NAS Drive Confiquration for Best Sound

[mikeg]

I am currently running a Qnap TS-251 with two red WD drives into via an upgraded Ethernet cable an Ayon Audio S-5 Network Player/Preamp. The NAS is powered by the Uptone Audio JS-2 which made a huge improvement.  Looking to configure a better sounding NAS drive.  I've done some research on the net and have found that a single SSD drive is preferred to two hard drives running in raid 1.  There reviews for the AudiophileOptimizer software if Windows 10 can be loaded on the NAS drive.  I was thinking the fan less Qnap HS-251 with Samsung Evo 850 2TB ssd.  Is it possible to load the Windows 10 operating system on this drive so I could load the AudiophileOptimizer software?  Would a NAS with 8GB of ram such as the Qnap TS-251+ be preferred.  Does more ram result in better sound in theory? Thanks.

[plissken]

The answer to all your questions are no. 

 

The NAS is simply storage medium that will serve up data on demand. The actual playback happens out of buffer on the local machine where the playback application runs.

 

You can start playback, disconnect the NAS, and still hear music. Did your music improve when the Ethernet cable from the NAS was disconnected?

[mikeg]

That couldn't be the furthest from the truth.  Before I went ahead and bought a Qnap TS-251+8GB and a single Samsung 850 Evo 2TB SSD, I spoke with people in the industry (Baetis Audio, Mark Porzilli of LT Memory Player, and the main guy who markets the AudioOptimzer software) and they thought and SSD would perform better.  I have it installed in my stereotimes.com reviewing system now and I can't believe how much better my system sounds.  Much more depth, vocals are more natural, I have better stereo separation, my speakers almost disappear, much cleaner sounding, and images are more holographic. It closer to my buddy's high end analog rig but without the disadvantages of analog.  It's a bigger improvement than adding the Uptone JS-2 power supply but I'm sure both changes are additive.  I was going to redo the crossover in my Acoustic Zen Crescendo with the best parts but now I don't see the need.  This set-up replaced the Qnap TS-251 with 2gb of ram and two 3TB spinning hard drives.  As before, both the Uptone Audio ps and NAS drive are sitting on top of an mechanical isolation platform that also provides RFI/EMI isolation.

[cjf]

On 8/20/2017 at 10:12 AM, mikeg said:

I am currently running a Qnap TS-251 with two red WD drives into via an upgraded Ethernet cable an Ayon Audio S-5 Network Player/Preamp. The NAS is powered by the Uptone Audio JS-2 which made a huge improvement.  Looking to configure a better sounding NAS drive.  I've done some research on the net and have found that a single SSD drive is preferred to two hard drives running in raid 1.  There reviews for the AudiophileOptimizer software if Windows 10 can be loaded on the NAS drive.  I was thinking the fan less Qnap HS-251 with Samsung Evo 850 2TB ssd.  Is it possible to load the Windows 10 operating system on this drive so I could load the AudiophileOptimizer software?  Would a NAS with 8GB of ram such as the Qnap TS-251+ be preferred.  Does more ram result in better sound in theory? Thanks.

I would bet the best sounding setup would be the one using local, to the music server, SSD storage but with that said it wont necessarily be the best idea. I guess it depends if you value SQ over reliability and fault tollerance? If you use the local SSD drive only as a means to cache the file being played and you already store the master file on redundant storage then its a non issue but if your only copy lives on the non redundant local SSD thats a different story.

 

Ill take long term reliability over performance any day, especially in this case where performance needs are low.

[phi1123]

Yes, music is played by the buffer. I would surmise that the system, or computer, would sound better with components that have a lower power demand variation, or an isolated, separate power source. Current demands/use go up,  voltage drops. We AR talking about an electrical system. Biggest difference I have experienced firsthand, however, was using realtime kernel. With only what you want (the music player running) your computer will be focused on getting that bitstream to (I would assume a DAC) and that only. 500 processes running, it is going to put extra layers of demands on your power source and as isolated as each bit of the system IS, every transistor, every chip, capacitor, resistor, IS connected together. Storing your music on a NAS device linked via WiFi eliminates some sharing of power source, assuming your NAS and computer are on separate power sources. Li-po batteries have excellent current regularity. When things are all connected, common ground is often one reason for declined performance.

