Building a DIY Music Server

[Nenon]

3 minutes ago, ASRMichael said:

@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. 

 

I can't comment for other people, but I can explain why you don't see my system details.

1. It constantly changes / evolves.

2. I rarely use commercial products. Most of my system is custom built components or DIY stuff. No one would really understand exactly what I have in my system as it's not as simple as listing brands / models.

3. I almost always have things in my system I can't really talk about. 

 

I was asked to describe my system recently, and here is my answer. 

On 9/28/2020 at 1:29 PM, Nenon said:

Sure, but it's a little difficult. The only off the shelf components I use is my recently acquired SoundApplication TT-7 power conditioner and my turntable/tonearm/cartridge/minusK stand. Everything else is custom built or DIY. 

 

My server is a DIY version of the Taiko Extreme server with some additions like the JCAT XE card, and with different power supply, chassis, RAM, etc. But it has a lot of similarities. And this is where DC cables make a big difference. My DAC is DIY I have been working on for several years. It's highly tweaked for my system/room/tastes and uses 6 rails of high quality LPS for the digital section and 2 for the analog. It's a NOS DAC with tube-based analog section. Took forever to find the best transformers for I/V conversion and all the other parts that went into that build. It's a one off that would never be seen in production. 

 

Line stage is a custom version of the Don Sach SP14. I just used better chassis and components. 

 

I am bi-amping with four SET monoblocks, some details here:

On 6/9/2020 at 11:29 AM, Nenon said:

1. I completed my tube SET monoblocks a couple of weeks ago. They sound amazing! Even better than they look. The Duelund Cast tinned copper caps are still breaking-in as well as all the Mundorf MTube and MLytic AG capacitors I used but they are 99% there. Add to that, the custom silver wires used throughout, Mundorf silver/gold wire for the power supply, custom transformers, custom chokes, Z-foil resistors, WBT silver connectors, NCF IEC inlets, Grade A Sophia Electric tubes, Grade A Shuguang WE6SN7 PLUS tubes, SR Orange fuses, etc. I used to love my Lundahl output transformers but those beat the shit out of the Lundahls! These monoblocks are something special. They really are a piece of art, and I am very proud of the end result. Sorry for the ugly photo, but that's the best I have at the moment.

 

Expand  

 

Speakers details:

On 6/9/2020 at 2:45 PM, Nenon said:

I built The Loudspeaker (TL-1) by Troels Gravesen about 18 months ago, and that has been my reference since then. Needless to say, I went overboard with my OCD. They also use Duelund tinned copper caps as well as Jantzen Amber-Z Caps, Path Audio resistors (not a fan of the Duelund resistors), special wire, WBT connectors, top quality 1.5'' thick baltic birch, etc.

They have 18'' woofers. The original design was using DSP and ClassD Hypex plate amps for the 18'' woofers, but I am very sensitive on bass details and definition. After a lot of testing, I settled on passive crossovers and bi-amping. Tried quite a few amplification options, including the Jeff Rowland 625 S2 amps as mentioned in one of my previous posts. But when I heard what those AudioMirror monoblocks did, I could not go back to anything else. That was not a self-intuitive test result. Typically one of the weaknesses of tube amplification is deep base. But I guess 95dB efficient 18'' woofers are a different beast. 

So, here I am, bi-amping with two sets of monoblocks (4 monoblocks total)!!! The question is what am I going to do in the summer months with those heaters... it was nice and cozy in the winter :). 

Expand  

 

 

 

However, what I need to link in my profile details is that I recently decided to help Sean Jacobs with his DC4 orders in North America. In order to be in compliance with the forum rules, I have stopped discussing his products. I also need to update my profile to reflect that but we are still finalizing some details. I have made Chris Connaker aware of all this. 

 

I was also working on a couple of products for myself, but if I partner with Sean I could make those available to other people too. 

