Network Improvements and their Impact on Sound Quality

The goal of this thread is to share experiences with the Network environment associated with streamed digital music. The hypothesis is that just like other areas of audio, improvements to and optimization of the quality of the network used to stream digital music can have a very positive impact on sound quality. While the underlying principals are relatively straightforward, networking, by virtue of the huge number of variables, is a very complex field. There are many ways to set up a perfectly functioning network and a huge variety of devices that can be employed....i would hope that through this thread and the experience and experimentation of key contributors, we may be able to narrow down the number of devices that are proven to work well for audio applications and develop a few networking concepts that function particularly well for the reproduction of music.
My own personal experience in optimizing my network has led to a number of very positive surprises....
  • large improvements in sound quality do not necessarily require large financial investments...in fact, many times the opposite holds true
  • standard 16/44.1 Redbook format files can sound incredibly good and highly satisfying, musically.....jaw droppingly so
  • internet radio at 128kbps can be hugely entertaining and a brilliant source of new music
  • some of the biggest improvements lie in the most unexpected of places
  • the law of diminishing returns doesn’t seem to operate the same way as classic audio...from a network standpoint, the better and more revealing your system becomes, the greater the impact of future network improvements
  • you may really need to rethink/revise your value scales....for example buying a $500 silver/gold DC cable for a $200 router may be an entirely sound decision (s’cuse the pun)
  • the efficacy of a device is highly dependent on the environment in which it‘s used. Because a device works well in one installation doesn’t necessarily mean it will work well in all installations...there are simply too many variables
  • careful implementation of a network will often achieve superior results compared to chasing the most highly reviewed hi-fi components and installing them in an less than optimum network environment
For the above reasons, I am keen to share networking experiences...I’m looking forward to benefitting from the wealth of knowledge and expertise of this forum’s participants.
 
Last edited by a moderator:

Comments

Apr 16, 2019
14
12
8
49
Hello,

Could someone please tell me if the BUFFALO BS-GSP Series BS-GS2008P is a comparable improvement to the models listed above in the Buffalo BS-GS switch list?

Thank you
Hello!
I think these models are equipped with PoE capability.
SOme users have got these models and if I am correct they were found to perform equally good.
Visit the other forum and there are quite a few comments about these switches.

I would like to underline that 2x switches make a stellar improvement in SQ but 4x switches make a stratospheric improvement. It seems that all relates to noise suppression. I tried 5 units and 6 units without any benefit at all. I also think that Paul Pang also uses 4x switches.
 

wil

Well-Known Member
Jul 22, 2015
353
171
128
SotM switches are for sure top performers. No doubts.
I have been contacted by one of the greatest network 'gurus' in Greece that switches for audio applications need to be paired for reasons that relate to noise rejection.
So as soon as you have 2x SotMs I can find no reason not to add another pair like the Buffalos. Configuration is not that difficult. Actually it is very easy. They take the IP address 192.168.1.254 if I am right. You can visit their web site, they have full description. So easy.
Petavgeris, thanks for the switch information.

Can you or the Greek Network Guru provide any technical explanation why even numbers of switches is better at noise rejection than odd numbers?

And would this apply only to the same switche model paired up, or would, for instance, an etherRegen paired with a Buffalo provide similar results?
 

Blackmorec

Well-Known Member
Feb 1, 2019
339
428
68
67
I love to read about improvements in sound quality posted by like minded individuals around the World. Often I’m willing to try some of the suggestions, but frankly this trend to cascaded switches is not floating my boat. If it takes 4 switches in series to get the best SQ, then I’m going to hang back until someone figures out what 4 switches do that‘s so beneficial. When you think about it, many installations don‘t actually need the functionality of a switch, so the improvement a switch makes is coming from something the switch does to the incoming data stream.
Also when you think about it, a switch is actually making a facsimile or copy of the incoming data stream....its resynthesising the stream based on what’s received. We can deduce from the fact the 4 switches keep improving the process, that its the quality of the incoming stream that results in a more accurately synthesised output.
All a switch is doing is taking DC power and modulating and timing it according to the pattern of bits it receives....so clearly the less noise and jitter on the incoming stream, the better the output. (Feel free to knock my logic if you see flaws :cool: )
So, again following the logic, the best way to improve a switch‘s output, is to improve it’s input. (Not exactly rocket science) and the best way to improve input is the make sure that the DC power that was used to create the input is as pure, noise-free and rippleless as possible. That way, clock errors and therefore jitter and phase noise are minimized, the resynthesised stream is more accurate and less noise is included with the stream to disrupt downstream processes and processing.