 

MAYBE the difference in quality between HDs, something I have never personally noticed, is due to the highly fluctuating power demands of some RAID drives and their motor systems. If I was concerned I would connect to my RAID NAS via wi-fi, and/or build a battery source for my RAID drive

 

Audiophile linux is great with rtc. Even debian with totally preemptive realtime kernel is 3 times better than standard as far as I can hear.

Realtime kernel makes a big diff. . You can set a few things if you compile your own.

Think logically, look for options, like clock speed, that may improve performance, compile a few kernels, be patient,

and see for yourself.

 

Or just try ubuntu studio and although their kernel is not fully preemptive, it is far from the standard kernel, and a difference in quality can be noticed.

 

Get your system right, you will have a time machine on your hands. perfect timing is the key along with clean power.

My laptop often sounds better on battery. Realtime kernel is all about timing. I use an Amanero USB to supply the PCM or DSD stream connected via USB, modified version with plenty of decent caps on the board to sufficiently isolate its power. Picked up an old lab supply, to power the DAC I use. (it was inexpensive, so it greatly benefits from the lab supply) I built my mosfet amp. The chassis comes from a digi002 which fate threw at me. the soldering is clean, the source is a good trafo, giant filter caps, the transistors are 2sa1941 (toshiba) and 2sc5200(toshiba). Speakers are fullrange. Paper speakers with whizzer cones. They say NASA on the back. I trimmed the whizzer cones to a few cm depth to get things in phase. Point source compared to a tweeter far from the woofer also makes a great timing diff for nearfield monitors. At a greater distance, not as important.

 

Philip S Bodenheimer.....  (a sci. guy who still trusts his hearing)

[phi1123]

I just saw a commercial for MQA at the top of the page. Bob Stuart is a genius. I did my Physics dissertation in wave mechanics. Deconvolving with a triangle function is genius! Why didn't I think of that? is all I can say. Meridian 206 is the best CD player I ever heard, imo. Shoutout to Bob Stuart.

 

Philip S Bodenheimer

[davide256]

On 8/30/2017 at 7:16 PM, mikeg said:

That couldn't be the furthest from the truth.  Before I went ahead and bought a Qnap TS-251+8GB and a single Samsung 850 Evo 2TB SSD, I spoke with people in the industry (Baetis Audio, Mark Porzilli of LT Memory Player, and the main guy who markets the AudioOptimzer software) and they thought and SSD would perform better.  I have it installed in my stereotimes.com reviewing system now and I can't believe how much better my system sounds.  Much more depth, vocals are more natural, I have better stereo separation, my speakers almost disappear, much cleaner sounding, and images are more holographic. It closer to my buddy's high end analog rig but without the disadvantages of analog.  It's a bigger improvement than adding the Uptone JS-2 power supply but I'm sure both changes are additive.  I was going to redo the crossover in my Acoustic Zen Crescendo with the best parts but now I don't see the need.  This set-up replaced the Qnap TS-251 with 2gb of ram and two 3TB spinning hard drives.  As before, both the Uptone Audio ps and NAS drive are sitting on top of an mechanical isolation platform that also provides RFI/EMI isolation.

You are chasing the least important component in a streaming solution. The only reason your system sounds better is because the OS/applications are now resident

on SSD vs slower HD. Go here for some insight from the Roon community on server build

https://kb.roonlabs.com/Roon_Optimized_Core_Kit

 

 

[jabbr]

On 8/20/2017 at 10:12 AM, mikeg said:

I am currently running a Qnap TS-251 with two red WD drives into via an upgraded Ethernet cable an Ayon Audio S-5 Network Player/Preamp. The NAS is powered by the Uptone Audio JS-2 which made a huge improvement.  Looking to configure a better sounding NAS drive.  I've done some research on the net and have found that a single SSD drive is preferred to two hard drives running in raid 1.  There reviews for the AudiophileOptimizer software if Windows 10 can be loaded on the NAS drive.  I was thinking the fan less Qnap HS-251 with Samsung Evo 850 2TB ssd.  Is it possible to load the Windows 10 operating system on this drive so I could load the AudiophileOptimizer software?  Would a NAS with 8GB of ram such as the Qnap TS-251+ be preferred.  Does more ram result in better sound in theory? Thanks.

 

General answer (I don't know your specific configuration)

 

ZFS likes RAM and caches files to RAM. If you have enough RAM you are playing the files out of memory, not disc.

