SQ comparison: roll-your-own vs. commercial servers

[CleverName]

As someone new to computer audio (but very familiar with building custom PCs), are there significant sound quality differences in a well-built dedicated PC vs. the majority of commercially sold music servers? What are the pros and cons?

[Audio_ELF]

With a few exceptions most commercially sold music servers are well-build dedicated PCs.

 

Exceptions would include devices such as Aurrender, Auralic Aries and devices such as Linn's DS/DSM ranges. The advantage if these tend to relate to plug and play ease of use / setup rather than specifically better sound quality (though sound quality may be better).

[ronalde]

It's really hard (for a consumer) to perform thorough blind tests with different computers/servers because one should have an equal number of equal DAC's and be able to use the same network, files etc. Furthermore, a broad comparisson of custom vs commercial computers seems nearly impossible because even changing a motherboard in your custom computer can change its sound (quality?).

 

In general, I noted when building my custom sound PC's, that the less load is on the system (CPU, RAM, pheripherals and busses), the better the sound quality. My current system therefore is fanless, uses a intel core i5, uses but 2GB RAM, has no local storage apart from a small (16GB) msata drive for loading the linux kernel image (initrd) and after that runs fully in RAM (OS+mpd in bit-perfect mode) and has no pheripherals apart from the USB-DAC and the network. It sounds better than anything I've built before, but my marks above do apply of course.

 

I don't think (but haven't listended them) that all-in-one NAS-ripper-player "servers" can achieve such results. I perform ripping and file and metadata management on my desktop PC and my music files are stored on the NAS (using NFS).

 

Good luck,

Ronald

[tesewinou]

It sounds better than anything I've built before, but my marks above do apply of course.

[k6davis]

It's really hard (for a consumer) to perform thorough blind tests with different computers/servers...

 

Agreed.

 

I achieve great results with a server that I built myself. I use Windows Server 2012 and the Audiophile Optimizer (both of which you can read lots more about on this site). Beyond that, I just choose my RAM, processor and SSD's carefully and added Paul Pang's USB card. Others have gone much further with power supply and inner cabling optimization, etc.

 

The prices that I see for the turnkey audio servers makes them a non-starter for me. The appeal seems to be to those who don't want to build a computer. I have much greater flexibility and a much lower cost building my own.

[CleverName]

Is there some advantage to storing the music files on NAS vs. on USB drive, say attached to the PC with a quality USB cable?

 

Also, as I am planning this system, is the general idea that the further you can separate the components (PSU from HDD/SDD from etc.) the better?

[Pepsican]

Is there some advantage to storing the music files on NAS vs. on USB drive, say attached to the PC with a quality USB cable?

 

Also, as I am planning this system, is the general idea that the further you can separate the components (PSU from HDD/SDD from etc.) the better?

 

A NAS has some advantages over a USB stick. First, from a file management and sharing perspective it is easier. From a soundquality perspective, it can possibly have a tiny effect. The USB stick on the port will use power, which will cause a tiny spike in RF noise.

 

The big issue with a pc compared to audio equipement is that there is alsmost no provision to counter ripple noise and RF noise. And noise tends to be cummulative and you need to fight it throughout the signal path. You cannot just let it rage on and then count on one component to eliminate it all at the end. So, it starts with the signal entering the computer. The one is the network cable, the other is the electricity cable. You need to make provisions there, like lineair power supplies, routers (routers reclock the signal) and insulated cables (Cat7). The second point are the PC internals. So, get a minimalist motherboard (B or Q chipsets rather than Z), ensure wires and connections are as short as possible. Insulate where you can and put filters in where possible. This is where things like PCIe cards (shortening connections as PCIe is northbridge while USB is southbridge), SATA filters (Rf noise) and special SATA and USB cables (insulation) come into play. Then you get to the signal out. It depends on what you use, but the cable matters for sure. And if you use USB ensure your PCIe card gets a clean power supply or use something else that cleans up the power (like iFi). Even when you DAC does not use the power from the USB, the rf noise is there and has a (minimal) impact on the signal.

