Jump to content

Building a DIY Music Server


Nenon

Recommended Posts

21 minutes ago, Dutch said:

ROG Maximum XI Gene

 

I am not referring to it, but I answer anyway. 😁


I use the ROG MAXIMUS XI GENE and it fits perfectly into Streacom FC9 Alpha (Black) - Desktop - micro ATX with HDPLEX 800W DC-ATX. I'm excited. 

Grigg Audio Solutions Owner

StreamFidelitys Setup

Sonus Faber Amati Futura | T + A M10 | T + A SDV 3100 HV | fis Audio PC | HFX RipNAS Solid V4 | GigaWatt PC2-EVO + | Solidsteel HJ-3 / HY-A

Share this post


Link to post
Share on other sites
13 hours ago, Nenon said:

See the part in red. It almost looks like it would not fit. 

 

Also, I like the extra space between the JCAT NIC and PinkFaun card in the ITX motherboard. The JCAT generates a lot of heat (probably EMI too) and would be nice to keep it further apart. 

 

I think it fits. I see the problem in the short distance between JCAT NET and PinkFaun USB. Mutual influences or heat development in the sensitive watch would not be good.

Grigg Audio Solutions Owner

StreamFidelitys Setup

Sonus Faber Amati Futura | T + A M10 | T + A SDV 3100 HV | fis Audio PC | HFX RipNAS Solid V4 | GigaWatt PC2-EVO + | Solidsteel HJ-3 / HY-A

Share this post


Link to post
Share on other sites

Hi @Nenon and others 😉

 

I am thinking about a build with a ROG MAXIMUS XI FORMULA mainboard. Can I connect the 4-pin connector (EATX12V_2) directly to a 12V / 8A Keces P8 or other LPS?

 

spacer.png

Grigg Audio Solutions Owner

StreamFidelitys Setup

Sonus Faber Amati Futura | T + A M10 | T + A SDV 3100 HV | fis Audio PC | HFX RipNAS Solid V4 | GigaWatt PC2-EVO + | Solidsteel HJ-3 / HY-A

Share this post


Link to post
Share on other sites

I started a new Audio PC project with a cable specialist. I would like to thank @Nenon for the many tips in this thread.

 

In order not to hijack this thread I started a new one. If interested: DIY Project High Performance Audio PC with high quality wiring.

Grigg Audio Solutions Owner

StreamFidelitys Setup

Sonus Faber Amati Futura | T + A M10 | T + A SDV 3100 HV | fis Audio PC | HFX RipNAS Solid V4 | GigaWatt PC2-EVO + | Solidsteel HJ-3 / HY-A

Share this post


Link to post
Share on other sites

@Nenon A really well thought out and structured article. Thank you for your thoughts.

 

58 minutes ago, Nenon said:

However, curiosity made me check that, and I was shocked to discover that RAM makes a significant difference.

 

Is there also a software-based solution for this? Possibly: MinorityClean. 

Quote

A utility that cleans up the CPU registers. Targets "General-purpose registers", "MMX registers", and "XMM registers", and matches the internal circuit standards of memory chips.

 

I was very skeptical and feared that I would have to pay a ransom for the encrypted hard drives. 😂

 

But everything is fine and it brings about improved clarity in my system. In jplay forum Pink HQ + Minorityclean you can find a lot of sound descriptions, which I think is mostly exaggerated. The effect is not immediately audible and takes a few hours for me. If I deactivate it, I missed these clear sound structures without sharpening. The tool costs nothing. Just try. But hey! It's not my fault if your hard drives are encrypted or if the Japanese military is listening. 😎

Grigg Audio Solutions Owner

StreamFidelitys Setup

Sonus Faber Amati Futura | T + A M10 | T + A SDV 3100 HV | fis Audio PC | HFX RipNAS Solid V4 | GigaWatt PC2-EVO + | Solidsteel HJ-3 / HY-A

Share this post


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

Did you also have the issue where the copper pipes didn't make it through the entirety of the heatsink of the CPU?

 

No Problem, look:

 

Grigg Audio Solutions Owner

StreamFidelitys Setup

Sonus Faber Amati Futura | T + A M10 | T + A SDV 3100 HV | fis Audio PC | HFX RipNAS Solid V4 | GigaWatt PC2-EVO + | Solidsteel HJ-3 / HY-A

Share this post


Link to post
Share on other sites
41 minutes ago, Dev said:

This leads me to believe that either the s/w and it’s communicating protocol being lossy, incur more latency or the network interface between the server and client is generating a noise pattern which is detrimental to SQ or a combination of both.