 

Linux at least, which runs many NAS mostly loads itself into RAM. I know this because I've used network PC renderers with no drive and which boot via iSCSI -- my NAS is a low powered Xeon server which runs Ubuntu Linux (headless) and I've got 64 Gb RAM -- I have terabytes of mirrored ZFS which has been rock solid for >10 years. The iSCSI booted PC still runs even when I reboot the NAS during which time it has no drive whatsoever.

 

Whatever drives you have should be largely irrelevant if you've got everything configured correctly. The OS running on the NAS should not affect SQ at all unless you've got a badly broken configuration, or very suboptimal NAS hardware.

[davide256]

2 hours ago, jabbr said:

 

General answer (I don't know your specific configuration)

 

ZFS likes RAM and caches files to RAM. If you have enough RAM you are playing the files out of memory, not disc.

 

Linux at least, which runs many NAS mostly loads itself into RAM. I know this because I've used network PC renderers with no drive and which boot via iSCSI -- my NAS is a low powered Xeon server which runs Ubuntu Linux (headless) and I've got 64 Gb RAM -- I have terabytes of mirrored ZFS which has been rock solid for >10 years. The iSCSI booted PC still runs even when I reboot the NAS during which time it has no drive whatsoever.

 

Whatever drives you have should be largely irrelevant if you've got everything configured correctly. The OS running on the NAS should not affect SQ at all unless you've got a badly broken configuration, or very suboptimal NAS hardware.

So out of curiosity, how much RAM do you need for  this effect to take place. Is it correct to assume RAM use is faster than M.2 SSD access like the EVO 960 series?

[phi1123]

I have an ssd and a motorized disk hd. Hear no difference between the two. I would imagine if there was a diff, as I said before, current (ampere) demands of the motor would be the cause of that. Pretty far fetched if your DAC is running on its own power source and has three accurate clocks within as it should. Preemptive realtime kernel, however, may account for audible improvement, if you are experiencing jitter issues due to lack of isolation between computer and sound source, or especially if you are powering your DAC or source via USB.

[jabbr]

1 hour ago, davide256 said:

So out of curiosity, how much RAM do you need for  this effect to take place. Is it correct to assume RAM use is faster than M.2 SSD access like the EVO 960 series?

Not a trivial answer but 8-16 Gb should be a good start. ZFS is high performance and can use SSD or 3D XPoint as a second level cache — not that the audio demands are difficult — they aren’t and bandwidth is low. For more demanding applications eg uncompressed 4-8K video or clustering etc., low latency NICs can Direct IO into Xeon cache. So there are layers and layers between the bits and the platter

 

I put an M.2 NVME on my son’s gaming rig

[KingRex]

On 9/1/2018 at 10:43 AM, davide256 said:

You are chasing the least important component in a streaming solution. The only reason your system sounds better is because the OS/applications are now resident

on SSD vs slower HD. Go here for some insight from the Roon community on server build

https://kb.roonlabs.com/Roon_Optimized_Core_Kit

 

 

David, is my interpretation of the Roon ROCK as follows correct.  Build yourself a computer. Or take a server such as my Mojo Audio which is a computer with excellent power supply and vibration control. Pull all existing SSD out and insert new. Take a flash drive with Roon ROCK On It and Boot the machine from the stick.

 

Once loaded the only operating system on this machine will be Roon.

 

Does this setup still allow HQ player via Roon.

 

Does the SQ improve when you operate the machine in Core mode via another machine on the network.

 

Have you done this yourself and found it superior to win server 2016 with AO, HQ via Roon.

 

 

[KingRex]

I am curious about some of these NAS alterations.  I pulled out my Cat6 home wired ethernet and replaced it all with blue jeans. When I went to assess the sound quality between the blue jeans or home network I found the buffer had stored so much information that in the time it took to switch cables the music never stopped. Because of this I heard no difference between the cables from NAS stored music. 

 

 

I also have a Linear Solution switch with a linear power supply in my music system.  Would that be negating any gains I may be getting from rewiring and re-powering my nas. 

 

[audiventory]

On 8/20/2017 at 5:12 PM, mikeg said:

Looking to configure a better sounding NAS drive

 

NAS model, configuration, speed (except extremely-extremely low) don't impact to sound quality.

 

But if you want to see movie, NAS and network speed may be matter. If you see hanging on TV screen.

