatic improvements.  As I already own (and have currently been using) SOtM's dCBL-CAT6 combined with their iSO-CAT 6 LAN isolator in my direct connection (which has resulted in decent gains in SQ over my BJC CAT6A by itself), I thought there would be no way their CAT7 Ethernet cable that incorporates a "filter" would do that much more.  Boy, was I wrong.  I elected to place SOtM's recommended "ultimate" combination of dCBL-CAT7 + iSO-CAT6 LAN isolator + dCBL-CAT7 between my Mac Mini and my Trend Net reclocking switch within the "direct connection" pathway and compared this against my dCBL-CAT6 by itself, a Supra CAT8 cable, and a BJC CAT6A cable and the improvement was nothing short of stunning in the form of a larger and more holographic soundstage and improved dynamics.  It's as if those other cables were suffocating my music.  The midrange sounded a bit more recessed than I would prefer but overall, the SQ improvement was tremendous and of similar impact as the reclocking switch itself!  In fact, it literally doubled the impact of the switch and nearly matched the improvement of the sMS-200 Ultra in terms of "air" and dynamics.  

 

Wow, this blows my mind. Ethernet is clearly still a frontier to be discovered and conquered.

 

One question: did you find the symmetry important? I think you compared:

  • dCBL-CAT6 > iSO-CAT 6 > dCBL-CAT6

with

  • dCBL-CAT7 > iSO-CAT 6 > dCBL-CAT7, correct?

Did you try any of these combos:

  • dCBL-CAT6 > iSO-CAT 6 > dCBL-CAT7, or
  • BJC CAT6a > iSO-CAT 6 > dCBL-CAT7?

Just curious if the principle of "nearest to the audio end" concept matter, and if there is a lower cost option to springing for 2 of these expensive cables.

 

7 hours ago, romaz said:

This has led me to conclude that Ethernet may not necessarily be better than other pathways such as USB as it seems just as susceptible to cables, at least within this highly revealing direct connection pathway.  As the weeks go by, it will be interesting to hear from others who buy the tX-USB Ultra or ISO Regen regarding whether they believe either of these USB endpoints by themselves can better their NAA counterparts like the mR or sMS-200 (Ultra or otherwise).  

 

Indeed it will. I've always felt these direct PC vs. Ethernet "wars" were silly, because so much of it is tied with optimizing cost, architectural preference, and most importantly where you're coming from. If you've already invested heavily in a PC/Mac based music player system, then you will (and should) make different choices than someone like me who already has an investment in endpoints, and none in the PC side.

Link to comment
32 minutes ago, romaz said:

 

Sorry to add confusion to all of this.  My point with hols finding and my recent observations with different Ethernet cables is that it appears a USB endpoint could possibly sound just as good as an Ethernet endpoint as both pathways seem equally susceptible to the same issues.  Personally, it is disheartening to me that a filtering Ethernet cable like the dCBL-CAT7 can make such a large difference suggesting there is plenty of noise in an Ethernet connection.  There are various threads here on CA and on Head-Fi suggesting that AOIP (Audio over IP) or the Ethernet is a superior transmission pathway for audio but now I'm not so sure that it is and when an endpoint is implemented well, it doesn't seem to matter whether it uses pure USB or Ethernet.

 

To be clear, however, I haven't yet heard the tX-USB Ultra (although I have one on order) or the ISO Regen (I missed out on the first run although another CA member who made it on the first run has agreed to loan me their unit) but thus far, my current setup which consists of a Mac Mini "directly connected" to my sMS-200 Ultra and with an intervening reclocking switch is the best setup I have ever heard, bar none (even if I remove the dX-USB HD Ultra from the chain).  I certainly don't want to discourage anyone from going down this Ethernet path.

that was incredibly clear, thanks so much Roy :) 

Link to comment
8 hours ago, romaz said:

 

Thank you for waking up the thread and for sharing your experience, hols!  The perspectives that you and @mozes have provided are very helpful.  

 

With the upcoming ISO Regen as well as with your experience with the tX-USB Ultra, it may turn out that a superior USB endpoint may be just as effective as an Ethernet endpoint (NAA).  This is a question I would like to answer for myself.  

 

With analog, it is generally accepted that the fewer components in the chain, the greater the transparency.  With digital, it could possibly go both ways.  My theory remains (based on my observations) that the more damage caused by bad clocks and noisy components upstream, the more reparative devices you need (like the tX-USB Ultra) downstream to restore signal integrity.  As I have added a reclocking switch, an sMS-200 Ultra and a dX-USB HD Ultra (4 clocks total) in series within this "direct connection," as long as each successive clock is at least equal to or better than the previous clock and as long as each device is powered to an equivalent high standard with a low noise and low impedance PSU, as many of you know, I have found that SQ improves and the cumulative improvement can be dramatic.  

