Jump to content

Tasso

Building the ideal(ish) Music Server

Recommended Posts

1 minute ago, rmpfyf said:

 

L2 means nothing unless you're actually putting the audio on a separate subnet.

 

(Why flat cable?)

Easier to hide.

 

I dont have a layer 2 switch yet but once I started looking at getting one I thought the rest of the setup is probably not much good any way. My intention would be to put audio on a separate subnet.  The other thing I was considering is connecting the roon server and renderer directly but the server doesnt have 2 network ports so which option would be better?

Share this post


Link to post
Share on other sites
2 hours ago, frednork said:

Easier to hide.

 

I dont have a layer 2 switch yet but once I started looking at getting one I thought the rest of the setup is probably not much good any way. My intention would be to put audio on a separate subnet.  The other thing I was considering is connecting the roon server and renderer directly but the server doesnt have 2 network ports so which option would be better?

 Show it loud and proud :D 

 

A direct connection is nice; it’s effectively a subnet of two and less electrical crud. Maybe a separate Ethernet card?

Share this post


Link to post
Share on other sites

Yes, it may be possible with a mini express network card. My server mobo is now a dh77df with an i73770s chip and passive cooling, so no room in this case for a standard card. Are there any that are more preferred than others. 

IMG_20190330_223138.jpg

Share this post


Link to post
Share on other sites


10 hours ago, frednork said:

USB to network adapter no good?

Was just about to suggest one... I'd give it a shot. Windoze or Linux? 

Share this post


Link to post
Share on other sites
11 minutes ago, rmpfyf said:

Was just about to suggest one... I'd give it a shot. Windoze or Linux? 

Still using snakeoil so Linux

Share this post


Link to post
Share on other sites
1 hour ago, frednork said:

Still using snakeoil so Linux

So long as Kith hasn't taken out the driver for whatever USB>Ethernet you get, it should run first time. 

 

Otherwise ask him to put it back in and you're also fine.

Share this post


Link to post
Share on other sites

Edited.

 

 

 

 

Edited by was_a

Share this post


Link to post
Share on other sites


Hi all, great thread! I have been dipping in and out of it with interest as I am currently upgrading my music server. At the moment I have:

 

- Minibox M350 case (https://www.cartft.com/catalog/gl/55) - a TINY mini-ITX case (2.5 litres)

- Intel D945GSEJT (Atom N270) mainboard

- 2gb single-sided RAM

- small internal SSD for OS and music (64GB?)

- Zero-Zone 12v 5a LPSU

 

I was using AES to my DAC via a Lynx AES16, but USB has leapfrogged it in the latest update, so right now I am just using the onboard USB.

 

Now, I had to upgrade to Windows 10 because I need USB Audio 2, and my DAC doesn't like Linux.  With the new OS the Atom cpu+onboard graphics are struggling to say the least.  So, I am revamping. I am getting:

 

- Mitac PD10RI (Celeron J3160) from here https://www.cartft.com/catalog/il/2064. Mitac are the OEM for the Intel DN2800MT and this is a compatible update of this board with a more recent CPU.

- 8gb single-sided RAM (maybe 4 is better?)
- PCIe riser kit for the M350 - specific to this motherboard - (https://www.logicsupply.com/mc-bp-mtx/)

- JCAT femto USB card.

 

The board is still 12vdc input so I can reuse the LPSU. Because of the riser kit I can reuse the case as well.

 

The question I have is about storage. I'll be getting an internal SSD large enough to hold all my music (say 500gb). In due course I'll get a 5v LPSU which can power both the USB card and the SSD. In light of that, would I be better off in terms of SQ:

 

- putting the OS on the same SSD as the music; or

- getting a separate, small, mSATA disk and putting the OS on that?

 

I can see advantages to having the OS on a separate disk. Against that, the mSATA is going to inject noise which I can't mitigate.

 

Any thoughts gratefully received!

Share this post


Link to post
Share on other sites

Do not use an mSATA card with conventional power from the motherboard! Even on idle the injection of noise is hideous in my experience. Rather, use two SSDs with the larger one containing your FLAC files.