 

If you want playback simultaneously 50...100 songs of different albums and sound have pauses or clicks, you need to check NAS and network speed too.

[mikeg]

On 9/1/2018 at 12:43 PM, davide256 said:

You are chasing the least important component in a streaming solution. The only reason your system sounds better is because the OS/applications are now resident

on SSD vs slower HD. Go here for some insight from the Roon community on server build

https://kb.roonlabs.com/Roon_Optimized_Core_Kit

 

 

Hi Davide,

 

Roon Core is non resident within the Ayon Audio S-5.  According to the US Distributor for Ayon Audio, this would be a software change that "could" be provided down the road.  The distributor also said that if Roon was loaded on my NAS drive, the Ayon S-5 wouldn't recognize Roon.  They are checking if the new Roon Optimized Core kit will work with my Ayon Audio S-5.  I hope is does but in the meantime, with a high end reviewing system that I have, any upstream improvements are significant.  The combination of the Uptone Audio JS-2 power supply, two Wireworld Silver Eclispe Ethernet cables with the SOtM CAT6 LAN Filter between the two Wireworld Ethernet cables, and a ssd drive installed in the QNAP NAS sounds sublime compared to the original system configuration.  I run my NAS output directly into the Ayon Audio S-5 Network Player/DAC/Preamp.  A router is not present in my system.

[davide256]

2 hours ago, KingRex said:

David, is my interpretation of the Roon ROCK as follows correct.  Build yourself a computer. Or take a server such as my Mojo Audio which is a computer with excellent power supply and vibration control. Pull all existing SSD out and insert new. Take a flash drive with Roon ROCK On It and Boot the machine from the stick.

 

Once loaded the only operating system on this machine will be Roon.

 

Does this setup still allow HQ player via Roon.

 

Does the SQ improve when you operate the machine in Core mode via another machine on the network.

 

Have you done this yourself and found it superior to win server 2016 with AO, HQ via Roon.

 

 

I'm just pointing out the hardware experience documented for Roon server builds

1) more than 8gb RAM  a waste for their core build

2) the disk your OS/application boots from/resides on should be fastest possible. In their case looks like EVO 860 generation of M.2 SSD is what works in tested NUC build, there are faster M.2's now

3) more powerful CPU (i3 ==> i7) becomes necessary as library size increases and oversampling DSP's are used

 

I do plan to follow these for my next server build but it will be a Windows based server to leave media player options open. NUC doesn't have good internal SATA drive storage capability for a growing library.

 

[davide256]

2 hours ago, mikeg said:

Hi Davide,

 

Roon Core is non resident within the Ayon Audio S-5.  According to the US Distributor for Ayon Audio, this would be a software change that "could" be provided down the road.  The distributor also said that if Roon was loaded on my NAS drive, the Ayon S-5 wouldn't recognize Roon.  They are checking if the new Roon Optimized Core kit will work with my Ayon Audio S-5.  I hope is does but in the meantime, with a high end reviewing system that I have, any upstream improvements are significant.  The combination of the Uptone Audio JS-2 power supply, two Wireworld Silver Eclispe Ethernet cables with the SOtM CAT6 LAN Filter between the two Wireworld Ethernet cables, and a ssd drive installed in the QNAP NAS sounds sublime compared to the original system configuration.  I run my NAS output directly into the Ayon Audio S-5 Network Player/DAC/Preamp.  A router is not present in my system.

I played around with HD vs SSD in Synology DS214play NAS. The difference was mainly that of stress on the power supply and external fan noise; SSD is less demanding on power supply and because the thinner drives ventilate easier, the NAS fan runs quieter. But running a good power supply there was no SQ difference, stress on power supply by physical drive motors is mitigated by more capable power supply. Using fiber Ethernet connection to your network player is  helpful, reduces irritants compared to a  wired connection. Still trying to figure out if router/switch CPU processor quality matters for network audio...

 

 

[KingRex]

34 minutes ago, davide256 said:

I'm just pointing out the hardware experience documented for Roon server builds

1) more than 8gb RAM  a waste for their core build

2) the disk your OS/application boots from/resides on should be fastest possible. In their case looks like EVO 860 generation of M.2 SSD is what works in tested NUC build, there are faster M.2's now

3) more powerful CPU (i3 ==> i7) becomes necessary as library size increases and oversampling DSP's are used

 

I do plan to follow these for my next server build but it will be a Windows based server to leave media player options open. NUC doesn't have good internal SATA drive storage capability for a growing library.