One of them is a pretty nice and solid audiophile chassis for the Ubiquiti Edge Router SFP with built-in two rails of Sean Jacobs power supply inside. It's a solid chassis with vibration treatment and damping, Gaia feet, and a world class 2-rail LPS, somewhere between a DC3 and a DC4 but closer to a DC3. The idea is to fit the Ubiquiti Edge Router SFP inside - you plug one DC rail to the router and have another DC rail to power up your cable modem or another network device. It's super easy and everyone who can use a screwdriver would be able to do it. It significantly improves the quality of the Edge router because of the better chassis, vibration treatment, and good linear power supply. I am designing this for me. It improves my streaming quality and helps me reduce the number of boxes and power supplies in my system. If other people are interested, I may turn this into a product. Still deciding how involved I want to be in the audio business to be honest.

I am very passionate about DIY audio and the way I look at it is that these projects I am thinking to get involved with would help me sponsor my hobby and do all the experiments that needed decent financing. I have tons of fresh ideas in my head that I would like to try and don't mind sharing any successful stories, so we can all build up on them. 

[Exocer]

35 minutes ago, Nenon said:

The idea is to fit the Ubiquiti Edge Router SFP inside - you plug one DC rail to the router and have another DC rail to power up your cable modem or another network device.

I would be interested in this. It would be nice to obtain more space and replace the Keces P3 powering my Edgerouter X SFP and SB 8200 modem. It would be great if the product offered the DIY option of dropping your router in and building the LPS or a complete product. I would probably choose the former for additional savings if possible as I already own the Edgerouter.

 

Speaking of the Edgerouter X SFP, I saw a post some time back, I believe it may have been @seeteeyou who mentioned specific routers which use either the same or similar Broadcom chip to the Buffalo/Melco switches we like. Do you think a SQ boost can be gained by using a "good sounding" router or am I totally out of line in my thinking? :)

 

Edit: I do not intend to insinuate that the Edgerouter X SFP limits sound quality or has any negative impact on what we hear.

[Nenon]

6 minutes ago, Exocer said:

Speaking of the Edgerouter X SFP, I saw a post some time back, I believe it may have been @seeteeyou who mentioned specific routers which use either the same or similar Broadcom chip to the Buffalo/Melco switches we like. Do you think a SQ boost can be gained by using a "good sounding" router or am I totally out of line in my thinking? :)

 

Possibly. It needs to be tried. However, a switch chip has different functions than a router chip. 

 

Another idea I've had was to try one of my older music servers - use two JCAT Network cards and run a small Linux distribution as a router. It would be a heck of a router. The only issue is that you use the CPU for all the routing functions, while on a router like the Ubiquiti all major routing operations are done by a hardware chip.  It would be a good experiment... if I only had the space for more network equipment :). 

[seeteeyou]

That's Juniper SRX300 with the same BCM53343 (Broadcom Hurricane 2 SoC)

 

https://www.juniper.net/us/en/products-services/security/srx-series/srx300/

https://audiophilestyle.com/forums/topic/30376-a-novel-way-to-massively-improve-the-sq-of-computer-audio-streaming/page/665/?tab=comments#comment-1058809

 

Last year used ones were like about 100 bucks each but now it's much more than that

 

https://www.ebay.com/itm/Juniper-SRX300-Networks-Services-Gateway-Security-Appliance/224174905635

https://forums.servethehome.com/index.php?threads/a-bunch-of-juniper-srx300-firewalls-dumped-cheap.23928/

Quote

Last week there were ~30 auctioned, and another ~30 this week. They all start at $95 and end between $120-140.

 

Though Buffalo did make their own wired router (with a different SoC?)

