Building a DIY Music Server

[cool_chris]

For the 2 Buffalows with Pink Faun clocks I am using for EACH switch :

Uptone lps1 - 7v out, followed by double regulated mpaudio SLS-Hpuln with 10x  LT 3045  for the clocks.

Vinnie Rossi mini supercap supplys with 3A and 5v out to the boards .

 

For 2 other Buffalows I use

Vinnie Rossi mini supercap supplys with 3A + 5V and

Vinnie Rossi mini supercap supply with 3A + 12 v .

 

I know it sounds crazy but planning to update this with Sean Jolacobs DC4 multirail. Just need little more time to collect the funds for that.

You cant go wrong with top supplys.

Best

Chris

 

 

[ASRMichael]

2 minutes ago, cool_chris said:

For the 2 Buffalows with Pink Faun clocks I am using for EACH switch :

Uptone lps1 - 7v out, followed by double regulated mpaudio SLS-Hpuln with 10x  LT 3045  for the clocks.

Vinnie Rossi mini supercap supplys with 3A and 5v out to the boards .

 

For 2 other Buffalows I use

Vinnie Rossi mini supercap supplys with 3A + 5V and

Vinnie Rossi mini supercap supply with 3A + 12 v .

 

I know it sounds crazy but planning to update this with Sean Jolacobs DC4 multirail. Just need little more time to collect the funds for that.

You cant go wrong with top supplys.

Best

Chris

 

 

Each to their own! working out total costs! 
 

4 switches 1k

4 PF clocks 6k

4 SJ rails 8k

+ wiring 1k

 

16k plus. I always say if you’ve got the money spend it on Hifi! (My wife says otherwise)

Enjoy & thanks for reporting 

[Blackmorec]

1 hour ago, cool_chris said:

Nenon is correct.

I got 4 Buffalo switches but sent 2 for the clock upgrade to Jord at Pink Faun.

He did fantastic work as you see on the pictures I submitted few pages ago.

Spering the money now to get other 2 upgrated. I need to have all 4 Buffalows  with Pink Faun clocks. Those are fantastic switches.

I wish you could hear the difference on my system. Night and day difference. 

There is no way to take any of the 4 Buffalows away or to get the 5th switch (Ether Regen) out of the chain.

 

I am glad Nenon confirmed the Telegartner M12 being excellent as a musical and sweetness adding solution. 

I wish I could afford to get one and install in my chain as the 6th switch.

 

For now I Prefer to keep Buffalows with PinkFaun clocks if they give 60 % of performance related to 40% of M12 performance ( Nenon rating).

So as far as I understand it correct

you get little more SQ for much less money.

Thank you Nenon !

Best

Chris

 

 

 

Hi Chris,

In my system I use Synergistic Research Atmosphere X Ref ethernet and USB cables. They are delivered with 2 ‘bullets’ which are exchangeable....the one bullet enhances the cable’s detail and spacial resolution, while the second bullet is more orientated around fullness and warmth, so with more than 1 cable you can fine tune exactly the balance between warmth, air and spacial resolution. 

[cool_chris]

2 minutes ago, Blackmorec said:

Hi Chris,

In my system I use Synergistic Research Atmosphere X Ref ethernet and USB cables. They are delivered with 2 ‘bullets’ which are exchangeable....the one bullet enhances the cable’s detail and spacial resolution, while the second bullet is more orientated around fullness and warmth, so with more than 1 cable you can fine tune exactly the balance between warmth, air and spacial resolution. 

 

Yes. You can do a lot with USB cables.

But I tested 25 different cables, solutions , cleaners , blockers , fibre and many more and got to the conclusion that they all work well, some very well, however only if your system is "mid class" 

If you have top system with perfected everything than removing all of the USB solutions gets you to the over the top sound quality. 

I liked very much the fibre usb isolation, the uptone iso regen and newest Intona 3.0 isolator .

 

[Blackmorec]

1 hour ago, cool_chris said:

 

Yes. You can do a lot with USB cables.

