My new 3 watt dead-silent USB Linux music server. - Page 3 - Head-Fi: Covering Headphones, Earphones and Portable Audio

nyc_paramedic · 05-21-2008, 07:42 PM

Originally Posted by Brewmaster

I hear you on the ambient noise issue of having a hard drive in the room. Perhaps a little NAS hard drive located elsewhere in the house. I just love this concept. Having small, low cost computer boards certainly opens up a lot of interesting ideas.

Up until a few days ago I researching low cost and low power NAS solutions, but for my needs I decided against it. My bedroom desktop is usually on most of the day. Why should I add another device that costs money, draws more electricity, and contributes to more clutter?

My desktop machine is a dual-core Athlon X2 4800+ & 2 GB RAM running Debian with a minimal GUI: Openbox window manager and a couple of X terminals. It's more than enough to handle being and NFS server.

But for anyone who does want to throw small and cheap NAS machine in the basement, the New Intel Mini-ITX motherboard with an Atom processor looks very interesting:
Intel D945GCLF Mini-ITX Motherboard

Add a couple of SATA drives and Linux and your good to go. If your Linux challenged then you could load up FreeNAS and control the whole thing from an easy to use web interface.

krrm · 05-21-2008, 09:19 PM

Originally Posted by nyc_paramedic

My desktop machine is a dual-core Athlon X2 4800+ & 2 GB RAM running Debian with a minimal GUI: Openbox window manager and a couple of X terminals. It's more than enough to handle being and NFS server.

As a nfs server for music with a limited number of concurrent users, about anything should do. I used a 800MHz Via C3 for 5 years (updated to Via C7 based thingy last year) for all server tasks. There was hardly any load on the server except when fetchmail popped my email account and spam assassin went to work, but that did not interfer with the music playing. Via C3 was not famous for speed.

As for silent disks please note that 2.5" disk is getting bigger and bigger. I am ready increase the size of my 2.5" disks in raid 1 and throw out the 3.5" storage disk when the capacity increases a little further and prices comes down a bit. When I upgraded my server I expected my current disk solution to last at least 3 years. I don't think I have to wait that long.

ephrank · 05-22-2008, 04:07 AM

Originally Posted by nyc_paramedic

Yes. You are correct. I was oversimplifying for the sake of some of the non-Linux techies here.

So, have u tested if it is bit-perfect?

nyc_paramedic · 05-22-2008, 04:34 AM

Originally Posted by ephrank

So, have u tested if it is bit-perfect?

How does one test for this bit perfectness most of you here so incessantly focus on? I'm not being sarcastic; I'll honestly admit when I don't know something. If you do know more more about the inner workings of ALSA, then please share.

I'll tell you this: This setup sounds much better than my Theta Data Basic transport ever did. And that was a result that I was NOT expecting. I was going the USB computer route out of mere convenience.

From what I understand, when you give ALSA a hardware address for the audio device --in my case "hw: 0,0", then the audio stream is passed along unadulterated.

ephrank · 05-22-2008, 11:22 AM

ALSA has a built-in mixer. Mixer is a OS feature because it enables multiple processes sharing a sound device. If the mixer volume is not set properly the output will not be bit perfect because the mixer changes the amplitude of the signal, results in integer rounding errors and distortion. Usually when the mixer is set to 0dB there is no change to the amplitude and therefore the output is bit perfect.

To test for bit-perfectness one can use a soundcard or audio interface with known bit-perfect capability, e.g. some M Audio or Edirol interfaces. To perform the test simply use a S/PDIF cable to connect the output & the input of the interface together. Start a sound record app, in the linux shell 'cat' a .wav file to the sound card device file and record whats been played to another .wav file. Compare the checksums of the original and recorded .wav files.

Octoploid · 05-22-2008, 01:45 PM

Originally Posted by nyc_paramedic

How does one test for this bit perfectness most of you here so incessantly focus on? I'm not being sarcastic; I'll honestly admit when I don't know something. If you do know more more about the inner workings of ALSA, then please share.

I'll tell you this: This setup sounds much better than my Theta Data Basic transport ever did. And that was a result that I was NOT expecting. I was going the USB computer route out of mere convenience.

From what I understand, when you give ALSA a hardware address for the audio device --in my case "hw: 0,0", then the audio stream is passed along unadulterated.

Yes, hw is bit perfect. Basically you just use the following /etc/asound.conf
file and everything should work for any USB-DAC:

Code:

pcm.!default {
  type hw
  card 0
}
ctl.!default {                                                                                                                                     
  type hw
  card 0
}

With these settings, you could play only one sound at a time and no mixing is used.

nyc_paramedic · 05-22-2008, 04:10 PM

Originally Posted by Octoploid

Yes, hw is bit perfect. Basically you just use the following /etc/asound.conf
file and everything should work for any USB-DAC:
Code:
pcm.!default {
  type hw
  card 0
}
ctl.!default {                                                                                                                                     
  type hw
  card 0
}
With these settings, you could play only one sound at a time and no mixing is used.

I actually don't even have that file in /etc. The aforementioned hardware address is from /etc/mpd./conf; the MDP configuration file.

There is no change in volume when I use the ncurses alsamixer program; with S/PDIF out I'm using I don't think it matters anyway. Only when I use the headphone output of the Trends UD-10 does the mixer work.

eehouse · 06-02-2008, 10:24 PM

The ALIX 3c2 has analog audio-out in addition to USB. Any idea if it's of adequate quality to drive a stereo? It'd be nice to skip the USB DAC if possible (for those of us who don't already own the gear.)

nyc_paramedic · 06-02-2008, 11:26 PM

Originally Posted by eehouse

The ALIX 3c2 has analog audio-out in addition to USB. Any idea if it's of adequate quality to drive a stereo? It'd be nice to skip the USB DAC if possible (for those of us who don't already own the gear.)

The board I have, the 3c2, does not have analog out. The 3c3 does, in addition to VGA. Can't speculate how it sounds, but for the price I'm sure it's not aimed at the Hi-Fi crowd.

TopPop · 07-09-2008, 02:52 PM

Hmmmm... found this post while searching in the forums, and it definitely piqued my interest...

...let's revive it, shall we?

I've been looking into doing something like this, actually, but had "decided" on a system like this, and would get an IDE to CF adaptor to use something like a 2-4 GB CF card as the drive. Planned on installing a stripped version of XP and run Foobar2000 on it, playing files from my external HDD.

However, this option is much cheaper, although I know precisely NOTHING about Linux based OS's, so I'm wondering if this would be over my head. It looks like an attractive option not only for the lower price, but also because Linux and MPD seem to use fewer resources than XP and Foobar. Plus, I'm not sure if Foobar has the same buffering system. I've heard that buffering in Foobar leads to lower sound quality... is this true for MPD?

All I want to do use play files from my external HDD, through a player using ASIO, and send the signal through USB to my Trends UD-10.1 USB transport. No DSPs... no added components... nothing. What do you guys think about this option for me?