 

Based on what one reads, the PCIe USB cards and linear power supplies, together with a decent USB cable (if you use USB) have the biggest impact.

[CleverName]

Thanks Pepsican. I currently have a docked netbook running W8 feeding a Peachtree Audio DAC with a carbon USB cable and I am driving my main speakers with a digital integrated amp and headphones with a dedicated headphone amp. But, as you can see, I am planning a dedicated PC build using all high-quality components.

 

Your information is helpful and very much appreciated.

 

Oh, I will add another comment/question, now. And I apologize to all if I am belaboring any of these points. I am someone who wants to gather and learn as much as possible (and yes, I AM reading the forum, but information changes all the time as our technology and learning improve over time).

 

In referring to my use of a USB drive, I was not clear. I am not using a USB stick, it is a 1tb external USB drive. This drive is powered by the computer's USB port, so your advice that this adds noise applies to my situation. That said, I can see that I may now want to continue to use this device in my build and will be switching to an SSD.

 

Aside from shielded SATA cables and filters are there additional shielding devices/methods? Because this brings me back to my original question about separation of components.

 

I see a lot of discussion and merit to shielded cables, filters and cleaners, but I also note that many are building dedicated PCs that are very, very small. Wouldn't tightly packing the components (SSD - even for the OS, PCI cards, etc.) into a very tight space introduce Rf bleed? I am sorry if this is a dumb question, but I am not an engineer. If the idea is to isolate Rf noise, it seems counter intuitive to me that shortening distances between components could/would be ideal, as it has the potential to introduce interference? Or am I wrong about this?

[CleverName]

That said, I can see that I may now want to continue to use this device in my build and will be switching to an SSD.

 

Sorry, that should have said "NOT" continue use of the USB powered HDD.

[Geardaddy]

As someone new to computer audio (but very familiar with building custom PCs), are there significant sound quality differences in a well-built dedicated PC vs. the majority of commercially sold music servers? What are the pros and cons?

 

Rolling your own? Is that choice of words a subconscious reference to your past?

 

In terms of technical particulars, Pepsican laid things out nicely. Due to inherent limitations in A+Bing this stuff, the puter is still a little bit of a black box in terms of what teaks have the most impactful effects on SQ. I am afraid a lot of our precious DIY tweaks would not be noticeable to a blinded lay listener.

 

Other than the obvious stuff (SS drive, software optimization, linear supply, etc), noise isolation through internal shielding is one thing I have long been curious about and read little on. Computers are a theoretical rats nest of electromagnetic radiation, etc. How does one effectively manage that? Also, people do claim to hear differences between networked (ethernet cable) versus USB for file storage.

 

Another area to consider is your player software player. HQplayer and XXXHighend are two to consider. Peter who designed the Phasure dac has been at the forefront of computer audio for some time and does not get props for that. He is by far the most sophisticated I have read when it comes to this stuff and he is worth following regarding computer matters.

 

Finally, if you are lazy like me, streaming (Aries) devices are a good option. You can also consider external devices (Empirical Audio Offramp) or other USB converter (Audiophileo) to improve SQ. That involves more $, but a lot of people swear by such things....

[amir57bs]

It's really hard (for a consumer) to perform thorough blind tests with different computers/servers because one should have an equal number of equal DAC's and be able to use the same network, files etc. Furthermore, a broad comparisson of custom vs commercial computers seems nearly impossible because even changing a motherboard in your custom computer can change its sound (quality?).

 

In general, I noted when building my custom sound PC's, that the less load is on the system (CPU, RAM, pheripherals and busses), the better the sound quality. My current system therefore is fanless, uses a intel core i5, uses but 2GB RAM, has no local storage apart from a small (16GB) msata drive for loading the linux kernel image (initrd) and after that runs fully in RAM (OS+mpd in bit-perfect mode) and has no pheripherals apart from the USB-DAC and the network. It sounds better than anything I've built before, but my marks above do apply of course.

