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

[auricgoldfinger]

5 hours ago, austinpop said:

I think SOtM would be well advised to ship the OCX-10 standard with the Evox caps and eABS-200 paper. The experience with the SR-7 underscores the advantage of having a device that allows external PSUs, but of course the aggregate cost of the OCX-10 with an SR-7 rail is then quite high.

 

For anyone who has not yet purchased a master clock but is considering doing so, I think Rajiv's statement above is the crux of the issue.

 

For those who already own a vanilla OCX10 and may now be experiencing buyer's remorse, here are my takeaway thoughts:

 

1. The biggest jump in SQ seems to result from the Evox capacitor upgrade.  In fact, the difference is startling.  I strongly recommend sending your clock to SOtM for this upgrade.  Have them install the eABS-200, as well.

 

2. Changing power supplies from the sPS-500 to the JS-2 also yields a significant improvement, but not as much as the Evox capacitors.  The JS-2 is a worthwhile option, and in my opinion, is an excellent value relative to an SR7.

 

3. If you own or expect to acquire an SR7 and have a rail to dedicate to an upgraded OCX10, you will get excellent performance.  If the REF 10 is a 100, the upgraded OCX10/SR7 combination is a 95-97.

 

Unfortunately, we only had the stock JS-2 copper DC cable on hand for our testing.  Based on my experience adding the Audio Sensibility silver DC cable to various PSUs in my system, I believe the performance gap between the JS-2 and SR7 (which had the PH XL silver DC cable) would be further narrowed, although not eliminated.

 

Thanks to @limniscate for hosting our get together.  I enjoyed listening to his system.  Thanks also to @austinpop for setting up the testing protocols and managing the equipment changes so that Eric and I could do our listening blindly.  It was a fun and enlightening afternoon!

[Johnseye]

2 hours ago, lmitche said:

Been there and done that with PXE boot. I ran my system here for two years from a pxe server. In the end, booting from a cleanly powered USB hard disk sounds cleaner. Hauling the OS activity data over the network just creates a ton of noise on both Linux and Windows. I guess one could try it again with ramboot.

 

Today, with Optane, the low latency reduces the impact of this activity to negligible levels.

 

A nuc7cjys has been there focus of my attention here over the last few days. The bios is a nightmare. Intel pulled legacy boot as an option in the latest bios release. Linux will uefi boot with Windows removed, but the console is blank afterwards so we have to remote into the box for setup. Hopefully we will be to give this a listen today running as a roonserver from emmc.

 

It looks like that NUC has a Celeron.  Do you intend to upsample DSD512 with it, or is it your endpoint?

 

[sig8]

SOTM OCX 10 has four outputs and lists at $3,500 without any mods, Mutec Ref 10 has eight outputs and list for about the same price, and Ref 10 gives better SQ without requiring mods or high end power supplies. If all that is true, then I think there is no contest, or am I missing something here?

[austinpop]

20 minutes ago, sig8 said:

SOTM OCX 10 has four outputs and lists at $3,500 without any mods, Mutec Ref 10 has eight outputs and list for about the same price, and Ref 10 gives better SQ without requiring mods or high end power supplies. If all that is true, then I think there is no contest, or am I missing something here?

 

Given that Ref-10s have already started appearing in the used market in the mid-2k’s, you’ll not hear an argument from me!

[auricgoldfinger]

31 minutes ago, sig8 said:

SOTM OCX 10 has four outputs and lists at $3,500 without any mods, Mutec Ref 10 has eight outputs and list for about the same price, and Ref 10 gives better SQ without requiring mods or high end power supplies. If all that is true, then I think there is no contest, or am I missing something here?

 

You're not missing anything. 

[austinpop]

7 hours ago, str-1 said:

Thanks for that report back.  Were you able to use the same cables when comparing?

 

To the extent possible, yes. We used exactly the same clock cables, just swapping back and forth between the 2 OCX-10's and the Ref 10. Eric's setup uses a 50Ω BNC to SMB cable (Pasternack or Digi-key) for his dX, and 75Ω BNC (my "mystery" cable from a few months back) to his tX.

 

Power cables were stock for the Ref-10 and SR-7, Pangea AC-14SE for the JS-2, and DC cables were stock (Belkin with Oyaide connectors) JS-2 cables, and (Jaeger to Oyaide) silver cables with the SR-7.

 

7 hours ago, str-1 said:

The weight of evidence seems to argue strongly for a reference clock in systems with numerous clocks to sync but I’m wondering how much of the reported benefit from a modded SR7-powered OCX-10 I would get in my Zenith SE/Blu2/DAVE system where I would only have a tX-USBultra to connect to.