 

https://product.rakuten.co.jp/product/-/16f909880dac0a27ce97a24c59719b3c/

https://www.amazon.co.jp/バッファロー-VR-S1000-BUFFALO-IPsec対応-VPNルーター/dp/B00H2E882E

 

---

 

At the end of the day we just want something with relatively low power consumption while all on-board regulators could be replaced easily, stuff like RPi could be an interesting option

 

https://www.runeaudio.com/forum/3x-powersupply-for-rpi-t4141-10.html#p18674

https://www.runeaudio.com/forum/3x-powersupply-for-rpi-t4141-50.html#p22497

https://www.runeaudio.com/forum/3x-powersupply-for-rpi-t4141-70.html#p25808

http://www.easyaudiokit.com/bekkan/Raspberry/Raspberry2.html

http://www.headphoneclub.com/thread-694573-1-1.html

http://www.headphoneclub.com/thread-698551-1-1.html

http://www.headphoneclub.com/thread-736184-1-1.html

 

EtherBerry (with separate rails for 5V / 3.3V and its 25 MHz clock could be removed with ease)

http://www.industrialberry.com/etherberry-v1-6/

http://www.industrialberry.com/wp-content/uploads/2017/04/Etherberry_sch.pdf

http://www.industrialberry.com/wp-content/uploads/2017/04/EtherBerry_Code_v1.6.zip

 

OpenWrt

https://github.com/coolsnowwolf/lede

https://downloads.openwrt.org/snapshots/targets/bcm27xx/bcm2711/

https://downloads.openwrt.org/snapshots/targets/scheduled_for_removal/brcm2708/bcm2711/

 

Running OpenWrt in a Docker container

https://forum.openwrt.org/t/running-openwrt-in-a-docker-container/56049

https://github.com/oofnikj/docker-openwrt/blob/gitlab-rpi/docs/rpi.md

https://hub.docker.com/r/minirailgun/openwrt-aarch64

 

Shield for Optane SSD

http://www.suptronics.com/miniPCkits/x872.html

 

Finally the kernel should be patched in order to improve the latency, and then the kernel frequency could also be changed accordingly

 

https://github.com/sam0402/pcp-8820hz/tree/master/xenomai3

[jean-michel6]

Buffalo bs gs 2016 . There are many report of the sq impact of using pf ultra ocxo clock on this switch .

Did anyone tried the same modification with their less expensive ocxo clock ?

[Soul Analogue]

1 hour ago, jean-michel6 said:

Buffalo bs gs 2016 . There are many report of the sq impact of using pf ultra ocxo clock on this switch .

Did anyone tried the same modification with their less expensive ocxo clock ?

I have used Connor winfield ocxo with great result...just note that the buffalo only works with LVCMOS output clocks....CMOS output does not work

[seeteeyou]

It really means that OCXO / LVCMOS options are limited then, most likely only SiTime @ ±5ppb could have the potential to beat Connor-Winfield @ ±10ppb but the lead time could be unpredictable

 

Quote

±5ppb
10-SMD, No Lead
0.354" L x 0.276" W (9.00mm x 7.00mm)
https://www.elementsearch.com/sitime-sit5711ai-kw-33n-25-000000f-oscillators

 

±5ppb
6-SMD, No Lead
0.551" L x 0.354" W (14.00mm x 9.00mm)
https://www.elementsearch.com/sitime-sit5711ai-mw-33n-25-000000f-oscillators

 

±5ppb
4-SMD, No Lead
0.787" L x 0.512" W (20.00mm x 13.00mm)
https://www.elementsearch.com/sitime-sit5711ai-qw-33n-25-000000f-oscillators

 

±5ppb
7-SMD, No Lead
0.984" L x 0.866" W (25.00mm x 22.00mm)
https://www.elementsearch.com/sitime-sit5711ai-tw-33n-25-000000f-oscillators

 

±10ppb
14-DIP Module, 4 Leads
0.800" L x 0.500" W (20.32mm x 12.70mm)
https://www.digikey.com/en/products/detail/connor-winfield/OH4610LF-025-0M/5641634

 

±10ppb
6-SMD, No Lead
0.555" L x 0.358" W (14.10mm x 9.10mm)
https://www.digikey.com/en/products/detail/connor-winfield/DOCSC012F-025-0M/7100177

 

±10ppb
14-SMD, No Lead, 4 Leads
0.800" L x 0.500" W (20.32mm x 12.70mm)
https://www.digikey.com/en/products/detail/abracon-llc/AOC2012XAJC-25-0000C/9998279

