Building a DIY Music Server

[seeteeyou]

On 10/11/2020 at 8:31 AM, Exocer said:

You're welcome, all I did was look up LVCMOS OCXOs @ 25Mhz on Digikey and @Soul Analogue confirmed the results. 

 

@tgb, looking forward to your project with the OH4510LF. Best of luck, and hopefully the results are worthwhile!

 

OH4610LF-025.0M = ±10ppb @ $55.1 (Minimum 1)

 

https://www.digikey.com/en/products/detail/connor-winfield/OH4610LF-025-0M/5641634

 

SIT5711AI-KW-33N-25.000000N = ±5ppb @ $59.68 (Minimum 500 = $29,840)

 

https://www.arrow.com/en/products/sit5711ai-kw-33n-25.000000n/sitime

https://www.sitime.com/part-number-decoder

 

---

 

Any SiT5711 OCXO @ 25 MHz = 1,080 RMB / $165 (Minimum 1)

 

https://item.taobao.com/item.htm?id=585481394277

 

SIT5711AI-KW-33N-25.000000F = $106.25 (Minimum 25 = $2656.25)

 

https://www.arrow.com/en/products/sit5711ai-kw-33n-25.000000f/sitime

https://www.sitime.com/part-number-decoder

 

---

 

A wider temperature range (-40 to 85 as shown above) should be superior to what JCAT picked for their USB Card XE (20 MHz) and NET Card XE (25 MHz) as follows

 

https://audiophilestyle.com/forums/topic/58899-usb-card-xe-xtreme-edition-usb-audio-output-with-ground-breaking-performance/page/2/?tab=comments#comment-1065007

On 7/3/2020 at 1:57 PM, Marcin_gps said:

The current lead time information for the OCXO SIT5711AC-KW-33E-20.000000F-ND is that they are expected to arrive at Digi-Key on July 8, 2020.

 

---

 

https://audiophilestyle.com/forums/topic/30376-a-novel-way-to-massively-improve-the-sq-of-computer-audio-streaming/page/644/?tab=comments#comment-1039516

On 3/31/2020 at 6:40 AM, romaz said:

According to Marcin, it uses a new OCXO rated at 5ppb but what he found to be the more important characteristic is that this clock is somehow immune to vibrations.  Apparently but not surprisingly, Marcin and his OEM (Adnaco) found that a high performance clock in a vibratory environment has the potential to perform more poorly than a lower performance clock in a vibration-free environment and so while others may be touting OCXOs with better lab measurements under ideal conditions, this card was designed to function optimally under real world conditions.

 

Minimum 500 would be quite a stretch unless we're able to split $29,840 among 50 or so participants, though $2656.25 wouldn't be THAT bad for 25 pieces of high caliber OCXO.

 

AFAIK plenty of stuff could take advantage of an OCXO @ 25 MHz since both Ethernet (NICs / switches etc.) and USB 3 could perform much better with an upgrade.

 

---

 

BTW, here's their latest DCOCXO that's even more interesting

 

https://www.digikey.com/en/products/detail/sitime/SIT5721AC-KW333-T-24-57600F/11696570

https://www.arrow.com/en/products/sit5721ai-kw033-t-50.000000/sitime

https://media.digikey.com/pdf/Data Sheets/SiTime PDFs/SiT5721_Web.pdf

Quote

The SiT5721 can be factory-programmed to any frequency between 1 and 60 MHz. In addition to industry leading performance, the DCOCXO is offered in a 9 x 7 mm package, the smallest Stratum 3E OCXO available. 

 

Theoretically we could get 45.1584 MHz / 49.152 MHz for upgrading DACs etc. but their lead time might be a concern.

[seeteeyou]

FYI - interesting post

 

https://cafe.naver.com/jazzkfpa

https://audiophilestyle.com/forums/topic/60871-korean-no-1-high-end-linear-power-supply/

Quote

Nayuta Flagship LPS is 3 rails / 4 rails / 5 rails.