All of which makes we wonder. What happens...if instead of adding 4 switches with 4 power supplies, 5 ethernet cables, 4 DC cables, 4 power cables and appropriate anti-vibration measures, you spend your $$$$s or ££££s on a single switch and a REALLY kick-ass power supply, power and DC cable, vibration control and output Ethernet cable.

Putting my money where my mouth is this is exactly what I’ve done. I started with an AQVox SE with its wall wart SMPS. AQVox state on a leaflet included with their switch that adding better power supplies does not improve performance and can cause SQ deterioration because the supplied SMPS is so carefully engineered and matched to the switch. Well let me tell you that in my experience that’s just simply incorrect. I have made several upgrades to my AQVox SE’s power supply and each upgrade has been rewarded by a significant, sometimes ‘jaw dropping’, shivers-down-the-spine improvement in SQ. And so far I have not reached a limit...in fact, each improvement seems to bring larger rather than smaller performance upgrades. I have not compared the improvements I have obtained with power supplies with multiple switches, but what i can say is that if I obtained the same level of improvement by adding a switch, I’d be really, really happy. If I compare the sound quality I’m getting now with the sound quality of the original AQVox SE + SMPS wall wart, you would judge the standard OEM supplied product to be severely impaired and missing oodles of detail and information, such as a fully focused, room expanding 3 dimensional sound scape populated by what sound like real musicians. With all its PS enhancements, the modest AQVox SE converts my listening room to a dimensionless soundscape (the dimensions you ‘hear’ are created by the recording) where I can sometimes hear the recording venue even before the music starts (depending on the recording obviously) And that from a standard Qobuz remote stream.
I have a further 2 upgrades in mind....one to the mains consumer unit (Gigawatt MCBs and Doepke differential switch) and upgrades to my Sean Jacob’s power supplies. Once that’s complete I’m probably going to stop upgrading and start saving for Magico’s S1 MkIII speakers...the ones which will very likely include a new aluminium honeycomb bass/midrange driver ala M9 and A5.
For me, cascading switches to improve SQ is essentially using an unexplained phenomena, logically likely to be noise and jitter related. That being the case, there’s a really good engineering solution that theoretically achieves the same goal, namely major PS improvements. And from what I’m hearing, the better the power supply used in making your datastream’s final synthesis, the better the SQ you’ll achieve. And again theoretically the improvements should keep coming until the synthesis is performed using perfectly noiseless, stable DC.
 

Abyss Man

Active Member
Jun 17, 2019
187
126
43
I’m on the same wavelength as you Blackmorec, I wanted to try out the Buffalo switches but then realised they are 100v devices and then to get another power cord, or LPS, isolation feet, ethernet cables and the space. Forget it, I’m already extremely happy with my dual soTm switch configuration. I have also changed out all my psu from sps-500 to Farad 3. That has brought about an excellent jump in SQ coupled with the Mundorf Silver/Gold DC cable. I’m a happy bunny.
 

gipe

New Member
Oct 15, 2019
20
3
3
France
Hello,

Thank you all for your help on this issue.

On the problem of noise and jitter on the incoming flow, I can make the following observation.

I'm essentially listening to Qobuz. I have a fibre optic Internet access on box Orange (France) connected to a HDPLEX 200 linear power supply. The current tajet is simple box > streamer Soundaware D300 Ref > DAC Terminator > Integrated Absolare Signature.

A passing friend installed his AQVox SE to try it. In a common opinion it brought little and did not justify at all the investment.

Perhaps the good power supply of the box improves the quality of the incoming flow to a large extent ?
 
Nov 10, 2018
1
0
6
64
Has anyone out there tried, or planning to try, Shunyata Research’s new Omega Ethernet cable which has filtering built into it, supposed to be big step up from their Sigma offering?

https://shunyata-uk.com/product/digital-cables/omega-ethernet/

Certainly would be simpler than 4 switches with associated power cords and interconnects.

I would certainly be interested in feedback and findings you may have.

Thanks

Graham
 

moby2004

Well-Known Member
Jan 21, 2018
133
121
113
I love to read about improvements in sound quality posted by like minded individuals around the World. Often I’m willing to try some of the suggestions, but frankly this trend to cascaded switches is not floating my boat. If it takes 4 switches in series to get the best SQ, then I’m going to hang back until someone figures out what 4 switches do that‘s so beneficial. When you think about it, many installations don‘t actually need the functionality of a switch, so the improvement a switch makes is coming from something the switch does to the incoming data stream.
Also when you think about it, a switch is actually making a facsimile or copy of the incoming data stream....its resynthesising the stream based on what’s received. We can deduce from the fact the 4 switches keep improving the process, that its the quality of the incoming stream that results in a more accurately synthesised output.
All a switch is doing is taking DC power and modulating and timing it according to the pattern of bits it receives....so clearly the less noise and jitter on the incoming stream, the better the output. (Feel free to knock my logic if you see flaws :cool: )
So, again following the logic, the best way to improve a switch‘s output, is to improve it’s input. (Not exactly rocket science) and the best way to improve input is the make sure that the DC power that was used to create the input is as pure, noise-free and rippleless as possible. That way, clock errors and therefore jitter and phase noise are minimized, the resynthesised stream is more accurate and less noise is included with the stream to disrupt downstream processes and processing.