 

While there is a growing consensus of empirical (or anecdotal) evidence reported here on the positive improvements with a reference clock, compelling explanations of why are lacking. To some, that is the end of story. "Not interested. Call me when you can explain why." To others, that's not a deterrent, and that's the people on this thread. We try things with our eyes open, knowing that explantations may be forthcoming in due course, and indeed the answers may reveal simpler and cheaper solutions. 

 

Sorry - that turned into a sidebar. Back to your question. Where I was going when I started the above paragraph was that the evidence seems to suggest that the benefit comes from the low phase noise of the reference clock. In fact, we haven't seen compelling evidence that it comes from synchronization, which is in fact the more common use of external clocks, especially in the pro world.

 

To your question - I am like you, I only use the Ref-10 to provide a reference clock signal to exactly one device, the tX-USBultra, and it still makes a huge difference. You do not need a plurality of devices to benefit from a reference clock. This is also what, I believe, Roy was using in his setup when he still had an SE.

 

Refer to the audio setup diagram at the link in my signature.

 

7 hours ago, str-1 said:

It will be a while before I can give this a try.  I’ve still to finalise the spec for my SR7 with Paul Hynes and won’t be getting that until sometime in the New Year, by when an Mscaler will have replaced my Blu2.  I will also have to upgrade my standard spec tX-U for clock input.  Probably best to go for three rails with the SR7 - one DR and two non-DR (if I can do that).

 

One can never have too many rails of SR-7. Especially as it is a resource that won't be around for long. I've gone whole hog on my SR-7 build, with 3 rails of DRXL, in a Streacon chassis. Hoo-ah!

[austinpop]

7 hours ago, jean-michel6 said:

Does you or anyone has tried powering the OCX-10 with an LPS 1.2 as in my experience the LPS1.2 is significantly better than the SPS-500 .

 

No we didn't try that, as we didn't have the LPS-1.2 on hand, but I too rate it better/higher than the sPS-500.

 

7 hours ago, jean-michel6 said:

The person who loaned me the ocx10 also reported that he has a friend who has a mutec ref clock in his system and has achieved significant sq increase by disconnecting the internal psu of the mutec and powering it by an sps-500 ...probably the impact of outside psu which is quite often a good thing as it removes vibrations and electrical noise from sensitive components .

 

I wish I had the chops to attempt a mod like this! I'm surprised that he heard an improvement even with an sPS-500. If so, imagine the possibilities with better PSUs!

 

Has this "friend of a friend" posted his experience anywhere? Not that I am likely to try this any time soon, but I'd like to know what's involved.

[seeteeyou]

And then there's the question, do we wanna invest in a reference clock plus relatively expensive Habst cable(s) or wait for Hugo M Scaler (roughly $4,500) instead?

 

https://www.computeraudiophile.com/forums/topic/30376-a-novel-way-to-massively-improve-the-sq-of-computer-audio-streaming/?page=148&tab=comments#comment-723032

On 9/25/2017 at 5:16 PM, romaz said:

Practically speaking, this results in a massive improvement in DAVE's resolution, so massive that the collective impact of my server mods which includes 8 clocks being replaced pales in comparison to what Blu Mk2 provides. For those of you who own a Chord DAVE, I would suggest you prioritize getting a Blu Mk2 beyond anything else discussed on this thread. Combined with Chord's upcoming "digital" amplifiers, there will be no more resolute or transparent way of listening to a digital file. Despite all of this, I am finding, however, that the quality of the music server still matters.

 

For owners of Chord (Hugo 2/Qutest/Hugo TT 2/DAVE) DACs, the latter one should be a better bet since we could take advantage of all 1 million taps.

 

For the others, we're talking about only 250K taps (i.e. only 192kHz via either single coaxial or optical) and therefore the improvements might or might not be as good as what a reference clock could offer.

[Forehaven]

55 minutes ago, austinpop said:

 

To the extent possible, yes. We used exactly the same clock cables, just swapping back and forth between the 2 OCX-10's and the Ref 10. Eric's setup uses a 50Ω BNC to SMB cable (Pasternack or Digi-key) for his dX, and 75Ω BNC (my "mystery" cable from a few months back) to his tX.

 

Power cables were stock for the Ref-10 and SR-7, Pangea AC-14SE for the JS-2, and DC cables were stock (Belkin with Oyaide connectors) JS-2 cables, and (Jaeger to Oyaide) silver cables with the SR-7.

 

 

While there is a growing consensus of empirical (or anecdotal) evidence reported here on the positive improvements with a reference clock, compelling explanations of why are lacking. To some, that is the end of story. "Not interested. Call me when you can explain why." To others, that's not a deterrent, and that's the people on this thread. We try things with our eyes open, knowing that explantations may be forthcoming in due course, and indeed the answers may reveal simpler and cheaper solutions. 