 

- 3 rails :  40Kg / Price (Included Shipping Fee) : $ 7,000

- 4 rails :  50Kg / Price (Included Shipping Fee) : $ 8,500

- 5 rails : 62Kg / Can't sell this version to the world

 

Nayuta Audio Linear Power Supply

https://www.monoandstereo.com/2020/02/nayuta-audio-linear-power-supply.html

https://www.facebook.com/permalink.php?id=1800464690274533&story_fbid=2572476489740012

 

---

 

https://www.custom-hifi-cables.co.uk/power-supplies/dc4-power-supply

Quote

£3000 (4000 USD for North America)
(1 output) e.g. Chord M-Scaler

 

£3700 (5000 USD for North America)
(2 outputs) e.g. a switch/clock combo

 

£4400 (6000 USD for North America)
(3 outputs) e.g. Chord DAVE or BLU mk2

 

[butifull01]

6 hours ago, seeteeyou said:

FYI - interesting post

 

https://cafe.naver.com/jazzkfpa

https://audiophilestyle.com/forums/topic/60871-korean-no-1-high-end-linear-power-supply/

 

Nayuta Audio Linear Power Supply

https://www.monoandstereo.com/2020/02/nayuta-audio-linear-power-supply.html

https://www.facebook.com/permalink.php?id=1800464690274533&story_fbid=2572476489740012

 

---

 

https://www.custom-hifi-cables.co.uk/power-supplies/dc4-power-supply

 

 

Your linked product is Nayuta Entry Level LPS.

 

Highly less than Flagship LPS.

 

I sell Goebel High End only speakers, Entreq only flagship, Light Harmonic only for Korea Full custom DACs, Skogrand Cable Flagship, Denafrips only Terminator Plus (The most cheapest product of mine), $ 600 Brimar Custom Fuses, etc.

 

I never sell entry product.

 

Nayuta Entry Model Name is 20th anniversary LPS for Korean frugal audiophiles.

 

I did AB test in 5 minutes, and 20th quality is 45% of my Flagship LPS.

[lmitche]

On 10/29/2020 at 11:49 PM, seeteeyou said:

since a more powerful CPU should sound (much) better under most circumstances

As a community service, we should stop promoting the falsehood that more CPU power means better SQ. It does depend on circumstances, and in most cases is just untrue.

[dminches]

13 minutes ago, lmitche said:

As a community service, we should stop promoting the falsehood that more CPU power means better SQ. It does depend on circumstances, and in most cases is just untrue.

 

In trying to refute someone's opinion you have injected your own.  I don't agree that "in most cases" it doesn't sound better.  It all depends on the design.

 

In my experience, if you can properly cool and power the higher powered CPU it will sound better than the lower powered counterpart.

 

[seeteeyou]

The thing is, not THAT many of us could / would have access to enough SR7 DR / DC4-tier (maybe OPTIMO_ATX down the road, too?) rails for obvious reasons

 

https://audiophilestyle.com/forums/topic/30376-a-novel-way-to-massively-improve-the-sq-of-computer-audio-streaming/page/614/?tab=comments#comment-1011955

On 12/10/2019 at 4:25 PM, austinpop said:

This super-combo just smokes both the red and blue combos! I have a hard time even articulating how good this sounds. The HDPlex Linear ATX supply is already very good, but using the SR-7 to power the server just takes it to another level. SQ improves across the board, in dynamics, transparency, bass articulation, tonality and imaging. I finally got to hear what @romaz has been raving about, and it is truly something!

 

And then there's something about having the right software that could actually take advantage of those substantial investments in hardware

 

https://audiophilestyle.com/forums/topic/30376-a-novel-way-to-massively-improve-the-sq-of-computer-audio-streaming/page/674/?tab=comments#comment-1062580

On 6/23/2020 at 12:43 AM, austinpop said:

What my experience showed me is that the 8700K sounded better than the 8700T at the same operating frequency:

• IF the PSU is big enough to easily energize both CPUs for all the factors above, and
• when using the performance governor (default) in Euphony that sets and holds the CPU frequency to the max value specified in the settings.

Does this prove anything? Nope. And if asking on any other OS than Euphony, without knowing how the OS will control CPU frequency dynamically, I wouldn't be able to say. But certainly for Euphony, my inclination would be to go with the K CPU, if you're confident of your PSU's capacity.

 

https://audiophilestyle.com/forums/topic/58164-building-a-diy-music-server/page/28/?tab=comments#comment-1077902

On 9/4/2020 at 6:07 AM, Nenon said:

It's the sum of all parts that matters. When I built my dual Xeon CPU server and ran Euphony it did not sound better than my single CPU AMD Ryzen with pinkFaun ultraOCXO motherboard clock. 

But after spending a few months to tweak a custom Windows LTSC, and after isolating the server and the streamer part on different CPUs, the dual Xeon is on another level. 

So there is no simple answer to your question. A single CPU server if tweaked well may sound better than a not so-well tweaked dual CPU server. There are so many variables, that it's hard to generalize. 

 

It's kinda like putting the right person, in the right place, at the right time.