But I tested 25 different cables, solutions , cleaners , blockers , fibre and many more and got to the conclusion that they all work well, some very well, however only if your system is "mid class" 

If you have top system with perfected everything than removing all of the USB solutions gets you to the over the top sound quality. 

I liked very much the fibre usb isolation, the uptone iso regen and newest Intona 3.0 isolator .

 

Sorry Chris, i don't understand what you’re saying. How does a top class server with a USB output connect to the USB DAC input if not with a cable?   I found my system to work best with a simple USB cable....all the other stuff downgraded the sound

[auricgoldfinger]

1 hour ago, ASRMichael said:

Each to their own! working out total costs! 
 

4 switches 1k

4 PF clocks 6k

4 SJ rails 8k

+ wiring 1k

 

16k plus. I always say if you’ve got the money spend it on Hifi! (My wife says otherwise)

Enjoy & thanks for reporting 

 

Your estimate is too low.  He needs 8 SJ rails for the 2 DC inputs on each PF Buffalo pluse a 9th SJ rail for the EtherRegen. 🙂

[ASRMichael]

2 minutes ago, auricgoldfinger said:

 

Your estimate is too low.  He needs 8 SJ rails for the 2 DC inputs on each PF Buffalo pluse a 9th SJ rail for the EtherRegen. 🙂

20k then!!  Getting closer to an Extreme price! 

[auricgoldfinger]

Just now, ASRMichael said:

20k then!!  Getting closer to an Extreme price! 

 

True, although many Extreme users have multiple switches.  Some have 2x M12 Gold.

[Nenon]

I just tried two PF Buffalo switches. I went from:

Arris8200 --> Ubiquiti Router --> Buffalo --> PF Buffalo --> Server

to:

Arris8200 --> Ubiquiti Router --> PF Buffalo --> PF Buffalo --> Server

 

I heard a nice improvement, but we are deep into the woods of diminishing returns. If we quantify the improvement of the first PF Buffalo as 100% (not saying that it improved my system 100%!), the second PF Buffalo subjectively brought 15% of that. It's sounding nice, but I might be able to spend my money better on other improvements. However, for people who have upgraded everything else, that switch rabbithole would bring some joy. 

 

It's hard to describe what I am hearing, but everything sounds more real. Nothing is standing out in particular. The soundstage is the same, the overall tone and frequencies are the same. It's just more engaging and real sounding. 

 

Not an Elvis fan, but Fever from the Qobuz album below is a reference track that I use sometimes. 

I typically listen to about a minute of the Fever track and move on. I found myself getting engaged and listening to the entire album. I did not even realize it until I started typing this post. This switch must be doing something good with the Qobuz stream I guess if it makes the music more engaging. But yes, it could be a crazy and expensive rabbit hole. This stuff is for people who have done everything else in their system (and have a high Hifi budget). BTW, I am not one of them! :)

[Exocer]

Would a non-PF Buffalo add anything to the above setup? 

 

If one PF Buffalo added gains 15%, does the non PF Buffalo take something away or just add way less than a 2nd PF Buffalo would?

[Nenon]

1 hour ago, Exocer said:

If one PF Buffalo added gains 15%, does the non PF Buffalo take something away or just add way less than a 2nd PF Buffalo would?

 

The 15% was a relative number.

 

Let's say we start with two stock clock Buffalos. I have written previously about the improvement of one stock clock Buffalo vs. two, three, four, five, etc. 

 

Then I replaced the stock clock on one of the Buffalos. And that was a huge improvement that I've also written about (let's call the level of improvement X). Replacing the clock on the second Buffalo did not give me the same (X) improvement... it gave me only 15% (of X) improvement .

In other words you get the best improvement from upgrading one of the switches. Upgrading more is an improvement but  diminishing returns. 

 

Hope that makes more sense. 

 

1 hour ago, Exocer said:

Would a non-PF Buffalo add anything to the above setup? 