 

Sorry - that turned into a sidebar. Back to your question. Where I was going when I started the above paragraph was that the evidence seems to suggest that the benefit comes from the low phase noise of the reference clock. In fact, we haven't seen compelling evidence that it comes from synchronization, which is in fact the more common use of external clocks, especially in the pro world.

 

To your question - I am like you, I only use the Ref-10 to provide a reference clock signal to exactly one device, the tX-USBultra, and it still makes a huge difference. You do not need a plurality of devices to benefit from a reference clock. This is also what, I believe, Roy was using in his setup when he still had an SE.

 

Refer to the audio setup diagram at the link in my signature.

 

 

One can never have too many rails of SR-7. Especially as it is a resource that won't be around for long. I've gone whole hog on my SR-7 build, with 3 rails of DRXL, in a Streacon chassis. Hoo-ah!

Rajiv, no concerns of multirailed ps's after we heard an improvement at John's place when he used one rail vs. two (I think that is what it was)?  JC

[austinpop]

1 minute ago, Forehaven said:

Rajiv, no concerns of multirailed ps's after we heard an improvement at John's place when he used one rail vs. two (I think that is what it was)?  JC

 

No, for 2 reasons. One, @Johnseye subsequently found an issue with his unit (can't recall offhand). And two, I've tried @limniscate's SR-7 in my system, and heard no ill effect from multirail. Note that both his rails have DR.

[PeterSt]

5 hours ago, seeteeyou said:

Did you try a distro that's running entirely from RAM? The real test could very well be comparing the latency of RAM versus that of Optane IMHO.

 

Maybe you remember I gave you a link about the "RAM-OS Disk". Possibly you did not see through it or thought I gave you a stupid SSD (indeed it was an SSD version I showed). Here the text which comes along with the HDD version of it - possibly it emphasizes the thing better :

 

Each OS can boot from RAM (all via remote control) and after that the HDD can be removed from the PC, with the idea that the PC contains nothing else. This improves Sound Quality largely (electrical noise is minimized).

 

Or maybe you want to do it all yourself ? I mean, it is a bit strange to see a 1000 links from you about the theoretical means while practice is just at your feet (in use by 100s of people for quite some years by now).

 

 

If you tell us that you are not interested in a commercial solution (because that's what it is) then at least I know that you understood it and read it. 

Btw, took about a year to develop with some pause-hurdles. But if you think this is stealing money from you ... I am fine with that thought. If it is regarded too much of a self-advertising, then I am also fine with that (and the post should be deleted). But I created it all for the better sound. And no ( @lmitche ), PXE boot did not do that.

 

Peter

[jean-michel6]

1 hour ago, austinpop said:

Quote

The person who loaned me the ocx10 also reported that he has a friend who has a mutec ref clock in his system and has achieved significant sq increase by disconnecting the internal psu of the mutec and powering it by an sps-500 ...probably the impact of outside psu which is quite often a good thing as it removes vibrations and electrical noise from sensitive components .

 

I wish I had the chops to attempt a mod like this! I'm surprised that he heard an improvement even with an sPS-500. If so, imagine the possibilities with better PSUs!

 

Has this "friend of a friend" posted his experience anywhere? Not that I am likely to try this any time soon, but I'd like to know what's involved.

To my knowledge he has not posted that anywhere but i will see if i can try to get some useful info  .

[Johnseye]

1 hour ago, Forehaven said:

Rajiv, no concerns of multirailed ps's after we heard an improvement at John's place when he used one rail vs. two (I think that is what it was)?  JC

 

1 hour ago, austinpop said:

 

No, for 2 reasons. One, @Johnseye subsequently found an issue with his unit (can't recall offhand). And two, I've tried @limniscate's SR-7 in my system, and heard no ill effect from multirail. Note that both his rails have DR.

 

Correct Rajiv.  I found one of my regulators to be faulty.  Paul shipped me a new one and things are back to normal.

 

Edit: I will note however that I'd prefer separate supplies and suspect that would be everyone's preference.  Independent transformers is ideal, it's just not cost or space effective.

[Confused]

10 hours ago, jean-michel6 said:

The person who loaned me the ocx10 also reported that he has a friend who has a mutec ref clock in his system and has achieved significant sq increase by disconnecting the internal psu of the mutec and powering it by an sps-500 ...probably the impact of outside psu which is quite often a good thing as it removes vibrations and electrical noise from sensitive components .

 

 

Can I just double check that the sPS-500 was used to Power a REF10?  The reason I ask is that it would seam a little radical to disconnect a LPSU and then power from a hybrid SMPS like the sPS-500.  Quite a few people have replaced the relatively cheap Meanwell SPS in the Mutec MC3+USB.   So this makes me wonder if your friend was actually using the sPS-500 to power an associated MC3+USB, rather than the REF10.  This would make more sense to me, hence this clarification, but you never know, some people like to try the radical!