 

±10ppb
5-SMD, No Lead
1.000" L x 0.866" W (25.40mm x 22.00mm)
https://www.digikey.com/en/products/detail/raltron-electronics/OX4150A-LZ-1-25-000-3-3-7/10272351

 

±20ppb
6-SMD, No Lead
0.555" L x 0.358" W (14.10mm x 9.10mm)
https://www.digikey.com/en/products/detail/connor-winfield/DOCSC022F-025-0M/5399016

 

±20ppb
6-SMD, No Lead
0.555" L x 0.358" W (14.10mm x 9.10mm)
https://www.digikey.com/en/products/detail/connor-winfield/DOC020F-025.0M/5322238

 

https://audiophilestyle.com/forums/topic/58899-usb-card-xe-xtreme-edition-usb-audio-output-with-ground-breaking-performance/?tab=comments#comment-1055309

On 5/27/2020 at 6:30 PM, Marcin_gps said:

I've just received an email from Digi-key regarding availability of the OCXO oscillator we use on the USB Card XE. The original shipping date when we placed the order was June, 6th. Now the date is June 30, 2020, with an additional, approximate 6 business days transit from the manufacturer to Digi-Key.

 

The SiT5711 is popular and SiTime has not been able to get ahead of the demand. Regardless, we are pushing to get the oscillators sooner, but it seems that the USB Cards XE will not ship sooner than in July.

 

https://audiophilestyle.com/forums/topic/58899-usb-card-xe-xtreme-edition-usb-audio-output-with-ground-breaking-performance/page/2/?tab=comments#comment-1069548

On 7/22/2020 at 7:46 PM, Marcin_gps said:

The manufacturer of the OCXO (SiTime) delayed the delivery again from July 8 to July 30, but after a conference call with them we managed to expedite the delivery to this week. Once we receive the oscillators, we will start the assembly. Hopefully we will be ship the orders by the end of July. 

 

https://www.eenewsanalog.com/news/apples-role-revealed-lossy-sitime-files-ipo

Quote

However, the filing also reveals that SiTime has been making losses. It reports that SiTime's annual revenue went from $101.1 million in 2017 to $85.2 million in 2018, a drop of 15.7 percent. Sales in the first nine months of 2018 were $62.4 million but fell to $56.0 million in the same period this year. That is a fall of 10.2 percent.

 

For 2017 SiTime declared a net profit of $4.7 milion but that turned into a net loss of $9.3 million in 2018 and for the first nine months of 2019 the net loss is $7.2 million.

 

https://www.globenewswire.com/news-release/2020/08/05/2073654/0/en/SiTime-Reports-Second-Quarter-2020-Financial-Results.html

Quote

In the second quarter of 2020 gross margins were $10.0 million, or 46.5% of revenue, operating expenses were $15.3 million, or 71.0% of revenue, GAAP loss from operations was $5.3 million, or 24.5% of revenue and net loss was $5.6 million, or $0.36 per diluted share.

 

[ASRMichael]

@seeteeyou your the best researcher on this forum. I find your posts great! Cheers

[cool_chris]

7 hours ago, jean-michel6 said:

Buffalo bs gs 2016 . There are many report of the sq impact of using pf ultra ocxo clock on this switch .

Did anyone tried the same modification with their less expensive ocxo clock ?

 

I know what you mean. 

I was thinking same way before shipping it for the clock upgrade. 

Did a lot of research and decided to go for the best possible clock. Pink Faun Ultra.

Yes it is very expensive but worth every penny in my opinion. 

Once you turn this thing on you forget the money you paid . It is that good. 

 

Also please keep in mind the numbers you see (ppb, phase noise)  is only the tip of the iceberg when it comes to the clock performance .

There is much more that plays significant role in the final performance. 

 

Nenon reported already that Buffalo + PF ultra clock combo is a touch better than 5000 USD M12 . 