 

While there is more to SQ than just the clock and quality power, the "before" and "after" where only the clock has been replaced has been very surprising to me.  I have never heard this kind of a delta before based on a clock change alone and so it suggests that SOtM's new clock is either really that good or that certain components, depending on their position within the chain, are simply more capable of manifesting the benefits of a good clock.  

 

Here's a different example of the impact of clock placement.  As many know, the sMS-200 contains 2 clocks -- one clock for the Ethernet input and another clock for USB.  Based on the experience of another CA member who replaced the clocks on his standard sMS-200 one at a time, he found that replacing the USB clock made a significantly larger difference than replacing the Ethernet clock.  This would make sense as I would expect the clock for the output stage to always have the greater impact.  It also makes sense from the standpoint that your clock at the output stage will always be that much closer to your DAC (or to the next component in the chain).

 

This brings up another point.  Since I have found that multiple reclockings can result in additive improvements, does the sMS-200 Ultra (with its 2 clocks) sound better than the tX-USB Ultra (with its one clock)?  

 

To add to the mystery, here's one more observation.  Over the weekend, a good friend lent me a pair of his SOtM dCBL-CAT7 ethernet cables ($500 each).  According to SOtM, combining a pair of these cables along with their iSO-CAT6 LAN isolator ($350) can lead to dramatic improvements.  As I already own (and have currently been using) SOtM's dCBL-CAT6 combined with their iSO-CAT 6 LAN isolator in my direct connection (which has resulted in decent gains in SQ over my BJC CAT6A by itself), I thought there would be no way their CAT7 Ethernet cable that incorporates a "filter" would do that much more.  Boy, was I wrong.  I elected to place SOtM's recommended "ultimate" combination of dCBL-CAT7 + iSO-CAT6 LAN isolator + dCBL-CAT7 between my Mac Mini and my Trend Net reclocking switch within the "direct connection" pathway and compared this against my dCBL-CAT6 by itself, a Supra CAT8 cable, and a BJC CAT6A cable and the improvement was nothing short of stunning in the form of a larger and more holographic soundstage and improved dynamics.  It's as if those other cables were suffocating my music.  The midrange sounded a bit more recessed than I would prefer but overall, the SQ improvement was tremendous and of similar impact as the reclocking switch itself!  In fact, it literally doubled the impact of the switch and nearly matched the improvement of the sMS-200 Ultra in terms of "air" and dynamics.  

 

Was it because my previous short run of Ethernet cable (1m in length) was causing that much harm or were these cables infusing their own proprietary goodness?  Because Ethernet assures bit-perfect delivery through its error correcting protocol, I have always assumed that the best an Ethernet cable can do is filter noise even though Ethernet is already partially galvanically isolated but is there really that much noise within this "direct connection" pathway that is already devoid of the direct impact of the router?  Of course, this led me to wonder if these filters were just removing RF or doing more.  I decided to directly compare this combination against my FMCs again but I'm not sure this answered my question.  Certainly, fiber should block RF at least as well (if not better) than these cables but just like before and similar to the experience of others, even with my receiving FMC powered by an LPS-1, the FMCs by themselves in the direct pathway didn't add much that was good and compared against the pair of dCBL-CAT7s + iSO CAT6 LAN isolator, it was no contest.  With the FMCs, the soundstage sounded flat and there was definitely a slight edge or grain that was undesirable.  

 

I opened up one of my FMCs (as I have done many times in the past) and I believe the problem lies within the FMC circuitry which contains a bad clock and numerous switching regulators among other components that have their own sonic signature, I'm sure:

 

0502172302a_HDR_resized.thumb.jpg.0ccba7a434e3e83d053b2b195f2b2e85.jpg

 

I then moved this combination of dCBL-CAT7 / iSO-CAT6 / dCBL-CAT 7 out of the direct connection pathway and between my router and music server.  It definitely brought about some of the same qualities but only a very small fraction compared to what I had heard when this combination was in the direct connection pathway, certainly not enough to want to pay $1,350.

 

This has led me to conclude that Ethernet may not necessarily be better than other pathways such as USB as it seems just as susceptible to cables, at least within this highly revealing direct connection pathway.  As the weeks go by, it will be interesting to hear from others who buy the tX-USB Ultra or ISO Regen regarding whether they believe either of these USB endpoints by themselves can better their NAA counterparts like the mR or sMS-200 (Ultra or otherwise).  