[str-1]

5 hours ago, austinpop said:

To your question - I am like you, I only use the Ref-10 to provide a reference clock signal to exactly one device, the tX-USBultra, and it still makes a huge difference. You do not need a plurality of devices to benefit from a reference clock. This is also what, I believe, Roy was using in his setup when he still had an SE.

 

Thanks.  That’s reassuring.

 

5 hours ago, austinpop said:

One can never have too many rails of SR-7. Especially as it is a resource that won't be around for long. I've gone whole hog on my SR-7 build, with 3 rails of DRXL, in a Streacon chassis. Hoo-ah!

 

I’ll have another chat with Paul about the possibilities for maxing out on the specs but  I’m thinking at the moment that either two rails of DRXL or three rails with one of DRXL will be as far as I will go.  If I go for two rails, I would still have the SR4, which I could assign to whichever of tX-U, OCX-10 and Mscaler shows least improvement from a SR7 rail.

[Confused]

13 hours ago, str-1 said:

 

The weight of evidence seems to argue strongly for a reference clock in systems with numerous clocks to sync but I’m wondering how much of the reported benefit from a modded SR7-powered OCX-10 I would get in my Zenith SE/Blu2/DAVE system where I would only have a tX-USBultra to connect to.

I can add a little to @austinpop‘s comments.  Recently my SOtM kit has been away to Korea for upgrade, and I have been running a microRendu to my Mutec MC3+USB / REF10.  The results are superb, and got even better when adding the Habst cable, so another example of a system with good results where only one device uses the reference clock.  This is a somewhat different case, as the MC3+USB is feeding a Devialet amp via AES3, so an asynchronous system, but I thought it worth mentioning.

[One and a half]

15 hours ago, Superdad said:

 

PTFE heat shrink?  That’s a new one for me. Very cool!  Given that Teflon is generally very resistant to high heat, the notion of that product seems to be an oxymoron. 

If the heat gun is capable of 2000C? ?

[RickyV]

Does anybody know what exact type of evox capacitor thay use?

could not be that difficult to buy the evox and let some electronics repair shop replace the caps. It will be a lot cheaper. 

[auricgoldfinger]

2 minutes ago, RickyV said:

Does anybody know what exact type of evox capacitor thay use?

could not be that difficult to buy the evox and let some electronics repair shop replace the caps. It will be a lot cheaper. 

 

Sorry, I have no idea.

[RickyV]

19 minutes ago, One and a half said:

If the heat gun is capable of 2000C? ?

Google “ptfe heat shrink tubing” and be amazed.

[RickyV]

Oh no, evox is now owned by Kemet. 

I get shivers hearing that name. ? ?

[Confused]

37 minutes ago, Confused said:

....... This is a somewhat different case, as the MC3+USB is feeding a Devialet amp via AES3, so an asynchronous system, but I thought it worth mentioning.

Syncronous, rather....

[BigGuy]

1 hour ago, One and a half said:

If the heat gun is capable of 2000C? ?

 

Almost!  From evoxinc.com...

 

PTFE heat shrink tubing requires approximately 650 °F ± 25 °F (340 °C ± 5 °C) to initiate shrinkage. While this is a liberally safe range, these temperatures are approximate.

[lmitche]

12 hours ago, Johnseye said:

 

It looks like that NUC has a Celeron.  Do you intend to upsample DSD512 with it, or is it your endpoint?

 

John,  I honestly don't know how best to use it. Obviously it's not going to upsample to dsd512 with hqplayer.  Yes, it could make a great naa machine. 

 

So far we have tested it running roonserver for a old 44.1 to 48 khz 24 bit max r2r spdif dac. High res and dsd files are downsampled by roon, others played native. SQ was fantastic for twenty minutes until it stopped working. I'm not sure what happened but the latest bios is crazy flaky, and something is just not right.

 

Anyway this is very promising, so likely worth the struggle.

 

Stay tuned, Larry

[BigAlMc]

On 8/19/2018 at 8:42 AM, BigAlMc said:

I've emailed Ghent and asked whether he has any plans to offer JSSS-360 USB or LAN cables or whether these can be custom ordered. 

 

Should hopefully help settle whether reluctant types like Beer and myself need to reach for braid, a heat shrink and Google instructions ?

 

Cheers, 

Non-DIY capable Al

 

Ghent replied that he can make JSSS-360 USB or LAN cables on request. 

 

Email him if you are interested. 

 

[email protected]

 

On a separate note I should get my Lush^2 today. 

 

Cheers, 

Alan