 

Hardware, power, cables, server / client combos, software player, bit-perfect versus upsampling, CPU affinity, process isolation, ramroot / RAM drive, BMC + disabling NIC, CPU offload with RDMA / SMB Direct, kernel timer frequency, BFS / MuQSS CPU scheduler, Xenomai kernel with RTDM, compiling codes with GCC settings for a specific CPU type, so on and so forth.

 

Since it's a thread for subjective discussions, just about any kinda truth / falsehood / truthiness etc. could be both right and wrong IMHO.

[lmitche]

2 hours ago, dminches said:

 

In trying to refute someone's opinion you have injected your own.  I don't agree that "in most cases" it doesn't sound better.  It all depends on the design.

 

In my experience, if you can properly cool and power the higher powered CPU it will sound better than the lower powered counterpart.

 

What OS and player software do you run?

[dminches]

26 minutes ago, lmitche said:

What OS and player software do you run?

 

I am running Euphony / Stylus.

 

I have a Xeon-based server with a Sean Jacobs DC4.

 

[lmitche]

57 minutes ago, dminches said:

 

I am running Euphony / Stylus.

 

I have a Xeon-based server with a Sean Jacobs DC4.

 

Thanks Dminches,

 

My experience is that in a computer running Euphony/Stylus more CPU power means better SQ. I believe there are many here that have had a similar experience with Euphony such as yourself.

 

The original comment was "since a more powerful CPU should sound (much) better under most circumstances". This is a complete generalization that makes no mention of Euphony. My objection is with the generalization.

 

The server mentioned in the post doesn't run Euphony. The post states it comes "with custom BIOS / stripped down version of Windows 10 IoT Enterprise LTSC" running on a Pentium J4205. While the Pentium is far from the fastest Intel processor, to imply that SQ must be below Euphony running on the fastest Intel processor is conjecture at best. We wouldn't know unless we have a listen. That is straight logic, not an opinion.

 

I believe people should listen to a solution before sharing an opinion based on some "theory" or extrapolation of others experience. There is no substitute for "hands-on" empirical evidence. An opinion on sound quality based on anything else is a disservice to the community.

[dminches]

43 minutes ago, lmitche said:

The server mentioned in the post doesn't run Euphony. The post states it comes "with custom BIOS / stripped down version of Windows 10 IoT Enterprise LTSC" running on a Pentium J4205. While the Pentium is far from the fastest Intel processor, to imply that SQ must be below Euphony running on the fastest Intel processor is conjecture at best. We wouldn't know unless we have a listen. That is straight logic, not an opinion.

 

I believe people should listen to a solution before sharing an opinion based on some "theory" or extrapolation of others experience. There is no substitute for "hands-on" empirical evidence. An opinion on sound quality based on anything else is a disservice to the community.

 

I think people need to be able to weed through all the information here and be able to separate opinions from fact-based data.  Every post is basically "YMMV."

 

Regarding high/low powered CPUs and different operating systems, I have run Audiophile Optimizer for years.  I followed Phil's recommendation for tweaking the BIOS by turning off turbo mode and other settings which allowed the CPU to operate in a high-power mode.  However, once I got a top LPS in the DC4 I found my Windows Server/Audiophile optimizer setup (I have 2 boot drives) sounded better without the BIOS tweaks.  I am a firm believer that the power supply is so critical with these servers.

 

 

[Dev]

30 minutes ago, dminches said:

I am a firm believer that the power supply is so critical with these servers.

power supply is critical to any component related to audio. If your DAC power supply has a broken design or your Amp’s power supply is busted, no matter how much DC4 you put into your server, will it sound good ? It won’t. The only reason why power supply for a server is deemed very especial is just because folks here are on the DIY route to these music servers. In other forums where folks mostly use commercially available gears, nobody talks about power supply of a server. Companies, like Ayre, Dagostino, Luxman, Esoreric, Gryphone and many more, won’t leave their power supply half hearted and still sound good. It’s just not possible.
 

The fact is, once you delve your hands into DIY for DACs and other analog gears, you will perhaps realize the important of power supply in all places makes for good audio. 

 

[RickyV]

Just wondering/ thinking out loud here but many people are already using the switching HDPlex 800 dc-atx power supply and feed with a linear power supply. What if there was a quality smps would that be better with it’s faster high current delivery? Does the low noise of the linear power supply count if there is a noisier switching atx supply after the linear one, I think the noise of the switching regulator counts. 
So why not use an audio grade smps from hypex? They have various power supplies for there ncore amps with output voltage ranging from 2x 36V to 2x63V all acceptable for the 800 dc-atx even the two banks. 