I did not try adding a non-PF Buffalo to a set up with two PF Buffalos. I don't have enough power supplies, DC cables, and network cables to do that at the moment. But @cool_chris already commented on that. He is using two stock clock Buffalos and two PF Buffalos. 

[Blackmorec]

15 hours ago, djwalter2000 said:

I found this video to be enormously helpful in my understanding of how digital can be suspectable to quality issues

 

https://youtu.be/grzoqEb2KMk

Thanks for posting. This really is an excellent video for the lay enthusiast.

[cool_chris]

11 hours ago, Blackmorec said:

Sorry Chris, i don't understand what you’re saying. How does a top class server with a USB output connect to the USB DAC input if not with a cable?   I found my system to work best with a simple USB cable....all the other stuff downgraded the sound

 

Sorry. You are correct this cant work 😆

I missed the one phrase  "except of the Intona Ultimate cable that is still in place" I removed everything.

But I keep everything , because the nirvana is not a stabile thing.

From time to time you want to change this or that OR after server / softwere update dosn't sound as " magical" anymore , so you need to experiment if one of the multiple USB solutions are improving the SQ or not.

So I come back and forth many times sometimes to set it to the new "absolute"

Especially Roon update can be very devastating to the SQ sometimes. Sometimes it improves SQ so you never know.

 

 

[cool_chris]

8 hours ago, Nenon said:

 

The 15% was a relative number.

 

Let's say we start with two stock clock Buffalos. I have written previously about the improvement of one stock clock Buffalo vs. two, three, four, five, etc. 

 

Then I replaced the stock clock on one of the Buffalos. And that was a huge improvement that I've also written about (let's call the level of improvement X). Replacing the clock on the second Buffalo did not give me the same (X) improvement... it gave me only 15% (of X) improvement .

In other words you get the best improvement from upgrading one of the switches. Upgrading more is an improvement but  diminishing returns. 

 

Hope that makes more sense. 

 

I did not try adding a non-PF Buffalo to a set up with two PF Buffalos. I don't have enough power supplies, DC cables, and network cables to do that at the moment. But @cool_chris already commented on that. He is using two stock clock Buffalos and two PF Buffalos. 

 

I agree with you Nenon, the only difference  I see is that in my case it is not X+ 15 % as it is in your case

In my system is rather X+ 50 % so not so much " diminshing returns "

It is solid very nice improvement.

so the first PF Buffalo is X ( in my case it is FEW levels up)

than senond PF Buffalo is about + 50% ( X+ 50 %)

As you remember I have 2 more Buffalows

so adding Nenon moded Buffalo to the above 2 PF Buffalows adds about 15 % ( X+ 50 %+15%)

Adding the 4th switch ( also Nenon moded Buffalo) adds additional 15 % ( X+50%+15 % + 15%)

Please keep in mind each time your starting point is much much higher !

So in reality once you get lests say about 100 % improvement (we call it X )with first swich adding second gives addtional 50 % form the new reference starting point not from the original starting point. I hope I am clear about that.

 

I had also an Ether Regen that was great but not so great as the stock Buffalo for sure.

I addesd it as the 5th switch ( A to A side ! = A side Ethernet to Startech SFP)

just after the modem as it added about 20 % I think it  double isolates the signal from modem.

It has build in transformers in ethernet ports as well as the Fiber opic cable isolates some noise futher. 

I know it is crazy but did try to remove one switch from that LONG chain, hoping I can sell it and invest in different regions of my system but I cannot 

listen to the music without one of the 5 switches. The loss is to painful to my ears.

 

Paul Pang is offering 4 x switch ( Reference QUAD SWITCH) in one enclosure in his e-store for a reason I think. He must have been one of the very early pioneering people in that area.  

It has one power cable but 4 internal linnear power supplys and 4 separate switch boards.