 

I don't think (but haven't listended them) that all-in-one NAS-ripper-player "servers" can achieve such results. I perform ripping and file and metadata management on my desktop PC and my music files are stored on the NAS (using NFS).

 

Good luck,

Ronald

 

what is your ripping software?

what is file format? wave , aiff , flac ,...

[zackthedog]

Thanks Pepsican. I currently have a docked netbook running W8 feeding a Peachtree Audio DAC with a carbon USB cable and I am driving my main speakers with a digital integrated amp and headphones with a dedicated headphone amp. But, as you can see, I am planning a dedicated PC build using all high-quality components.

 

Your information is helpful and very much appreciated.

 

Oh, I will add another comment/question, now. And I apologize to all if I am belaboring any of these points. I am someone who wants to gather and learn as much as possible (and yes, I AM reading the forum, but information changes all the time as our technology and learning improve over time).

 

In referring to my use of a USB drive, I was not clear. I am not using a USB stick, it is a 1tb external USB drive. This drive is powered by the computer's USB port, so your advice that this adds noise applies to my situation. That said, I can see that I may now want to continue to use this device in my build and will be switching to an SSD.

 

Aside from shielded SATA cables and filters are there additional shielding devices/methods? Because this brings me back to my original question about separation of components.

 

I see a lot of discussion and merit to shielded cables, filters and cleaners, but I also note that many are building dedicated PCs that are very, very small. Wouldn't tightly packing the components (SSD - even for the OS, PCI cards, etc.) into a very tight space introduce Rf bleed? I am sorry if this is a dumb question, but I am not an engineer. If the idea is to isolate Rf noise, it seems counter intuitive to me that shortening distances between components could/would be ideal, as it has the potential to introduce interference? Or am I wrong about this?

 

No need for apologies, we're all learning here. :-) And searching this huge forum for specific answers can be a pain.

 

There are several things that argue against storing your music on a USB drive.

 

1) You're possibly adding noise to the USB circuitry.

 

2) You've got incoming and outgoing packets on the same chip, possibly degrading the signal to your DAC.

 

3) A system crash (or even an update!) can corrupt that drive in a heartbeat and leave your music collection in limbo unless you're able to resurrect the drive and salvage your files.

 

You're much better off storing your music on a NAS and *backing it up* to a detachable, self-powered USB storage device that's stored safely somewhere. A NAS is not expensive and it allows access to your music from anywhere on your network. Generally speaking, your collection is safely isolated from most disasters.

 

Most well-designed playback software nowadays either loads the files into memory or uses enough buffering to avoid any playback problems over your network.

 

As for building your own computer... I am not an expert but I think that unless you really understand what components go into a truly high-end audio PC, you're asking for a lot of time and effort with possibly not the results you want. The folks who design these things do rigorous testing to determine which motherboards, chipsets, power supplies and other components really offer the lowest noise and cleanest signal transfer.

 

Personally, I think your best bet would be a fanless mini-PC with an Intel i5 or better, preferably quad core, with at least 8GB of memory, and running Windows 10. This is enough to accomodate even the most demanding playback software around today, including those that do complex upsampling. And you can easily control it from a laptop or pad. Look at a Gigabyte or Zotac. But I'm sure others will chime in. And you could look at the CAPS DIY articles here on the forum.

 

Software is another matter. The basic standard, IMO, is JRiver does just about everything one could want, and the sound is better than decent. XXHighEnd is the best I have heard--nothing gets the top of the piano quite as right--but it's complicated and you have to be a fearless experimenter. But software is easy and not terribly expensive to experiment with.

 

I hope this helps.

[plissken]

 

 

In referring to my use of a USB drive, I was not clear. I am not using a USB stick, it is a 1tb external USB drive. This drive is powered by the computer's USB port, so your advice that this adds noise applies to my situation. That said, I can see that I may now want to continue to use this device in my build and will be switching to an SSD.