 

That's a good point. I think that low latencies are crucial. In a dual-PC system, data traffic must be optimized using Ethernet settings. JCAT gives its NET Card Femto a good manual.

 

There are also different ideas about the dual-PC system. Many says that the computing-intensive tasks (upsampling) are carried out in the 1st PC. The 2nd PC is only used as a low power system for rendering.

 

I'm goes the other way round. The music files and the music management (Roon Server) are in my 1st PC (Control PC). The power consumption is low and the server runs around the clock. It sends the data natively (without upsampling) to the 2nd PC (Audio PC). It runs in high power mode to give the HQPlayer enough computing power for the EC modulators and conversion to DSD. Everything else is switched off, especially the noisy SATA controller. It sounds fantastic and very analog. Just my subjective impressions.

Grigg Audio Solutions Owner

StreamFidelitys Setup

Sonus Faber Amati Futura | T + A M10 | T + A SDV 3100 HV | fis Audio PC | HFX RipNAS Solid V4 | GigaWatt PC2-EVO + | Solidsteel HJ-3 / HY-A

Share this post


Link to post
Share on other sites
42 minutes ago, frederick184 said:

but wondered what the typical bios settings were that the group has found have an effect on sound quality? 

 

There is no typical BIOS setting, as many parameters are important. Is it a fanless design or is it actively cooled? Is overclocking or underclocking required depending on the application? How stable is the power supply?

 

My suggestions should therefore be used with the necessary caution. 😉

 

 

Grigg Audio Solutions Owner

StreamFidelitys Setup

Sonus Faber Amati Futura | T + A M10 | T + A SDV 3100 HV | fis Audio PC | HFX RipNAS Solid V4 | GigaWatt PC2-EVO + | Solidsteel HJ-3 / HY-A

Share this post


Link to post
Share on other sites
7 hours ago, Blackmorec said:

This switch business may be ‘crazy’

 

Yeah it's crazy But maybe it has something to do with the difference between analog and digital reproduction.

 

My theses:

 

1. If I play an analog source, for example from a turntable, I inevitably have losses on the analog route via the amplifier to the loudspeaker. No matter how good the cables are, something is being taken away. Be it the frequencies (skin effect) or the speed. I suspect that if we pass 70% or 80% of the original source on to the loudspeakers in this way, that's a lot.

 

2. With a digital source, however, there is a 100% chance of passing the original source signal on to the DAC. This is due to the bit-perfect forwarding of 0 bit and 1 bit. Based on this, optimization measures should no longer play a role. But they do and in my opinion this has the following reasons:

 

2.1 jitter
We can measure jitter and know how it affects music playback. The human ear is very sensitive to timing errors.

 

2.2 Interfering signals, for example high frequency
There are observations that high frequencies can demodulate the useful signal. Even if repeated packet requests are able to restore the original code due to mismatched checksums, high frequency can interfere with other devices.

 

3.3 Conclusion
In my experience, jitter is minimized when there are several high quality clocks in the signal path. Galvanic isolation, for example by means of glass fiber, completely eliminates high-frequency components. Perhaps this is where the secrets lie, why cascading Switches has very positive effects on the SQ. From the DAC onwards, we naturally have to struggle with the analog transmission of the music again. Ultimately, we will never achieve 100% identical reproduction even with digital sources. But with digital it works a little better. 😉

Grigg Audio Solutions Owner

StreamFidelitys Setup

Sonus Faber Amati Futura | T + A M10 | T + A SDV 3100 HV | fis Audio PC | HFX RipNAS Solid V4 | GigaWatt PC2-EVO + | Solidsteel HJ-3 / HY-A

Share this post


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

Power Supply ( Keces 19V x 2 ..not fully decided)

 

Do you mean Keces 19V x 2 single output?

 

1 x single 19V and 1 x single 12V would be correct. The separate 12V supply of the CPU increases the SQ gigantically. 😉

 

Alternatively 1 x dual 9V / 12V + 18V / 19V. The temperature development could become a problem.

Grigg Audio Solutions Owner

StreamFidelitys Setup

Sonus Faber Amati Futura | T + A M10 | T + A SDV 3100 HV | fis Audio PC | HFX RipNAS Solid V4 | GigaWatt PC2-EVO + | Solidsteel HJ-3 / HY-A

Share this post


Link to post
Share on other sites
30 minutes ago, BCRich said:

It would be two Separate 19v Keces LPS’s powering both inputs of the 800w HDPlex.

 

I had that and it worked fine.