Why is that ? 

I am pretty sure it is related to the clock performance and its supply .

Pink Faun clock is not just a clock.

You get a full board with a power supply and proper grounding .

 

[Darryl R]

Following up on some earlier posts about building with a Z490, I've not been able to get one up and running.  With 3 different mobos -- ASUS ROG Strix Z490-G, -E, and MSI MPG Z490M -- all will not boot into what appears to be a successful install of Ubuntu 20.04 LTS Desktop (which I use to test out builds).  It just keeps looping around to the GRUB menu and displaying Q-CODE F6.  If I use "Try Ubuntu" it seems to work fine.  Since the RAM is the only common denominator with the above machines, I think it's worth changing this out.  I was using Apacer 2 x 4GB UNB DDR4 2666 CL19 WT from an older build.  Can anyone recommend some RAM that worked on a Z490 build?  Thanks.

[bit01]

@Darryl R it could be you need more juice -See PM

[Darryl R]

1 hour ago, bit01 said:

@Darryl R it could be you need more juice -See PM

 

The thought had crossed my mind.  I'm temporarily using a Streacom 150W SMPS.  And for all the boards I was using the i7-10700.  It's very close for the ATX, but should be enough for the microATXs.

[austinpop]

On 10/5/2020 at 8:06 AM, dctom said:

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?

 

Yes, on the same machine. The idea came up as a way to exploit both CPUs, since HQPlayer 4 Desktop seems to not recognize the 2nd CPU. So we're running HQPlayer on one CPU (logical CPUs 0-19), and NAA on the other CPU (logical CPUs 20-39), using Process Lasso to assign CPUs. This boosts SQ noticeably, but may only apply to machines like the Extreme.

[rando]

2 hours ago, austinpop said:

This boosts SQ noticeably, but may only apply to machines like the Extreme.

 

This boosts SQ noticeably compared to what exactly.

 

Roon?

 

TAS?

 

Something else falling further outside the designed implementation?

[Soul Analogue]

Continue working on the full linear atx PSU for the EPYC server. All the rails are choke filtered and regulated by my discrete voltage regulators

 

In this picture there are 4 rails (3.3, 5, 5v standby and 12v)...there will be a few more smaller current rails for add-on cards...

 

[austinpop]

7 minutes ago, rando said:

 

This boosts SQ noticeably compared to what exactly.

 

Let me spell this out for you: 

• config 1: HQP directly connected to DAC via ASIO driver
• config 2:
  • NAA alloted to CPU0, connected to DAC via ASIO driver
  • HQP alloted to CPU1, connected to NAA.

Music playback directly via HQPlayer. No Roon, No TAS.

 

Config 2 sounds better than config 1.

 

7 minutes ago, rando said:

 

Something else falling further outside the design prospective?

 

I have no idea how to decode this sentence.

[seeteeyou]

Let's just borrow this from NASA since it's easy to find on Google, the original SGM Extreme was closer to the following setup

 

https://www.nas.nasa.gov/hecc/support/kb/skylake-processors_550.html

https://www.whatsbestforum.com/threads/taiko-audio-sgm-extreme-the-crème-de-la-crème.27433/page-7#post-557810

Quote

It’s the Asus WS C621E Sage with 2X Silver 4114.

 

---

 

And then here's another one that could be compared to the revised version of SGM Extreme

 

https://www.nas.nasa.gov/hecc/support/kb/cascade-lake-processors_579.html

https://www.whatsbestforum.com/threads/taiko-audio-sgm-extreme-the-crème-de-la-crème.27433/page-165#post-640513

Quote

We switched to the 4210 when it was released which "sounds" identical to the 4114. The 4214 does not have the same midrange contrast levels.