 

Would this be useful to you?

 

I pipe the output of my DAC into my ADC and capture the tracks. I'll do a few tracks from an attached USB Drive (Flash or SSD let me know) with the DAC on the same USB bus.

 

Next I'll capture output over Wireless Ethernet and post the files. You can listen at your leisure with out the burden of knowing which is which. You can draw your own conclusions and I'll let you know afterward which was which.

[audiventory]

As someone new to computer audio (but very familiar with building custom PCs), are there significant sound quality differences in a well-built dedicated PC vs. the majority of commercially sold music servers? What are the pros and cons?

 

Main difference may be in noise level at analog output of connected DAC.

 

I prefer build own PC. I know how manage the noise. Of course, need measurement equipment.

 

I done ten years ago custom software for music management and managed by remote control.

 

It is hobby. I very like it. Though now I have no time for it.

 

Otherwise, ready decission give nice looked design of hardware unit and management software. It is integrated ready-to-use system - connect and play.

 

If music server provide bit-perfect playback, then no reasons to bothering about sound quality of server as itself.

 

Anyway quality may be checked by test signals, if we can capture digital audio stream at DAC output.

 

Resume:

 

1. If you like build PC, programming and have time for it, you can build own media server.

 

2. If you want professionally assembled system with remote control and other integration features - buy a commercial server.

 

3. Sound quality of DYI unit depend on used components, measurement tools and your skills.

[zackthedog]

No need for apologies, we're all learning here. :-) And searching this huge forum for specific answers can be a pain.

 

There are several things that argue against storing your music on a USB drive.

 

1) You're possibly adding noise to the USB circuitry.

 

2) You've got incoming and outgoing packets on the same chip, possibly degrading the signal to your DAC.

 

3) A system crash (or even an update!) can corrupt that drive in a heartbeat and leave your music collection in limbo unless you're able to resurrect the drive and salvage your files.

 

You're much better off storing your music on a NAS and *backing it up* to a detachable, self-powered USB storage device that's stored safely somewhere. A NAS is not expensive and it allows access to your music from anywhere on your network. Generally speaking, your collection is safely isolated from most disasters.

 

Most well-designed playback software nowadays either loads the files into memory or uses enough buffering to avoid any playback problems over your network.

 

As for building your own computer... I am not an expert but I think that unless you really understand what components go into a truly high-end audio PC, you're asking for a lot of time and effort with possibly not the results you want. The folks who design these things do rigorous testing to determine which motherboards, chipsets, power supplies and other components really offer the lowest noise and cleanest signal transfer.

 

Personally, I think your best bet would be a fanless mini-PC with an Intel i5 or better, preferably quad core, with at least 8GB of memory, and running Windows 10. This is enough to accomodate even the most demanding playback software around today, including those that do complex upsampling. And you can easily control it from a laptop or pad. Look at a Gigabyte or Zotac. But I'm sure others will chime in. And you could look at the CAPS DIY articles here on the forum.

 

Software is another matter. The basic standard, IMO, is JRiver does just about everything one could want, and the sound is better than decent. XXHighEnd is the best I have heard--nothing gets the top of the piano quite as right--but it's complicated and you have to be a fearless experimenter. But software is easy and not terribly expensive to experiment with.

 

I hope this helps.

 

I owe the OP an apology--I missed the part where he says he's familiar with building custom PCs! Please ignore my comments about "knowing what you're doing"! :-)

[Sal1950]

Irrelevant of the hardware used, if any two servers are delivering the same bit perfect data steam from a music file to your DAC, they will sound identical.

[audiventory]

Irrelevant of the hardware used, if any two servers are delivering the same bit perfect data steam from a music file to your DAC, they will sound identical.

 

There may be different noise by wires and air (lesser significant than by wires) due different hardware. For careful implementation need check.

