minimalist network configuration for a digital source

[aKnyght]

audiophile warning: this post contains material which some may find heretical — reader discretion is advised

given all the recent discussion on digital source networking here, thought i would pass along my specific implementation for anyone who might be interested…

having an asynchronous DAC but no other legacy digital source equipment, i was able to design an entire digital chain without constraint according to one overriding principle: simple and elegant solutions win.

the resulting digital audio network configuration (red type) is as follows:

modem <—cat7—> router <—cat7—> FMC <—fiber—> transport <—usb—> DAC

the transport is self designed and built as a one box solution specifically to minimize the necessary supporting network infrastructure. it runs roon ROCK and provides storage, streaming, transport and rendering. consequently, the network is used only for streaming internet files (qobuz) and for control (roon iOS app). for anyone interested, more information here.

this network configuration is implemented entirely with quality, consumer-grade networking equipment and has been transformative to the SQ of my system.

here, the elimination of all upstream electrical noise provided by the direct input of fiber ethernet to the transport in conjunction with data correctness provided by roon’s use of TCP/IP cause me to believe that expensive audiophile networking gear such as switches, ethernet cables, power supplies, dongles, etc would provide no benefit over quality, consumer-grade gear in an audio network topology and system configuration such as this.

furthermore, as a copper ethernet connection to the transport provides a conduit for the continuous injection of electrical noise, the fiber ethernet connection not only benefits the SQ of streamed files but also locally stored files.

the last mile… the one outstanding item in this overall configuration is to replace the USB connection from the transport to the DAC with a fiber connection. however, to my knowledge, there is only one DAC having a non-proprietary optical fiber input. so, this upgrade will have to wait for more manufacturers to add this IO.
_______________________________

caveat 1: as nothing i have offered here is being taught at carnegie-mellon or cal tech… ymmv

caveat 2: i am firmly in the camp that everything (potentially) matters and will leave open the possibility that SQ could benefit from specialized audiophile network gear. it is, however, a possibility i feel in no way compelled to investigate at this point.

 

[Swisstrips]

Thanks for sharing. Is your main point that the fiber has provided the increase in SQ due to its inherent isolation?

Have you applied / tried any LPS with any of the network components?

There seems to be endless threads on power mgmt with regards to network specific components and then computer items (servers, streamers, endpoints, etc..)

If one is in the everything matters camp it seems that power supply, specifically at the proper location has the potential to bring about significant improvements. Where to apply quality power is system dependent and or the big question in so far as spending $ wisely.



Sent from my iPhone using Tapatalk

 

[UltraFast69]

aKnyght

The big part is making this fun in all of the effort, aka the hobby.

Thanks for sharing!


Sent from my iPad using Tapatalk Pro

 

[aKnyght]

Thanks for sharing. Is your main point that the fiber has provided the increase in SQ due to its inherent isolation?

Have you applied / tried any LPS with any of the network components?

i think the main point is that because fiber provides complete electrical isolation, i do not have to worry about any upstream network gear beyond ensuring that it performs its intended function as per industry specifications.

this includes power supplies. the benefit of an LPS would be to reduce extraneous electrical noise... noise which would not survive the optical isolation provided by the fiber. consequently, an LPS would be superfluous.

in my system configuration, SQ would benefit from an LPS for the transport which is downstream from the fiber. however, my plan is to replace the USB connection from the transport to the DAC with fiber, thereby, extending optical isolation all the way to the DAC --- but, this needs to wait until more DACs include fiber IO.

 

[aKnyght]

aKnyght

The big part is making this fun in all of the effort, aka the hobby.

Thanks for sharing!

totally agree!! i found all the research to learn about the issues, principles, theories, etc. involved and then the use of that knowledge to design and build the entire digital source (network + transport) to be extremely enjoyable. then to switch it on and hear the results which exceeded all expectations to be fully satisfying. i am just sad that process is now close to being over.

 

[Swisstrips]

i think the main point is that because fiber provides complete electrical isolation, i do not have to worry about any upstream network gear beyond ensuring that it performs its intended function as per industry specifications.

this includes power supplies. the benefit of an LPS would be to reduce extraneous electrical noise... noise which would not survive the optical isolation provided by the fiber. consequently, an LPS would be superfluous.

in my system configuration, SQ would benefit from an LPS for the transport which is downstream from the fiber. however, my plan is to replace the USB connection from the transport to the DAC with fiber, thereby, extending optical isolation all the way to the DAC --- but, this needs to wait until more DACs include fiber IO.