 

---

 

Notice the numbers between those pairs of UltraPath Interconnects as shown above. And then here's what we're getting from Ethernet with its PCIe x1 interface

 

https://audiophilestyle.com/forums/topic/59632-optical-networking-sfps/page/2/?tab=comments#comment-1061654

On 6/19/2020 at 7:33 AM, ray-dude said:

The  StarTech PEX1000SFP2 PCIe optical Network card in my Taiko Audio Extreme

 

Basically Ethernet couldn't even hold a candle, ain't it kinda silly to put HQP and NAA on 2 different boxes while they're connected to each other via Ethernet? Of course we've gotta have the right power supplies for a single box, especially 2 excellent 12V rails (i.e. high current ones) for powering both Xeon Scalable processors with 85W TDP. Obviously it's limited to either PCM upsampling or NOS in this case.

 

---

 

BTW, here's an affordable day to compare the differences between single Xeon versus dual Xeon

 

https://audiophilestyle.com/forums/topic/57525-server-and-endpoint-or-dual-cpu-server/?tab=comments#comment-1080109

[rando]

3 hours ago, austinpop said:

I have no idea how to decode this sentence.

 

Process lasso and everything else is reaching beyond officially recognized, nee designed specifically for at exclusion of all else, programs.  With exclusion of HQP following direct recognition of what Jussi has accomplished.  

 

I was asking how 'roll your own' you are getting with Extreme Server these days.

[dctom]

On 10/7/2020 at 12:18 AM, austinpop said:

 

Yes, on the same machine. The idea came up as a way to exploit both CPUs, since HQPlayer 4 Desktop seems to not recognize the 2nd CPU. So we're running HQPlayer on one CPU (logical CPUs 0-19), and NAA on the other CPU (logical CPUs 20-39), using Process Lasso to assign CPUs. This boosts SQ noticeably, but may only apply to machines like the Extreme.

 

thanks for the explanation

 

[Nenon]

Just read the AMD Ryzen 5000 announcement. Sounds interesting and the better performance might be beneficial for my now older AMD-based Euphony server that I still keep. The best part is, it would be a drop-in CPU replacement after a BIOS upgrade. 

[ASRMichael]

@nenon If only we could get an LPS to power the Threadripper! Series! 128 threads would do!!!!!

[Exocer]

On 10/6/2020 at 1:45 AM, Soul Analogue said:

I have used Connor winfield ocxo with great result...just note that the buffalo only works with LVCMOS output clocks....CMOS output does not work

This sounds interesting. It would be great if you could share some more details of the Connor Winfield clock you've used.

[Exocer]

On 10/9/2020 at 2:15 PM, Exocer said:

This sounds interesting. It would be great if you could share some more details of the Connor Winfield clock you've used.

After a brief chat with Soul Analogue, he confirmed that the following clocks should work with the Buffalo Switches:

 

• https://www.digikey.com/en/products/detail/connor-winfield/OH4610LF-025-0M/5641634
• https://www.digikey.com/en/products/detail/connor-winfield/DOCSC012F-025-0M/7100177
• https://www.digikey.com/en/products/detail/connor-winfield/DOCSC022F-025-0M/5399016

 

I would not know where to begin so I will do some digging and report back when time permits.

 

Brief update on the new Mundorf Silver/Gold "24" pin ATX connector... I have doubled the ground and 5v cables and the system feels a bit more smooth/effortless. Not sure how else to describe it. 

 

[tgb]

Thanks Exocer & SoulAnalogue for feedback vs clocks.

 

The question is : what's important => 

- the stability of the clock ?

to me, the 10ppb of these Connor-Winfield clock is quite close to other clocks below 10ppb (PF is at ?)

- the current to power the clock ? 

Something special in the PinkFaun power supply ? far far better than a MPaudio module (LT3045 based) ?

 

I think I'll have a try with the OH4510LF. I have a MPaudio module to deliver a nice a 3.3VDC... I guess it could work.

 

Thanks SoulAnalogue vs the info / "LVCMOS" output ! It's such a mess to know which output type is the right one...

Rgds

[Exocer]

22 minutes ago, tgb said:

I think I'll have a try with the OH4510LF. I have a MPaudio module to deliver a nice a 3.3VDC... I guess it could work.

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!