 

Isn't the sMS-200 Ultra that superior endpoint?  An NAA with best of class DDC USB clock?  I just don't follow why a DDC like the dx-USB HD Ultra would need to follow the sMS-200 Ultra.  Unless the clocks are better in the dx-USB HD, then why wouldn't the best clocks available be used in the sMS-200 Ultra?

 

If you take this process from a minimalist perspective, if you are powering your PC/server (all clocks replaced with the best available) with the SR7 then wouldn't a direct connection to the DAC be the cleanest, simplest solution?  What is being gained with each device added? Another power filter or another clock.  Let's assume that the noise generated by the server, despite the low power consumption and SR7, is enough to require a "filter stage", then that next stage should have the cleanest power (SR7 rail) and best USB clock available.  Why would it matter if it's an NAA endpoint or DDC?  This is assuming your server is in close proximity to your DAC and that separation benefit isn't a consideration.

 

Why would multiple reclockings matter unless one clock is better than the other?  If multiple reclockings have additive improvements when all clocks and other components are the same how many clocks does it take before diminishing returns negate the benefit.

 

With regards to ethernet cables, this is a huge can of worms.  There is an ongoing debate in the Suggestion: Ban all cable debates thread (don't go there if you can help it) which gets into the question of validity. A lot of the discussion is around scientific method and ABX testing which is why I hesitate in using subjective terms to describe sonic improvements.  It's a reasonable and valid consideration.  Fleshing out the difference between what is perceived subjectively and specific to ones own equipment, room, etc vs objectively or as objectively as humanly possible is a substantial effort.  So let's take all that out of the equation and just trust the reviewer's perception for what it is.  Not you specifically, but anyone posting their opinion.  Data packets are what they are and ethernet is bit perfect so we can eliminate any possibility of sound modification due to packet manipulation.  This means we are only managing the noise coming over the ethernet cable.  Many factors come into play when considering the noise.  The iSO-CAT6 LAN isolator as a filter, the cable's category, jacket, twist, shielding, termination, grounding, cable material such as copper vs silver should all have some kind of impact.  The level of each material's impact and its effect on your system are variable.

 

You bring up the point of FMCs and I believe your conclusion is correct.  The FMCs themselves, while stopping the noise path over ethernet copper or silver also have their own issues.  Whether that be noisy power, clocks, regulators or whatever, if you've improved your upstream to where the FMCs produce a negative effect in your sound then I would consider that a success.  Of course you could always search for a better FMC, replace the clocks if possible or improve it in some other way where it may have an impact, but its function is a noise filter and nothing more.

 

My goal is to reproduce the sound of the music as close to its intended recording as possible, without adding coloration.  It is very hard to avoid coloration or certain sound signatures and that often times is exactly what we're doing; mixing and matching sound signatures to what sounds best to us.  Eliminating the noise and distortion helps to achieve that neutrality as best as possible, as well as allowing for more of the recording to be heard through lowering of the floor.  I am very curious to try different ethernet solutions, after I get my SR7, based on your findings and opinions.  I'm still after the cleanest simplest solution possible as I believe in our efforts to improve our chain through lowering noise and distortion we may be only changing colors or signatures.

 

 

8 hours ago, romaz said:

 

I wonder if this is because the clock in the Singxer may not be as good as the clock in the Ultra and that perhaps the Singxer's clock is negating or at least limiting the impact of the Ultra?  This has been my experience any time I have placed a "better clock" before a "merely good clock." Of course, improving the Singxer's PSU would undoubtedly improve things also.  You should try moving your LPS-1 to the Singxer and using the stock PSU on the Ultra and see what happens.

 

What intrigues me about the Singxer is its I2S interface.  Similar to AES/EBU but without the limitations.  Provided your DAC has an I2S interface this could be the best option.  My DAC has I2S but it's a 5 pin DIN termination while the Singxer is an HDMI termination.  The electrical standard must be LVDS and I need to determine if my DAC's I2S is the same.  You wouldn't happen to know someone talented enough to build a 5 pin DIN to HDMI I2S cable would you?

 

7 hours ago, ElviaCaprice said:

What?  You have a budget?  :P

I agree, I have always thought that the Ethernet route was a lost detour.  Take USB head on.  Interesting times.  Thanks again for your thoughts.  I wish sotm would have designed the tX-USB Ultra more input/output friendly.  Maybe the TX-Pcie card is the way to go with the S-clk ex clocking board separate?  Waiting to hear about your motherboard clocking findings.  I have a feeling that this will be the tipping point to favour USB direct.

 

Yes, if it had an I2S out I'd be a happy camper.  Does this bring things back full circle to USB direct out of the server?  I am also interested in the mobo clock findings.  Provided that all the updated clocks and power from the SR7 can eliminate the need for an endpoint or DDC this would be a worth while effort.  I expect the noise generated by the server itself regardless to be enough to warrant some kind of separation or filtering.  However, even if there were substantial gains by implementing high quality clocks on the mobo it is something I'd consider.