 

Why are you saying your next server is windows?   I am reading, but trying to validate my interpretation is there is no windows OS.  It's just a (probably Linux) Roon OS.  Nothing else. Is that what others read.  

 

I spoke with a Roon Rep at a show and he was quick to note the Roon black box server is not "audiophile".  It's for good quality playback across many rooms (kids, living etc) and users all at once. A hub.  That is not to say it won't be audiophile software if installed in a purposed machine with the right linear PS, voltage regulators,  isolation, vibration control etc.   I did not ask that question. 

[davide256]

Just now, KingRex said:

Why are you saying your next server is windows?   I am reading, but trying to validate my interpretation is there is no windows OS.  It's just a (probably Linux) Roon OS.  Nothing else. Is that what others read.  

 

I spoke with a Roon Rep at a show and he was quick to note the Roon black box server is not "audiophile".  It's for good quality playback across many rooms (kids, living etc) and users all at once. A hub.  That is not to say it won't be audiophile software if installed in a purposed machine with the right linear PS, voltage regulators,  isolation, vibration control etc.   I did not ask that question. 

If you want "audiophile" quality off the shelf, you can buy disk less SonicTransporter for ~$800, and it can run HQP and others... a better buy than NUC, similiar i5 architecture. SInce I'm not a Linux educated person, my builds are  WIndows based as is my current Roon server.

[KingRex]

I run a Mojo Audio CAT server into the Mojo Mystique V3 DAC.  I originally for ease had J river and win 7 OS on this server.  With help from Alrainbow I made a partition with duel boot.  I loaded win server 2016, AO, Roon with HQ. The change in performance was nothing less than astounding.  A whole new level with just modifications to software.  Now bettering (in many but not all ways) than my newly upgrades $7k vinyl.

 

Software seems to be very important.  I think there is great potential in Roon OS if all the other noise from processes unrelated to music are eliminated.  However, I have heard others say Linux by design is thin in tone, so who knows.  Windows stripped down with Roon as a program may play more to a listeners taste.  I just wish I was more proficient and had more time to experiment.  I try to find others  who have tried different variations and can give a good Sonic description of what they hear.  

[jabbr]

On 8/20/2017 at 10:12 AM, mikeg said:

I am currently running a Qnap TS-251 with two red WD drives into via an upgraded Ethernet cable an Ayon Audio S-5 Network Player/Preamp. The NAS is powered by the Uptone Audio JS-2 which made a huge improvement.  Looking to configure a better sounding NAS drive.  I've done some research on the net and have found that a single SSD drive is preferred to two hard drives running in raid 1.  There reviews for the AudiophileOptimizer software if Windows 10 can be loaded on the NAS drive.  I was thinking the fan less Qnap HS-251 with Samsung Evo 850 2TB ssd.  Is it possible to load the Windows 10 operating system on this drive so I could load the AudiophileOptimizer software?  Would a NAS with 8GB of ram such as the Qnap TS-251+ be preferred.  Does more ram result in better sound in theory? Thanks.

 

I think you are mixing "NAS" and "music server" together. Try leaving the NAS as just a NAS for file storage, and get a proper server to run software such as HQPlayer or A+ and/or Roon and then optimize the server. One big advantage of Linux over Windows is that its much leaner and will pretty much automatically run in RAM if you have enough RAM.

 

I mean if money is tight, then fine and don't worry about it but with a NAS that's just a NAS you don't have to worry about the "sound" of the discs.

[feelingears]

Apologies for being too lazy to look up the reference, but someone who has pro IT chops over in the Naim forums tested some NASes and found that an NAS with fewer features, fewer disks, and fewer things going on in the background seemingly generated less noise and thus sounded better. I recall it had something to do with Ethernet network timing.

 

I'm paraphrasing here and I'm mentioning it only because it's an intriguing idea if you're already on board with the concept of fewer processes, etc. in an NAS equals better audio performance. Something to consider if you're already going down this path, I suppose.

 

 

[marce]

On 8/31/2017 at 12:16 AM, mikeg said:

  As before, both the Uptone Audio ps and NAS drive are sitting on top of an mechanical isolation platform that also provides RFI/EMI isolation.

It would have to surround the NAS drive to provide rfi/emi isolation.