 

There was someone here in the one of the forums mentioning long ago that

 

" Signal going around the glob via hundreds of switches and servers gets devasteted with inpinting the 

character of the each switch / clock on the way to your home ,

so you might need FEW switches to rerender back some of the original content " 

 

I think this makes sense. I was able to confirm it is doing something to the signal / noise that brings the 

positive effects each time. But stopped at 5 switches.

Expecting to power it better than it is nnow and see if Taiko swich will come soon as the 6th swich. 

I dont know the answer how this is happening we are getting better SQ with so many switches.

All I know I get a way better SQ in my room each time.

Tried to calculate that using "Nenons fomula"  but I agree with Nenon that is is very hard to describe the 

difference because everything gets better.

So I think what we were doing until the SWICH era , we were modulating the SQ in one domain ( electrical equelizing ?)

Now with the use of the OXCO clock modified switches we are modulating something else ( time domain ?)

I wish someone can explain that soon.

 

 

 

[Blackmorec]

The video that DJWalter posted switched on a few Mmmmmms in my mind. From that video I can see how a signal may be damaged and how buffers will have no effect on removing those defects. Its also clear that damage can be gradual and cumulative....the more bad switches and power supplies, the more damage occurs....then by the same token the more ‘good quality’ regenerations, the more the signal defects are gradually corrected. 

Given that some of those regenerations happen in network modules other than switches, it explains why good quality LPSs have a positive effect on SQ wherever they are placed.  Similarly, why anti-vibration measures will positively effect the quality of the regeneration, and why good cables will prevent further de-generation. 

 

Why will we see multiple switch related quantitative differences in SQ improvements between different systems? It must depend on the quality of the stream reaching those ganged switches. The more profound the arriving data stream damage, the greater the level of SQ improvement multiple switches will bring. 

 

 

[dminches]

@cool_chris What DAC and other equipment are you using?  Just curious.

 

[dctom]

On 9/22/2020 at 9:45 PM, austinpop said:

posted by mistake!

 

[Exocer]

So if someone were to get their hands on a PF Ultra OCXO (with streaming quality being the highest priority) would it make the most sense to place the clock on the Buffalo switch OR the motherboard?

[seeteeyou]

Not exactly easy to acquire BS-GS2016 these days unless we're willing to import them from their country of origin.

 

Regarding the motherboard, maybe it kinda depends on the cost of motherboard itself since it's a bit too risky to fry an expensive one?

 

And then we also have to figure out where the on-board Ethernet is connected to, perhaps its original 25 MHz clock ain't worth the replacement if that were connected to the PCH? Sometimes we're lucky to find something that's connected the CPU directly but it might not be that common to begin with.

 

Ultimately we still have to take this into consideration if local files were all we need for a motherboard with BMC (baseboard management controller)

 

https://audiophilestyle.com/forums/topic/30376-a-novel-way-to-massively-improve-the-sq-of-computer-audio-streaming/page/656/?tab=comments#comment-1053701

On 5/22/2020 at 3:11 PM, darkfrank said:

I must say that even I'm already using the Telegartner M12 switch, by using ROON to play local file only, unplug the network cable still improve the SQ for about ~20%.

 

https://audiophilestyle.com/forums/topic/30376-a-novel-way-to-massively-improve-the-sq-of-computer-audio-streaming/page/699/?tab=comments#comment-1077821

On 9/4/2020 at 12:56 AM, austinpop said:

Disconnected is best: My preferred mode of listening is local playback (from local SSD) using HQPlayer, with the network disconnected. Not only does this sound sublime, but it forms a reference for what the network chain needs to achieve.

 

https://audiophilestyle.com/forums/topic/58164-building-a-diy-music-server/page/32/?tab=comments#comment-1082113

On 9/29/2020 at 1:55 PM, Nenon said:

But do we really need to spend that much money to buy both? I am not really sure. Even with both of these switches connected to my system, playing local files completely disconnected from the network sounded best. 