 

31 minutes ago, BCRich said:

I guess I should be Powering the CPU directly

 

Yes! 👍 This is the better option for me. In my system I already had short current peaks of 12V / 7A. So be careful! An i9-9900K wants power, especially if the BIOS has not yet been optimized.

Grigg Audio Solutions Owner

StreamFidelitys Setup

Sonus Faber Amati Futura | T + A M10 | T + A SDV 3100 HV | fis Audio PC | HFX RipNAS Solid V4 | GigaWatt PC2-EVO + | Solidsteel HJ-3 / HY-A

Share this post


Link to post
Share on other sites
5 hours ago, lmitche said:

FYI - Over here there is no better way to suck the life out of the music than to enable turbo mode and multi-threading.

 

That's what I once thought too. However, it depends on what kind of player is being used and whether a conversion requires a lot of processing power. So the HQPlayer's EC modulators for DSD are absolutely great. Hyper Threading enables better utilization of the CPU cores and the cores must be clocked to at least 4GHz. Many roads lead to Rome. 😉
 

Grigg Audio Solutions Owner

StreamFidelitys Setup

Sonus Faber Amati Futura | T + A M10 | T + A SDV 3100 HV | fis Audio PC | HFX RipNAS Solid V4 | GigaWatt PC2-EVO + | Solidsteel HJ-3 / HY-A

Share this post


Link to post
Share on other sites
11 hours ago, GUTB said:

One thing I'd like to understand is what you guys are using as sources and network interfaces for your DACs in this network scenario. PC, QNAP / Synology, etc common network sources? Where does Roon live? What is the network endpoint?

 

I think there is a broad consensus in this forum for the best SQ to use a dual PC system. The system structure differs. First a definition, as I see it:

 

Audio Server : Control PC
In general, all applications and files used for music playback are located in the Control PC. In detail, these are, for example, the music files stored in the NAS ( Network Attached Storage) in the network to play provided. The music needs to be managed, for example sorted by music genre, artist and possibly with information on the musicians. In addition, it must be possible to control the music, including selecting, playing, stopping and possibly regulating the volume. This is made possible, for example, by the Roon software . Other additional functions can be the ripping of CDs, for example with dBpoweramp. Or measurements for room sound correction such as with Acourate . When in doubt, many programs are active. As a rule, there is an Internet connection and therefore a firewall and virus scanner are required. This is bad for trouble-free music playback.

 

Audio End Point : Audio PC
This brings us to the functions of an audio PC, which should have less running processes as possible. Therefore, the Audio PC should only render the music. Then it is up to the user whether the DAC receives the music data natively or with oversampling. In addition, to choice whether the DAC, the data in the format of PCM ( Puls Code Modulation ) or DSD ( Direct Stream Digital) to receive. All of these functions can be performed, for example, with the Roon software. Anyone who values the very highest sound quality use ofr example HQPlayer. The internet connection should be blocked. The firewall and virus protection are no longer required.

 

We come to the implementation of a dual PC system. Without claiming to be complete, a list:

 

1. Dual PC solution with maximum computing power in the Control PC / Audio Server
With this solution, the Control PC would do the main work. This would have to be computationally powerful, as it might also have to calculate the oversampling, the conversion of, for example, PCM to DSD and the convolution for the room correction in realtime. The Audio PC would only render the data natively. The advantage lies in the Audio PC, which can be operated with very little computing power. The disadvantage lies in the increase in applications in the Control PC, which, for example, would have to transport a very large amount of data to the audio PC in the case of oversampling.

 

2. Dual PC solution with the highest computing power in the Audio PC / Audio End Point
The Control PC could manage with a relatively lean computing power, since the computationally intensive tasks such as oversampling, converting, for example, PCM to DSD and the convolution for room correction in the Audio PC would take place. The tasks of a Control PC could, for example, be carried out with older computers. In the Control PC only the music control and the "native" transfer of the data to the audio PC take place. The advantage lies in the Audio PC, which prepares the music data for the DAC in the highest quality undisturbed by other applications. Another advantage is the permanent low-energy operation in the Control PC.

 

In the example below, solution 2. was implemented. The Control PC runs as a music server (NAS) 24 hours a day and consumes little electricity due to its low computing power. The Audio PC is switched on for music operation and this renders the data in the best resolution and sends them to the DAC without interference.

 

fis-Audio-PC-Musicplayer-Renderer.png?w=

 

3. Dual PC system in one chassis

Taiko Audio Extrem uses a mainboard with a dual CPU. I find a very elegant solution. Different tasks can be assigned to the two cores. The DIY community has not really made any progress with the replica. Nenon has created a prototype.