Share this post


Link to post
Share on other sites
12 hours ago, was_a said:

Do not use an mSATA card with conventional power from the motherboard! Even on idle the injection of noise is hideous in my experience. 

Second that.

Share this post


Link to post
Share on other sites
On 10/10/2019 at 12:05 PM, was_a said:

Do not use an mSATA card with conventional power from the motherboard! Even on idle the injection of noise is hideous in my experience. Rather, use two SSDs with the larger one containing your FLAC files.

 

On 11/10/2019 at 12:12 AM, rmpfyf said:

Second that.

Thanks, both. For simplicity, I'm using one SSD at the moment and that seems to be sounding pretty good.

 

This PD10RI motherboard isn't great. Much noiser than the D945GSEJT. I guess the DC-DC is inferior. I'm thinking I might try an ATX motherboard with picoPSU instead. Does anyone have a recommendation for a mini-ITX Celeron/Atom with ATX power and a PCIe slot? The reccos on this thread are old enough to be impossible to source.

 

... with this being said, with the JCAT USB card fed external 5v power from a @Gieseler Audio LPSU the sound is a LOT cleaner than my previous server. This is a very high quality setup for the price.

Edited by rhgg2

Share this post


Link to post
Share on other sites


For the last fanless DC server I built, I couldn't get the "200" pico Tasso recommended (that we did the big filter cap mods to); I got a "160" instead. Something like  

 

http://www.mini-box.com/picoPSU-160-XT

 

Previously I'd used Asrock mITX boards J3700 and J3710, but had to get an updated one this time: J4205, still with very low TDP.

 

I use single SSDs - not sure if going double would improve matters.

 

I have 2 pairs of 4GB single sided RAM sticks sitting here if anyone has trouble finding some (Kingston 1600 DDR3). (I went to 8GB single sticks, but it didn't improve the stutter problem as hoped).

 

Interesting choice with the JCat Femto USB card - is this the current favorite out there? I got the Clones Shaar Deluxe, but it seems to have gone from their site, and I'd love to hear if someone's done a comparison. Or with the Pink Faun, which now does DSD apparently. And with some rather costly clock options.

Share this post


Link to post
Share on other sites

 

13 minutes ago, BioBrian said:

For the last fanless DC server I built, I couldn't get the "200" pico Tasso recommended (that we did the big filter cap mods to); I got a "160" instead. Something like  

 

http://www.mini-box.com/picoPSU-160-XT


The other 200w pico that was recommended still seems to be available, so am thinking I might go for that...

https://www.aliexpress.com/item/32836338204.html

 

13 minutes ago, BioBrian said:

Previously I'd used Asrock mITX boards J3700 and J3710, but had to get an updated one this time: J4205, still with very low TDP.

Thanks! will look into that.

 

13 minutes ago, BioBrian said:

Interesting choice with the JCat Femto USB card - is this the current favorite out there? I got the Clones Shaar Deluxe, but it seems to have gone from their site, and I'd love to hear if someone's done a comparison. Or with the Pink Faun, which now does DSD apparently. And with some rather costly clock options.

 

The JCat card more or less came about by chance - someone I knew was selling one at a good price. (In fact, I actually had no plans to upgrade my server until that came up.) Don't know if it's the preferred option, but it does seem to be very good.

 

BTW, I got my 1R ram from MSY:

https://www.msy.com.au/kingston-so-dimm-kvr1333d3s9-kvr13s9s8-4gb-single-ddr3-1333-notebook-memory

 

Share this post


Link to post
Share on other sites

I find with two SSDs a major improvement in sound quality. For operating system a 256GB Samsung 860 Pro (smaller capacity uses less power) and a larger Samsung 860 QVO for storage of audio files. (As an experiment on the storage drive I have a couple of smaller partitions - around 64GB - for critical listening). 

 

 

 

 

Edited by was_a

Share this post


Link to post
Share on other sites
On 30/09/2019 at 10:51 AM, rhgg2 said:

Hi all, great thread! I have been dipping in and out of it with interest as I am currently upgrading my music server. At the moment I have:

 

- Minibox M350 case (https://www.cartft.com/catalog/gl/55) - a TINY mini-ITX case (2.5 litres)

- Intel D945GSEJT (Atom N270) mainboard

- 2gb single-sided RAM

- small internal SSD for OS and music (64GB?)