[Nenon]

25 minutes ago, RickyV said:

Just wondering/ thinking out loud here but many people are already using the switching HDPlex 800 dc-atx power supply and feed with a linear power supply. What if there was a quality smps would that be better with it’s faster high current delivery? Does the low noise of the linear power supply count if there is a noisier switching atx supply after the linear one, I think the noise of the switching regulator counts. 
So why not use an audio grade smps from hypex? They have various power supplies for there ncore amps with output voltage ranging from 2x 36V to 2x63V all acceptable for the 800 dc-atx even the two banks. 

It would be a decent way to get a computer powered in a more affordable way than using high quality LPS. Definitely worth a try.

The HDPlex 800 dc-atx is quite noisy, and I don't use it to power my entire computer. But it's the best I have found on the market. For me powering up just the ATX connector (well just the 3.3V and 5V in my case) with the HDPlex fed by a good 5A LPS is an acceptable compromise. The CPU and USB card are powered by different rails. 

 

Acopian makes linear power supplies with really good specs. They are built really well, measure pretty good, and you can get a high current top model easily at a very reasonable price on eBay. But they never sounded good in my system. I don't really recommend them for a high-end system, but if someone is on a tight budget an Acopian feeding an HDPlex 800 dc-atx power supply might give surprisingly good results for the money you pay. I am assuming you get the Acopian for $200 or less on eBay.

The Hypex might be a decent way to do that as well. We can call it HyPLex :). 

[dminches]

The market is wide open for someone else to build a DC-ATX converter.

 

[RickyV]

5 hours ago, Nenon said:

It would be a decent way to get a computer powered in a more affordable way than using high quality LPS. Definitely worth a try.

The HDPlex 800 dc-atx is quite noisy, and I don't use it to power my entire computer. But it's the best I have found on the market. For me powering up just the ATX connector (well just the 3.3V and 5V in my case) with the HDPlex fed by a good 5A LPS is an acceptable compromise. The CPU and USB card are powered by different rails. 

 

Acopian makes linear power supplies with really good specs. They are built really well, measure pretty good, and you can get a high current top model easily at a very reasonable price on eBay. But they never sounded good in my system. I don't really recommend them for a high-end system, but if someone is on a tight budget an Acopian feeding an HDPlex 800 dc-atx power supply might give surprisingly good results for the money you pay. I am assuming you get the Acopian for $200 or less on eBay.

The Hypex might be a decent way to do that as well. We can call it HyPLex :). 

The idea only surfaced because the hypex is build for audio and to be near audio equipment. I would not go for an Acopain or any other non audio smps.   
I am not in the market for atx server I am happy where I am for now. I would have probably done the 3.3v and the 5v linear too, as you have in the past.

Maybe especially for the cpu because of the enormous current peaks. Just wondering out loud 😀

[Soul Analogue]

The hardware and PSU part of my EPYC server are almost done

have successfully separated power supply an intel i350 lan card (removing the switching regulators on the card) by 1.0v and 1.8v

the 12v will power a PCIE SSD ... probably an optane 😁 

there will be spare 5v and 3.3v rails....probably for clock modifications or usb/i2s card

[RickyV]

1 hour ago, RickyV said:

Maybe especially for the cpu because of the enormous current peaks. Just wondering out loud 😀

It could be that that’s the reason why the extreme uses that big capacitor bank, to provide the cpu with a puddle of energy to draw those current peaks from.

[cool_chris]

1 hour ago, Soul Analogue said:

The hardware and PSU part of my EPYC server are almost done

have successfully separated power supply an intel i350 lan card (removing the switching regulators on the card) by 1.0v and 1.8v

the 12v will power a PCIE SSD ... probably an optane 😁 

there will be spare 5v and 3.3v rails....probably for clock modifications or usb/i2s card

 

This looks fantastic. 

Congratulations 👏 

Could you please be so kind to share more about how you did 1 v and 1.8v  ?

We had similar issues whith powering the Buffalo 2016 switch .

It has 1.5v and 1v sps regulators on board .

Some were able to power it up but I am not sure what is the best way of bypassing those noisy regulators.  

I have already Pink Faun clocks installed, 

but wanted to power with LPSs only bypasding this regulators

 

 

[Soul Analogue]

On 10/30/2020 at 7:23 AM, Soul Analogue said:

Good question...

there is no commercially off the shelf choke of this kind of specification available in the market

i have to buy separated parts (EI laminations, mounting, bobbin, etc) myself. Get the bobbin wound by local transformer maker... assemble, tune, test (for noise, both electronically and physically) each individual choke for the target voltages and current... and finally lacquer, before putting into the circuit

Pi filter chokes are technically much less demanding to make... than choke input filter chokes...

Process of chokes stacking

[Soul Analogue]

7 minutes ago, cool_chris said:

 

This looks fantastic. 