 

The two follow-up projects are not making any progress. 

 

Sorry for the long text in bad English.

Grigg Audio Solutions Owner

StreamFidelitys Setup

Sonus Faber Amati Futura | T + A M10 | T + A SDV 3100 HV | fis Audio PC | HFX RipNAS Solid V4 | GigaWatt PC2-EVO + | Solidsteel HJ-3 / HY-A

Share this post


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

I'm looking to reduce all audio latency

 

I also consider low latencies to be a key to increasing SQ. The setting in the ASIO USB driver depends on the sample rate. With DSD 256, 32 samples can be specified as the lowest buffer. Then it is 1.18 ms input latency and 1.50 ms output latency for USB.

 

With LatencyMon the process latency can be measured. I have measured the sensationally low value of 12.20 µs = 0.0122 ms since using the XILINX Solarflare Flareon Ultra SFN8522 PCIe card.

 

40082845xm.png

 

My tips:
- Check the process latency with every change. Disturbing drivers are displayed with LatencyMon. A driver update usually helps.
- Select and configure everything that transports data according to low latencies (CPU, USB, Ethernet, RAM, SSD, etc.)
- Disable unnecessary processes

Grigg Audio Solutions Owner

StreamFidelitys Setup

Sonus Faber Amati Futura | T + A M10 | T + A SDV 3100 HV | fis Audio PC | HFX RipNAS Solid V4 | GigaWatt PC2-EVO + | Solidsteel HJ-3 / HY-A

Share this post


Link to post
Share on other sites
19 minutes ago, ASRMichael said:

Apacer ECC ram is here

 

model - D31.23185S.001

 

Thanks for link. 

I don't know if it means anything. The RAM latency is rather average: CL19 / 2666 MHz = Calculated nanoseconds 14.25.

I use CL10 / 2400 MHz = Calculated nanoseconds 8.33.

Grigg Audio Solutions Owner

StreamFidelitys Setup

Sonus Faber Amati Futura | T + A M10 | T + A SDV 3100 HV | fis Audio PC | HFX RipNAS Solid V4 | GigaWatt PC2-EVO + | Solidsteel HJ-3 / HY-A

Share this post


Link to post
Share on other sites
4 minutes ago, MarcelNL said:

can you post a link perhaps?

 

Yes of course. But it is not an ECC RAM. I attach great importance to cooling, because I run a fanless PC and the lowest latencies.

 

DOMINATOR® PLATINUM 16GB (2 x 8GB) DDR4 DRAM 2400MHz C10 memory kit

Grigg Audio Solutions Owner

StreamFidelitys Setup

Sonus Faber Amati Futura | T + A M10 | T + A SDV 3100 HV | fis Audio PC | HFX RipNAS Solid V4 | GigaWatt PC2-EVO + | Solidsteel HJ-3 / HY-A

Share this post


Link to post
Share on other sites
6 minutes ago, Exocer said:

I'm not sure if you've already covered this,but, have you had a chance to compare the Xilinx to an audiophile network card? What were the tonal differences if any?

I compared it to the JCAT NET Femto Card.

 

Sound description XILINX Solarflare Flareon Ultra SFN8522 PCIe card:

 

The sound is just addicting. A sensational speed. Flamenco or drums are a real delight. This clarity and accuracy of the notes are the best I have heard so far. The separation of instruments is outstanding, combined with an excellent spatiality. And the whole thing garnished with a beautiful flow of music without any sharpness. Despite the high resolution, voices have a nice melt and the bass is physical and precise.

Grigg Audio Solutions Owner

StreamFidelitys Setup

Sonus Faber Amati Futura | T + A M10 | T + A SDV 3100 HV | fis Audio PC | HFX RipNAS Solid V4 | GigaWatt PC2-EVO + | Solidsteel HJ-3 / HY-A

Share this post


Link to post
Share on other sites
17 hours ago, Nenon said:

What switch are you connecting it to? And what transceivers are you using? 

 

I use the Startech 1000BASE-EX SFP -SM LC-40 KM transceiver for this.

 

The connection is:

Router -> 11GTek (10GTek with Sellarz Clock) -> XILINX Solarflare Flareon Ultra SFN8522 

 

For more details: 

 

Grigg Audio Solutions Owner

StreamFidelitys Setup

Sonus Faber Amati Futura | T + A M10 | T + A SDV 3100 HV | fis Audio PC | HFX RipNAS Solid V4 | GigaWatt PC2-EVO + | Solidsteel HJ-3 / HY-A

Share this post


Link to post
Share on other sites


×
×
  • Create New...