 

---

 

OTOH, we could also put our switches between the music server and the USB DAC instead

 

https://www.whatsbestforum.com/threads/sotm-snh-10g-network-switches-x2-the-new-king-of-usb-network-gadget-setup.27758/page-19#post-621571

[Nenon]

18 hours ago, dminches said:

@cool_chris What DAC and other equipment are you using?  Just curious.

Chris has an amazing system but does not like to talk much about it. Don't be surprised if you don't hear from him, just take my word for it :). 

 

49 minutes ago, Exocer said:

So if someone were to get their hands on a PF Ultra OCXO (with streaming quality being the highest priority) would it make the most sense to place the clock on the Buffalo switch OR the motherboard?

Good question. But I don't think there is a good general answer. I believe it is system dependent. In some cases you would get better results by replacing the clock on the motherboard. In other cases you would be better doing the switch. The only way to find out is to try it. If I was to gamble on it, I would pick the motherboard... But I have no desire, intentions, or interest to replace any of the the clocks on my Asus SAGE motherboard. Hence, I did my switch. 

 

12 minutes ago, seeteeyou said:

Not exactly easy to acquire BS-GS2016 these days

I will be putting one for sale in the classified section soon. Actually, I have several things to put there. I can never find time for that...

[seeteeyou]

Wise words from Emile

 

https://audiophilestyle.com/forums/topic/30376-a-novel-way-to-massively-improve-the-sq-of-computer-audio-streaming/page/616/?tab=comments#comment-1014096

On 12/16/2019 at 7:37 AM, romaz said:

You really want the least possible variation in load, and higher cpu power / bandwidth systems are better at that with very low load music playback processes."

 

https://www.nas.nasa.gov/assets/pdf/papers/NAS_Technical_Report_NAS-2014-01.pdf

Quote

One of the desirable features associated with InfiniBand, another network fabric technology, is its Remote Direct Memory Access (RDMA) capability. RDMA allows for communication between systems but can bypass the overhead associated with the operating system kernel, so applications have reduced latency and much lower CPU utilization. This results in much faster network performance rates than traditional TCP/IP.

 

At some point maybe we've gotta leave TCP/IP behind since they aren't any good for our CPU(s) in terms of the least variation in load. Hopefully RDMA / RoCE / iWARP could get us one step closer to the "ideal" conditions of being unplugged / disco.

 

---

 

On the software side, there's a way to bypass the TCP/IP stack altogether

 

https://www.symphonic-mpd.com/forum/topic/97/rpi4版-v1-0系のサポートはこちら/300

Quote

一つは「カーネルのTCP/IPスタックをバイパス」すること。

 

Even both IP header and UDP header could be omitted

 

https://www.symphonic-mpd.com/forum/topic/97/rpi4版-v1-0系のサポートはこちら/302

Quote

直結に限定すれば、IPヘッダ・UDPヘッダを省くことも技術的には可能です。

 

Speaking of the least variation in load, IMHO most Linux distros and even stripped-down Windows 10 LTSC 2019 couldn't get anywhere close to this guy

 

https://audiodigitale.eu/?p=228

Quote

I built SqueezeCore because I needed a lightweight, minimal, plug&play OS to use it as a squeezelite player. The whole OS is 17MB and when running it occupies around 60MB of RAM; SqueezeCore must be installed on a USB drive and will run completely in RAM.

 

[cool_chris]

20 hours ago, Blackmorec said:

The more profound the arriving data stream damage, the greater the level of SQ improvement multiple switches will bring. 

 

I must say this is very true in my system. 

I started long ago with radio based Internet.  

Not many know that in fact this are the very strong WiFi antennas (not radio broadcast !) with transmitting distance upto 5 km . This is a very strong signal.

There is more.

Usually this kind of Internet is used in difficult terrain and by that they use much stronger antennas to transmit it further. 

So in my case there was 3 transmission antennas and 3 receiving antennas before it reached the server og the internet provider.

I was able to get the difference with 5GHz and 2,4 Ghz and 5 Ghz destroyes the signal much more than 2,4 antenna.