All of which makes we wonder. What happens...if instead of adding 4 switches with 4 power supplies, 5 ethernet cables, 4 DC cables, 4 power cables and appropriate anti-vibration measures, you spend your $$$$s or ££££s on a single switch and a REALLY kick-ass power supply, power and DC cable, vibration control and output Ethernet cable.

Putting my money where my mouth is this is exactly what I’ve done. I started with an AQVox SE with its wall wart SMPS. AQVox state on a leaflet included with their switch that adding better power supplies does not improve performance and can cause SQ deterioration because the supplied SMPS is so carefully engineered and matched to the switch. Well let me tell you that in my experience that’s just simply incorrect. I have made several upgrades to my AQVox SE’s power supply and each upgrade has been rewarded by a significant, sometimes ‘jaw dropping’, shivers-down-the-spine improvement in SQ. And so far I have not reached a limit...in fact, each improvement seems to bring larger rather than smaller performance upgrades. I have not compared the improvements I have obtained with power supplies with multiple switches, but what i can say is that if I obtained the same level of improvement by adding a switch, I’d be really, really happy. If I compare the sound quality I’m getting now with the sound quality of the original AQVox SE + SMPS wall wart, you would judge the standard OEM supplied product to be severely impaired and missing oodles of detail and information, such as a fully focused, room expanding 3 dimensional sound scape populated by what sound like real musicians. With all its PS enhancements, the modest AQVox SE converts my listening room to a dimensionless soundscape (the dimensions you ‘hear’ are created by the recording) where I can sometimes hear the recording venue even before the music starts (depending on the recording obviously) And that from a standard Qobuz remote stream.
I have a further 2 upgrades in mind....one to the mains consumer unit (Gigawatt MCBs and Doepke differential switch) and upgrades to my Sean Jacob’s power supplies. Once that’s complete I’m probably going to stop upgrading and start saving for Magico’s S1 MkIII speakers...the ones which will very likely include a new aluminium honeycomb bass/midrange driver ala M9 and A5.
For me, cascading switches to improve SQ is essentially using an unexplained phenomena, logically likely to be noise and jitter related. That being the case, there’s a really good engineering solution that theoretically achieves the same goal, namely major PS improvements. And from what I’m hearing, the better the power supply used in making your datastream’s final synthesis, the better the SQ you’ll achieve. And again theoretically the improvements should keep coming until the synthesis is performed using perfectly noiseless, stable DC.
I am also interested in Sean Jacob’s power supplies and notably the DC4. The “ perfect” power supply for switch(es) ?
 

kswanson27

Active Member
Nov 22, 2018
132
32
33
76
I am also interested in Sean Jacob’s power supplies and notably the DC4. The “ perfect” power supply for switch(es) ?
Anyone have any idea if the EdgeRouterX configured as a switch adds any more or less SQ than the SOtM or ER switches?
 
May 12, 2018
62
22
15
Perth, WA
Hi kswanson27.

I'm no expert and I don't have the audiophile switches you mentioned.

I do have the EdgeRouterX-SFP which I use as a router to replace my ISP Wifi Router.

I think Nenon said best regarding the EdgeRouter back in post #245 on this thread.

G
 

XCop5089

Well-Known Member
Sep 5, 2015
95
44
105
Winchester, UK
Hello!
I think these models are equipped with PoE capability.
SOme users have got these models and if I am correct they were found to perform equally good.
Visit the other forum and there are quite a few comments about these switches.

I would like to underline that 2x switches make a stellar improvement in SQ but 4x switches make a stratospheric improvement. It seems that all relates to noise suppression. I tried 5 units and 6 units without any benefit at all. I also think that Paul Pang also uses 4x switches.
Hi Peter

I asked you a similar question over on Audiophile Style, but your post got deleted!!!!

As a UK-based enthusiast, I have found it impossible to source a Buffalo BS-GS2008/2016 switch!

Do you recommend the Buffalo BS-GU2016 unmanaged switch as an alternative with the same performance uptick please?
 