 

I don't say about jitter, about noise floor only. Because it is not matter, because, as rule, DAC work by own buffer with internal clocking.

[Pepsican]

Irrelevant of the hardware used, if any two servers are delivering the same bit perfect data steam from a music file to your DAC, they will sound identical.

Not true. A digital sound signal consists of two components: the bits and the timing (which is why clocks impact sound quality). When something is bit perfect, you have not lost any digital information, but your timing may differ. One way this shows up is in the form of jitter.

 

Bits are definetely bits. But that is only half of the digital audio signal equation.

 

And then we haven't discussed yet how that digital signal converts to analogue before your ears can listen to it....

[audiventory]

Not true. A digital sound signal consists of two components: the bits and the timing (which is why clocks impact sound quality).

 

1. Most cases jitter is not matter, because DAC have own buffer and own clock.

 

2. Jitter may impact if SPDIF or other digital-interface clock used as clock for DAC. But it is very rare case. DAC may have switch between mode #1 and #2.

[Sal1950]

Not true. A digital sound signal consists of two components: the bits and the timing (which is why clocks impact sound quality). When something is bit perfect, you have not lost any digital information, but your timing may differ. One way this shows up is in the form of jitter.

Bits are definetely bits. But that is only half of the digital audio signal equation.

And then we haven't discussed yet how that digital signal converts to analogue before your ears can listen to it....

True, but we're talking the servers. Proper handling of timing (jitter) should be done at the DAC.

In any case I was referring in the main to all the silliness and malarkey that gets expended (everything matters) on digital sound transmission.

[plissken]

Not true. A digital sound signal consists of two components: the bits and the timing (which is why clocks impact sound quality). When something is bit perfect, you have not lost any digital information, but your timing may differ. One way this shows up is in the form of jitter.

 

Bits are definetely bits. But that is only half of the digital audio signal equation.

 

And then we haven't discussed yet how that digital signal converts to analogue before your ears can listen to it....

 

USB DACs for about the past 10+ years have been Asynchronous. Which means there is no timing of the audio on the USB bus. The clock is applied by the DAC.

[zackthedog]

Irrelevant of the hardware used, if any two servers are delivering the same bit perfect data steam from a music file to your DAC, they will sound identical.

 

That may be true, but with the many different software options available now, it's important to consider what kind of computer and OS you want to invest in. Many of the high-end playback apps have different ways of handling the data, different methods of upsampling, and demand different levels of CPU performance and memory. And I don't care to get into a big argument about "bits are bits," but in my experience a Raspberry Pi, for instance, simply doesn't sound as good as a well-made PC, even when they are delivering the same "bits."

[Sal1950]

And I don't care to get into a big argument about "bits are bits," but in my experience a Raspberry Pi, for instance, simply doesn't sound as good as a well-made PC, even when they are delivering the same "bits."

 

Of course you are entitled to a opinion, but to present it as a fact requires supportable evidence for that claim. Something I doubt can be supplied..

Here's a snipet of the required system load to deliver the Linn 24/196 test file to my DAC using Clementine's bit perfect stream. An easy-peasy task.

[zackthedog]

Of course you are entitled to a opinion, but to present it as a fact requires supportable evidence for that claim. Something I doubt can be supplied..

Here's a snipet of the required system load to deliver the Linn 24/196 test file to my DAC using Clementine's bit perfect stream. An easy-peasy task.

[ATTACH=CONFIG]34024[/ATTACH] [ATTACH=CONFIG]34025[/ATTACH] [ATTACH=CONFIG]34026[/ATTACH]

 

I agree that it doesn't necessarily take a lot of CPU or memory to deliver bit-perfect audio.

[audiventory]

USB DACs for about the past 10+ years have been Asynchronous. Which means there is no timing of the audio on the USB bus. The clock is applied by the DAC.

 

I don't know history of USB DAC. But computer audio devices has own clock always - since 1990s. Some audio interfaces with digital input have additional external sinchronization mode. User manually select between these modes.