- Zero-Zone 12v 5a LPSU

 

I was using AES to my DAC via a Lynx AES16, but USB has leapfrogged it in the latest update, so right now I am just using the onboard USB.

 

Now, I had to upgrade to Windows 10 because I need USB Audio 2, and my DAC doesn't like Linux.  With the new OS the Atom cpu+onboard graphics are struggling to say the least.  So, I am revamping. I am getting:

 

- Mitac PD10RI (Celeron J3160) from here https://www.cartft.com/catalog/il/2064. Mitac are the OEM for the Intel DN2800MT and this is a compatible update of this board with a more recent CPU.

- 8gb single-sided RAM (maybe 4 is better?)
- PCIe riser kit for the M350 - specific to this motherboard - (https://www.logicsupply.com/mc-bp-mtx/)

- JCAT femto USB card.

 

The board is still 12vdc input so I can reuse the LPSU. Because of the riser kit I can reuse the case as well.

 

The question I have is about storage. I'll be getting an internal SSD large enough to hold all my music (say 500gb). In due course I'll get a 5v LPSU which can power both the USB card and the SSD. In light of that, would I be better off in terms of SQ:

 

- putting the OS on the same SSD as the music; or

- getting a separate, small, mSATA disk and putting the OS on that?

 

I can see advantages to having the OS on a separate disk. Against that, the mSATA is going to inject noise which I can't mitigate.

 

Any thoughts gratefully received!

 

6 hours ago, rhgg2 said:

 

Thanks, both. For simplicity, I'm using one SSD at the moment and that seems to be sounding pretty good.

 

This PD10RI motherboard isn't great. Much noiser than the D945GSEJT. I guess the DC-DC is inferior. I'm thinking I might try an ATX motherboard with picoPSU instead. Does anyone have a recommendation for a mini-ITX Celeron/Atom with ATX power and a PCIe slot? The reccos on this thread are old enough to be impossible to source.

 

... with this being said, with the JCAT USB card fed external 5v power from a @Gieseler Audio LPSU the sound is a LOT cleaner than my previous server. This is a very high quality setup for the price.

 

As usual getting all isolated/buffered/reclocked downstream of the PC hopefully makes all our PC tinkering irrelevant. 

 

In the meantime :) ...

 

Your DAC should have no feelings about Linux. Do you know what's inside it (as in what the USB>I2S interface actually is)?

 

Major thoughts on motherboards:

  • Minimum on-board devices mean less to power and less injected noise from components doing their own thing
  • You're battling lower power use vs higher number of power phases vs pipeline width (if your software uses it) vs device ordering (especially if you're using onboard graphics and the GPU is time-sliced out of CPU operation, later Core CPU designs use this). My experience has been to avoid Atom CPUs (two CPU power phases make for comparatively noisy rails - look for four minimum), T-series CPUs, no onboard graphics past the 4th gen Intel Core CPUs (though I'd like to try an Intel Xeon D/expensive tho), compile the OS carefully to take advantage (lots of tricks in the last point - not possible with Windows). Run a 'big' CPU slowly, there's more control and less variance over power consumption, and some software will take advantage of the bigger pipelines inherent. Nothing over 35W TDP and hopefully less with modern CPUs. 
  • Board clocks/implementations do make a difference, more so if you're using onboard USB. Some boards have very direct paths, some have very crap phase noise characteristics. All can be improved but this is at the bleeding edge of a**ing about. 
  • Ideally a M2 NVMe on a separate power supply would be ideal from an interrupt/latency management perspective. Maybe a shielded and independently powered PCIe riser? In the meantime just power the SSD's independently, and ideally off the same SATA device. 

Single-sided RAM as per earlier comments.

 

Share this post


Link to post
Share on other sites


Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


  • Recently Browsing   0 members

    No registered users viewing this page.

×
×
  • Create New...