Apr 16, 2019
14
12
8
49
Petavgeris, thanks for the switch information.

Can you or the Greek Network Guru provide any technical explanation why even numbers of switches is better at noise rejection than odd numbers?

And would this apply only to the same switche model paired up, or would, for instance, an etherRegen paired with a Buffalo provide similar results?
Hello!
My local friend who is a network administrator, providing network admin services to various companies, has told me that for audio purposes the switches need to be paired *under any circumstances*. He told me that all relates to noise suppression that can be very easily measured. He even told me to use the same ports for the consecutive switches. That means you should better connect port #1 on Switch 1 to port #1 on Switch 2.
I am far from being a network expert. I just pass his recommendation.
Regarding my experience, yes, 2x Brand A in series with another 2x Brand B, did make a positive difference. Mixing various switches in different fashion than the one described above, did not lead to better results. So his assumption regarding noise suppression seems to be real.
 
Apr 16, 2019
14
12
8
49
Hi Peter

I asked you a similar question over on Audiophile Style, but your post got deleted!!!!

As a UK-based enthusiast, I have found it impossible to source a Buffalo BS-GS2008/2016 switch!

Do you recommend the Buffalo BS-GU2016 unmanaged switch as an alternative with the same performance uptick please?
Yes! I wanted to reply to your comment but they deleted the post.
I do not know why they delete posts. Some guys there have ironic behavior and they make fun by leaving ironic comments and flames 'under the table'. I'm not that kind of person, so I decided to terminate my presence there and never leave again any comment. Bye bye!

So, regarding your question, all I can say is that the Buffalos have been vanished in Europe and USA (if I am correct). What I wrote and was deleted is that I cannot find any reason to spend thousands on anti-vibration and supporting gear without having spent the basics for the most critical parts that are the network equipment. In 3 systems tried so far, the most dramatic improvement was the addition of 4x switches. The power supplies that were used for the comparison of the 4x pieces were SMPS and not LPS. We had compared 2x units (Melco & 2008) with LPS, OK, no need to write more on this. On two other systems, when trying 2x vs 4x 2008, the difference was night-and-day by using SMPS. I have not compared 2x on LPS vs 4x on SMPS but I guess that the results should be similar.

You can still find some Buffalos on eBay Japan. IMHO you should go and get at least 2x 2008 because it is a matter of days that everything will be vanished. They play so damn good. Ubiquiti as switch was tested on the other forum and did not have so good results.

You can send me a PM if you want and if you think I could help further.
Greetings from Athens.
Peter
 

nenon

Active Member
Jan 29, 2020
50
195
35
39
I was just told by a US reseller of network equipment that they have 10 x Buffalo BS-GS2016 switches in their warehouse. That's the switch with the same board as the Melco switch. I'll talk to them tomorrow to confirm. If anyone is interested, send me a PM.
 
Apr 16, 2019
14
12
8
49
I was just told by a US reseller of network equipment that they have 10 x Buffalo BS-GS2016 switches in their warehouse. That's the switch with the same board as the Melco switch. I'll talk to them tomorrow to confirm. If anyone is interested, send me a PM.
If I didn't have any switches at all and if I was living in usa, I would go and buy them all at once. I would use them in stacked quads and would never look back again, regarding network upgrades. It is by far the minimum expense with the maximum reward related to SQ uplift.
 

Gladiator

Well-Known Member
Oct 9, 2015
36
1
88
I was just told by a US reseller of network equipment that they have 10 x Buffalo BS-GS2016 switches in their warehouse. That's the switch with the same board as the Melco switch. I'll talk to them tomorrow to confirm. If anyone is interested, send me a PM.
Sent you a PM.
 

JGlacken

Well-Known Member
Mar 4, 2011
45
2
413
Sent you a PM.
 
Jul 16, 2020
1
0
1
71
I was just told by a US reseller of network equipment that they have 10 x Buffalo BS-GS2016 switches in their warehouse. That's the switch with the same board as the Melco switch. I'll talk to them tomorrow to confirm. If anyone is interested, send me a PM.
PM sent
 

nenon

Active Member
Jan 29, 2020
50
195
35
39
Here is some background on the Buffalo / Melco switches.

The Buffalo BS-GS2016 switch was designed as a business switch. Some audiophiles in Japan figured out it's really good for audio and started modifying these switches with linear power supplies, custom configurations, etc. Buffalo then released a limited audio version of that switch for the Japanese market called Buffalo BS-GS2016A (A stand for audio I guess). It was no different than the the regular Buffalo BS-GS2016 switch, but it had the top ports disabled and some settings changed - disabled all top RJ45 ports, disabled LLDP, enabled DHCP. They included a better cable and sold it a little bit more expensive.