Also I paid to get installed direct fibre to exclude the 3x plus 3x wifi demage and the outcome is clear.

Using the same system , same stacked switches music is by far better in every possible way. 

I left the radio internet (wifi) for home / office use and 200 mbps fibre is exclusive for music server only.

For those with radio Internet on the roof it will be much better investment to get the fibre home than anything else. 

[dctom]

On 9/22/2020 at 9:45 PM, austinpop said:

 

I would highly recommend you try bypassing your NUC entirely, and going straight to the DAC. This is based on one key assumption I'm making regarding your use of HQPlayer. You said your were upsampling to DSD256 and PCM384. My assumption is that you're only doing DSD-to-DSD and PCM-to-PCM, in which case the CPU load on the server even with upsampling should be low. If, on the other hand, you are running at high CPU utilization doing PCM-to-DSD conversion, then the following does not apply.

 

Assuming the low-CPU use case is true, do try the direct path and see what you think. When you do, keep in mind that the USB ports on the Z390 Gaming ITX/ac mobo don't all sound the same. I found the USB 3.1 Gen 2 ports sound better. Feel free to do some trial and error and pick the best by ear. For me, the USB 3.1 Gen 2 port directly under the Ethernet port sounded best. Give it a try.

 

Finally, if you do find the server direct path to sound better, with more dynamics, then you could apply the cost of the NUC towards a tX-USBultra. Or sell the NUC and DC3, and get a Phoenix.

 

I have now got around to using the server on its own without the NUC as naa.

This has improved the sound quality further. The SQ with the NUC naa is slightly harder, less fluid and less transparent, so thank you again for the advice Rajiv. I did try this comparison a while ago, when the server was less well developed, and preferred the 2pc set up - obviously as the configuration changes, better PSUs etc, the interaction between components combine to produce a final result that can't be predicted!

I read your report on the Pheonix so will try and get hold of one to try out.

When commenting on your SGM you mentioned you are using it with HQP and naa. Is that a naa within the Extreme itself (2nd cpu) or an external naa?

As others have found, HQP is an amazing tool to tweak and optimise the sound. I play all the music locally having my music files stored internally  on a NVMe drive, this has been an improvement over an external NAS.

[SimoneF]

12 hours ago, seeteeyou said:

Wise words from Emile

 

https://audiophilestyle.com/forums/topic/30376-a-novel-way-to-massively-improve-the-sq-of-computer-audio-streaming/page/616/?tab=comments#comment-1014096

 

https://www.nas.nasa.gov/assets/pdf/papers/NAS_Technical_Report_NAS-2014-01.pdf

 

At some point maybe we've gotta leave TCP/IP behind since they aren't any good for our CPU(s) in terms of the least variation in load. Hopefully RDMA / RoCE / iWARP could get us one step closer to the "ideal" conditions of being unplugged / disco.

 

---

 

On the software side, there's a way to bypass the TCP/IP stack altogether

 

https://www.symphonic-mpd.com/forum/topic/97/rpi4版-v1-0系のサポートはこちら/300

 

Even both IP header and UDP header could be omitted

 

https://www.symphonic-mpd.com/forum/topic/97/rpi4版-v1-0系のサポートはこちら/302

 

Speaking of the least variation in load, IMHO most Linux distros and even stripped-down Windows 10 LTSC 2019 couldn't get anywhere close to this guy

 

https://audiodigitale.eu/?p=228

 

 

I'm the dev behind Squeezecore. While I have not updated the thread here in audiophilestyle I've released new updates. If anyone found something not working correctly (DAC not recognized for example) feel free to PM me.

[ASRMichael]

13 hours ago, Nenon said:

Chris has an amazing system but does not like to talk much about it. Don't be surprised if you don't hear from him, just take my word for it :). 

 

 

@nenon. I’ve noticed over time not everyone shows their system details on their profile. Including yourself. 
 

I’ve always asked myself why is this? Any ideas? I actually find it really helpful by looking at other people’s setup. Really helps.