Link to comment
3 hours ago, seeteeyou said:

Well, could a single music server with Adnaco-S3B (i.e. occupying PCI-E slot without going through PCH) do the trick by any chance? The clocks and switching regulators on both ends could be replaced accordingly

 

http://www.adnaco.com/doc/Adnaco-S3B-DS.pdf
https://www.adnaco.biz/collections/s3b/products/s3b-01-000-p-e

 

And then we could power Adnaco-R1USB30 with LPS-1 easily (~1.5W to 1.75W) so we're good to go

 

http://www.adnaco.com/doc/Adnaco-R1USB30-DS.pdf

 

Still "Direct USB" in a somewhat indirect way?

 

Based on @lmitche's strong endorsement of the Adnaco S3B a while back, along with another friend, I have indeed been exploring this unit.  Obviously, what is intriguing about the Adnaco solution is it utilizes fiber and based on theory, there should be no better way to provide complete galvanic isolation since fiber carries zero DC.  Despite quite a bit of reading, it remains unclear to me how fiber transmission at 101MHz impacts signal integrity, especially from the standpoint of jitter but based on the rave reviews by some including Larry, I felt it worthwhile to look into it.  Having had lengthy communication with Adnaco, they have been extremely helpful and so we are now in the process of purchasing one for modification and comparison testing.

 

This unit consists of a PCIe card that converts a PCIe data stream to a fiber stream and in the process, this card performs one reclocking using a 101MHz clock.  A standard fiber cable then connects this card to an outboard fiber-to-USB 3.0 hub and It turns out this receiver module uses another 101MHz clock to re-time the incoming fiber signal and then a 24MHz clock to re-time it a 3rd time as the signal is converted to USB.  From my standpoint, the opportunity for 3 reclockings at strategic points could make this unit an ideal endpoint.  Moreover, there are no inaccessible subclocks that I need to worry about.

 

The USB 3.0 host controller used is the Texas Instruments TSUB7320, which I believe is the same host controller SOtM uses.  The receiver module is a very straightforward design that uses only 2 switching regulators -- 3.3V @1A and 1.1V @1A with an input voltage of 5V.  While 3.3V @1A will consume potentially 3.3 watts which equates to a fair amount of heat, Adnaco tells me the draw is more typically around 500mA at 3.3V and so converting to linear regulators should be no problem with the use of appropriately sized heat sinks.  Just to make sure heat isn't an issue, we plan to order this receiver board without an enclosure and will find a suitable larger enclosure at a later time.  Now if I can only convince @paulhynes to provide us a pair of his linear regulators to replace the two that come on this board, I think that in theory, this could prove to be a superb endpoint, maybe the best one yet.

 

As a counterpoint to this Adnaco solution, I am exploring one other fiber endpoint option.  As previously mentioned, FMCs are not a good solution because the circuitry used in these $30 units is less than ideal for audio.  Nonetheless, I have secured an Intel PCIe dual port fiber LAN card for only $90.  This card is identical to the Intel PCIe dual port LAN card I recently sent to SOtM except that instead of 2 copper-based RJ-45 connectors, it incorporates 2 SFP slots for fiber.  Unlike the newer SFP+ cards that run at a much faster 10GbE, this card runs slower at 1GbE but consequently consumes much less power (only 2.2w) and generates much less heat.  It utilizes a single 25MHz clock that will be replaced and I plan to bridge the 2 fiber ports.  I will then be sending SOtM 2 TP-Link FMCs and each has a clock that will be replaced.  One FMC will be used to improve  the stream from my router to my server and the other FMC will be used to connect to my sMS-200 Ultra.  Just like with my reclocking switch, the noisy switching regulators will be replaced and capacitors added.

 

Yes, this is a crazy set of experiments, especially when what I have already is sounding so spectacularly good but because several within my local audiophile society have been involved with my experimentation from the beginning and are intrigued by what these fiber options might provide, we have decided to pool our resources to fund all of this testing.  While I cannot say enough good things about the sMS-200 Ultra (thus far, the very best endpoint I have yet heard), the dX-USB HD Ultra (thus far, the very best USB-to-SPDIF converter I have heard) and this reclocking switch that has proven to be an uber bargain, ultimately, I am bothered that SOtM's dCBL-CAT7 Ethernet cables can still make such a huge difference.  My guess is that there still even more noise in the line that can be removed and an optimized fiber solution may hold the key.

Link to comment

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now



×
×
  • Create New...