Congratulations 👏 

Could you please be so kind to share more about how you did 1 v and 1.8v  ?

We had similar issues whith powering the Buffalo 2016 switch .

It has 1.5v and 1v sps regulators on board .

Some were able to power it up but I am not sure what is the best way of bypassing those noisy regulators.  

I have already Pink Faun clocks installed, 

but wanted to power with LPSs only bypasding this regulators

 

 

Hello Chris

I think i already post in the buffalo thread about my trial on BS2016/A

it was my finding about the current requirement of the three rails on the BS2016

I have my discrete regulator designed to go down to 1.0v and had made it work to power up the buffalo (in the trial)

I will wrap up the buffalo project after this EPYC

 

[RickyV]

1 hour ago, cool_chris said:

It has 1.5v and 1v sps regulators on board .

Some were able to power it up but I am not sure what is the best way of bypassing those noisy regulators.  

I soon have some pictures of how I did that without to much heat in one spot.

[sgouris]

First post in the forum. Just assembled my computer server following the general direction of the first post. 

 

Primarily my post is aimed to those who may be intimidated to start building their server - my recommendation just DO it. The sound quality improvement from stock computers is fascinating. 

 

My build:

HDPlex H5 Chassis -https://hdplex.com/hdplex-h5-fanless-computer-case.html

Mobo: Asus Rog Stryx z390-i (as in the first post)

RAM: Apacer 2x4GB ECC memory (as in the first post)

CPU: Intel i9-9900k (as in the first post)

Storage: 32GB Optane (as in the first post)

Power supply: HDPlex 400W DC-ATX and AC-DC module

No special cabling.

OS: Euphony (as in the first post)

No special network card

USB: Matrix Audio Element-H https://www.matrix-digi.com/en/products/333.html

 

Sound: Sublime, layered, detailed. It is a big step change from the TotalDac own mini-computer and of course any Mac mini etc.

 

Big thanks to Nenon (for the fantastic work in trying out options), HDPlex and Euphony for the truly amazing support!

 

Things that I wish I had know before:

- Commercial cable extensions are just that - extensions so of no value beyond aesthetics. One needs to remove the pins which I did not want to deal with.

- Euphony forces the frequency to max set warming up the CPU - change the max to lower (around 2.7GHz seemed optimum for me)

- Conductonaut Liquid metal thermal paste is electrically conductive - bought but never used ...

- Insert the RAM carefully and securely - if not it is not seen

- To disable the fan check in ASUS bios you need to enable! the fan and change the control 

 

[al2813]

Hi, I just bought a second hand server with what looks a pretty nice HW implementation (see pics below). The server has Linear power supply with a Toroidal transformer (this is the exact reference https://www.toroidal-transformer.com/shop/toroidal-transformers/120va/57/2x-12v-toroidal-tranformer-120va.html). It is currently feeding a Jetway NF9M - N2930 mobo/CPU. I understood from the seller that the Mobo is customized but looking at the layout I do not see any major change compared to the picture in the Jetway website. With a pretty weak CPU and only 8Gb of RAM, I am unable to do Roon oversampling, plus not sure I can run Audiolinux in ramroot with Roon server on. 

 

Can one of the more expert people here determine from the pictures if I can upgrade the Mobo to a socketed one that will allow me a more powerful CPU (and also lore RAM)? Also what makes would be recommended. I understand the Jetway ones are quite good for audiophile setups (I believe a variant of this Mobo was used by Antipodes for their DX), but they are hard to come across in Europe. 

 

 

 

[cool_chris]

4 hours ago, Soul Analogue said:

Hello Chris

I think i already post in the buffalo thread about my trial on BS2016/A

it was my finding about the current requirement of the three rails on the BS2016

I have my discrete regulator designed to go down to 1.0v and had made it work to power up the buffalo (in the trial)

I will wrap up the buffalo project after this EPYC

 

Sounds great !

Could you please share some info of how did you like your sound quality of the Buffalo after " triple bypass" ?.

Inneed yo learn how yo do it or send my Buffalo to expert for the " bypass surgery" 

 

[cool_chris]

3 hours ago, RickyV said:

I soon have some pictures of how I did that without to much heat in one spot.

 

That would be fantastic if you can share it with us. Pictures would be great. 

 

I have number of MP audio modules so can get most of voltages required. 

Just not sure if the 3A max will work 

Also have no idea how and where to solger the 3 different voltages .

If this is easy I am willing to do that myself. 

If not I would be happy to pay for it and will be asking you to do it if possible or ask some local expert for help.

 

Can't wait for your pictures Ricky !