Dedicated Music server vs Mac mini on a Devialet via Ethernet (AIR) connection

Roon and Tidal are working fine now via AIR (EtH) in a Devialet 200. My only digital source is a Mac mini 2012/16GB RAM/ SSD - AIR (eth) direct in. My speakers: Raidho X1.

My concern is that the Mac Mini is not optimized for audio so I wonder what upgrade path to take.

Either I tweak the MM with software and new LPS. But I wonder if that's worth it as I'm not using the USB input on the DEVialet. USB is known for its electric interference. But is not Ethernet a lot cleaner?? a proper Paul Hynes SR5 would cost about €700,-?

The other route is going for a standalone server like Innuos or Sonictransporter i5. (€700-€900) Would that offer a significant improvement on SQ over Ethernet?
If so, what switch is recommended? they also tend to perform better with a better PSU!?

Or just go for a cheap NUC? (€100-200)

Requirements: able to handle Roon core.

My goal is to improve SQ on a budget, but improvement should be significant as in better transparency/separation.
What upgrade path would you choose??

I doubt if tweaks to the Mini will have any effect.  I have a maxxed-out Mini for USB, but I don't believe it has any effect when I play through Ethernet.

If you want an effect, put a fast responding LPS with earth ground connected to DC common on your Router and then wire direct to the DAC from the Router.

There are ways to do this with WIFI also, but it's a trade secret.

Steve N.

Empirical Audio

The biggest issues with the Mac Mini seems to be the power supply. If you can solve that, either by using it on battery only or getting a linear supply you are good.


Thanks for your input Steve, but I believe the Devialet works best when the Mac is connected directly via ethernet. And use wifi for Tidal.
Otherwise ethernet is probably overloaded when functioning both as download(TIDAL) and upstreaming (DAC).
So the router is only connected via wifi. Would there still be gains with a LPS on the router?

The only way to substantially improve sound is to get a better DAC. Any upstream changes are just band aids because any noise or distortion or jitter or other issues are ALL coming from your DAC. If the DAC is well designed it will be well isolated from any contamination from digital incoming signals, clock timing and power supply noise.

Think about it - the DAC design goal is high fidelity of the source signal
- so a well designed DAC should remove all extraneous noise or contamination that is not the audio signal. A poorly designed DAC will be sensitive to which input is used and possibly even the type of cable.

Blaming mac mini or cabling or power supply for DAC inadequacies is misplaced logic. The DAC has a job to do and a key design requirement: to accurately reproduce the source signal without any extraneous factors influencing the sound.

Part of the problem of multifunction devices is that the complexity multiplies the possibility for errors in design of hardware or firmware and signal contamination from shared power supply. A dedicated device has simpler goals and is likely to perform better.

FWIW - Mac/PC and pro tools is how most music is produced these a computer core running roon and feeding a dedicated DAC is an excellent way to enjoy multiple digital formats.

You should not be "overloading" your ethernet unless you have multiple devices that are gaming and streaming HD video simultaneously - two channel audio is not a bandwidth hog.

For example, my wifi speed and download speed from my internet provider are both around 150 Mbps and two channel audio uses only 1% of this bandwidth.
shadorne,  I'm using Roon to send files from my iMac via ethernet over home wiring to an Ayre QX-5 Twenty DAC.  The SQ is excellent, especially with downloaded DSD files.  I've wondered if it might be better to store the files on my Mac Mini and connect it directly to the QX-5 via an ethernet cable.  As I understand your post, the answer is no, given the QX-5 is processing received files to analog, it that correct?

Just a couple notes:

if your Ethernet DAC is anything like a Merging Technoloies DAC, jitter is not an issue. That’s probably the biggest advantage of Ethernet DACs. 

The other thing, a 24/192 file uses (I believe) about 2000 kbps (2Mbps).  The cheapest routers today push 100 Mbps, and frankly, anything even remotely modern is pushing 1 Gbps. Even a 2010 Mac Mini has a 100 Mbps Ethernet Socket. To put it in perspective, Merging Tech can play DSD256 *multichannel* to multiple DACs from one NAS.  

Bottom line, don’t worry about bandwidth. 

I would expect the sound quality to be the same via the iMac or the Mac Mini if configured the same.

Post removed 
Thanks, shadorne.  The only difference in configuration would be ethernet through house wiring v direct cable from Mini.

@shadorne thnks for your view. I believe Devialet has a great isolated DAC so that should be covered. In this article though the top line Innuos does seem to better SQ. Does it make sense that a noise free server serve better quality bits into the DAC??

btw: this price range is out of my league, but I might go with the mini Innuos if there is a significant SQ improvement

Digital is digital. Bits are bits. So you dont need to worry about noise if you assume Devialet has built a robust device with good power supply and have separated and isolated the digital logic circuits from contaminating the DAC section (this can occur through a shared  power supply or mere proximity of circuits). The only complication with audio is the timing signal. This timing problem (jitter) is well understood and modern equipmemt should function asynchronously which means the jitter clock is ignored. If you are providing ethernet to the DAC then your DAC is behaving like a small computer and it will not be receiving timing information over ethernet like it would with SPDIF or Toslink - so jitter will be a function of the electronics and clock stability of the Devialet itself. The Devialet should not care about the path of how the ethernet signal reaches it - so upstream digital devices should have no effect at all (apart from contamination of the house AC power from SMPS supply and which Devialet should be designed  to cope with.

@shadorne your explanation seems to make sense. Which makes me an even happier Devialet owner as it safes me quite some money on extra boxes and fiddling around with all the great options out there. 
I guess its the good old upgraditis which tries to find a reason to invest more in this hobby.
it’s a good opportunty to use the money on Roon lifetime. The 1.3 version sounds great and has nice DSP features that offer perfectly audible changes. My guess is that we’re in an era where software is going to be more important to our hobby/experience. Like Microsoft was back in the days.
how is shadorne's post different?

he lists the following criteria:
- good power supply
- separated and isolated the digital logic circuits from contaminating the DAC section
- timing/jitter & which clock is used
- contamination of the house AC power from SMPS supply

What would you add to that list?

Ethernet is great so is optical. Optical has perfect galvanic isolation and a jitter immune DAC will just reject the higher jitter on an optical feed. However it all depends on a good DAC. If the DAC works perfectly then it will sound the same if the bits arriving are the same - no matter if they come over optical, coax, USB or ethernet. (not saying all DACs sound the same but a good one ought to sound the same on any digital input if it receives the same bit perfect data)

It boils down to the DAC performance - and performance leaves a lot to be desired for many products out no wonder folks report differences with all kinds of tweaks and different inputs, bridges etc.

That's what Rob Watts (designer of the Chord Dave) and Chris Hansen (Ayre) say as well: their DACs are immune to upstream signal issues.  

That was not true when I tested the Chord DAVE and the Ayre QX-5 Twenty.  Both still sounded better with SOtM gear in front of them.
I have not delved into the jitter rejection system on the Chord or Ayre.

Unfortunately most methodologies still allow low frequency jitter to affect performance even if all high frequency jitter is rejected and it performs well on the AES test that Stereophile perform. (Low frequency jitter can be quite audible)