Yes agreed and I think that is the "general" consensus however, there is never really any consensus/agreement in this hobby

I'm also running fiber on my home network (home and listening room) and am discovering a fair amount of info that still suggest LPS will still benefit (server source, switches, NAS) but as with anything its really up to the indiv to experiment and determine if the $ is worth it.

I'm looking to replace/upgrade my mU to something else and in the process of researching this, it really opens up a can O worms, thus my comments above on LPS.

A relatively inexpensive item to try would be the sonore optical modules to replace the consumer grade FMC's and would ideally need two - one at the switch and one at DAC. My L2 managed switch is AC powered and has two SFP ports but even in this instance, "they" say go Ethernet out of switch > optical module > fiber > 2nd optical module > streamer.

OR as an alternative fiber from switch (in network rack in my configuration) to a "dumb" switch (in audio rack) from some of newly developed "audiophile" type of switches (sotm, uptone, liner solution). These audiophile switches offer "better" clocking/isolation but are $ so a lot of available choices.

 

[aKnyght]

agreed... all theories need to stand up to empirical confirmation. plenty of room for experimentation!!

A relatively inexpensive item to try would be the sonore optical modules to replace the consumer grade FMC's and would ideally need two - one at the switch and one at DAC. My L2 managed switch is AC powered and has two SFP ports but even in this instance, "they" say go Ethernet out of switch > optical module > fiber > 2nd optical module > streamer.

i am not sure the sonore OM would add anything over the SFP ports on an upstream switch... the electrical>optical signal conversion is the last thing that takes place (in the SFP transceiver not the OM). anything upstream of that is irrelevant for the purposes discussed here. on the downstream end, the OM could be beneficial if its circuitry post optical>electrical conversion by the SFP module is less noisy -- which i suspect might be the case.

...as an alternative fiber from switch (in network rack in my configuration) to a "dumb" switch (in audio rack) from some of newly developed "audiophile" type of switches (sotm, uptone, liner solution). These audiophile switches offer "better" clocking/isolation but are $ so a lot of available choices.

certainly worth trying to find out empirically. my a priori thoughts would be that:

> the isolation could only be as good as fiber but, most likely not.

> the "better" clock would add nothing to SQ... for an asynchronous DAC, the only clock that matters is the one on the DAC used to time the audio stream being feed directly to the digital-to-analog conversion circuity (chip). all other clocks/clocking in the digital chain are used for more mundane things like network data transmission, process scheduling, operation of the CPU, etc...

...more generally: as long a piece of network gear faithfully implements its respective industry standards/protocols and operates within specifications, full interoperation with and full functionality of audio devices (i.e. DAC) is guaranteed. however, this breaks down when a piece of audio gear requires network gear to function beyond its industry specifications to achieve full performance… and, i would argue that such audio gear has not been fully designed or adequately specified.

again, all this is just my theory ...ymmv

 

[W9TR]

If you’d like to experiment with fiber isolation, here is a way to do it that won’t break the bank. Buy two of these:
TP-Link Fast Ethernet SFP to RJ45... Amazon.com

Buy one of these:
2 Meter OM1 Multimode Duplex... Amazon.com

Total cost is $55.

Buy the linear power supply of your choice, 9V 500mA, to use on the ‘silent’ end of this fiber break.

I have seen a marked improvement using this low cost solution. I built my own 9V linear supply for a few dollars. I can share if others are interested.

 

[nicoff]

I am always curious by all the talk about ‘electrical noise’ that somehow some folks can “hear” that benefits from isolation to the point that it generates all kind of accolades.
I would love to see some graph and/or objective measurements and test results that actually show and quantify this noise.

 

[W9TR]

I am working on a way to instrument this end-to-end on my system - that is from a test file stored on my NAS, played through Roon, to various endpoints/DAC’s and then into my Audio Precison System 222A analyzer. It’s still a work in progress. It would be nice to validate what I’m hearing.