It's interesting that they did not use the 8-port version. The 16-port version with disabled top ports (which makes it an 8 or 9-port switch) was prefered by those Japanese audiophiles. Later on Melco also picked the 16-port version and did not expose the top copper ports. Perhaps that is because the 16-port also has SFP ports for fiber connectivity. Or perhaps there is more into it...

In February, a member of AS hinted that the board on the Buffalo BS-GS2016 looks like the board Melco used on their switch. I bought one of these Buffalo's on the next day:
https://audiophilestyle.com/forums/...eaming/page/635/?tab=comments#comment-1032742

We were curious to compare them side by side and most importantly to see what kind of clock Melco used. At the time we expected that Melco used the same board and upgraded the clock. However, the clock is hidden under the heatsink and we could not tell. A member who had the Melco switch offered to send it to me, so I can take a further look. To my surprise, I discovered that the clock on the Melco was not upgraded. I was hesitant to write much about this, from respect to Melco, and only shared this in a small circle originally. The word quickly got out, and it was not a secret anymore. I was getting hammered with messages after someone posted this publicly and decided to create a post I can point people to. Here is my post comparing the Melco and Buffalo switches:
https://audiophilestyle.com/forums/...eaming/page/651/?tab=comments#comment-1047951

The Buffalo BS-GS2016 was widely available on the US market before that post. And I managed to order a few more of those switches, which I started modifying, testing different DC cabling inside the switch, different vibration treatments, chassis damping, capacitors, different configurations, etc.
Shortly after my post, that particular Buffalo model disappeared from the US market. We can only speculate why, but I think there were multiple reasons. They were slated to go end-of-sale already. I am guessing audiophiles bought a bunch of them and emptied the local warehouses. Not sure if Melco had any influence or not. Buffalo and Melco are affiliated. I managed to buy a couple more switches from Europe. Shortly after that, they disappeared from the EU market too. Before those posts, I was able to find them for $200. After the post, I was lucky if I could get one for $300. At the end I was able to find them for $350 and the last switch I bought was $400 (because of the hefty shipping fee).
That reminds me of the PlanetTech transceiver shortage the day after Emile posted that he liked them :).

I have tried 2,3,4, and more of these switches in series.
Screen Shot 2020-07-16 at 9.31.24 AM.png

Yes, 2 is better than 1. And 3 is better than 2. But we are talking about diminishing returns here. I would prefer two switches with better LPS than four switches with the stock SMPS. Those switches have a really noisy switching mode power supply inside. They sound super harsh with that power supply. I remove it and install a DC connector on them. Power up with a good linear power supply and they start delivering. I was able to power the switch with an Uptone Audio LPS-2.1 running at 12V, but I did not try passing any traffic, just saw it boot. But that means they consume less than 1.2A during boot. What you see on the bottom of the picture is my custom DIY Sean Jacobs DC3 LPS, which I used to power my network devices. They take about 3 weeks of non-stop running to get through the initial break-in period.

It was a bit shocking the first time I realised that the DC cable you use to power those switches (and network devices in general) makes a difference. I can easily hear the difference between different DC cables in my system. What is even more shocking is that I can easily hear a difference between two identical DC cables, one with passive JSSG360 shielding and one without. So, I started paying attention to those things. Tried all kind of different wires and my favorite one is the 15.5 AWG Mundorf silver/gold wire - https://www.partsconnexion.com/MUNDORF-72181.html. There is a DIY DC cable recipe I have posted. You can find it here, just follow the references. This is the best DC cable I have heard. by far! If AudioQuest or another big cable company was offering this cable, my guess is it would cost $2,000 or more. If you can solder, you can make one for $250-$300 I guess. I use the same wire as the internal hookup for the DC connector on the switch.

BTW, Marcin from JCAT was kind enough to help me figure out the pinout of the M12 Switch Gold. I just made one of these silver/gold for my friend's M12 switch and he seems thrilled.

Back to the Buffalo switch... it responds really good to vibration treatment and chassis damping too. I am now playing with various capacitors and clocks. The next step is to replace the cheap clock with a PinkFaun ultraOCXO clock (€1250) and power up with a two rail Sean Jacobs DC4 power supply (£3700) - one for the switch and one for the clock. I just need to find some more time for that. Then I will get a M12 Switch Gold and compare. BTW, PinkFaun installed their ultraOCXO clock on a Melco switch. The owner who I have been in touch with really liked it.

I may try to bypass the switching mode regulators too as one other person successfully did here - https://audiophilestyle.com/forums/...-3-external-pss/?tab=comments#comment-1067961
But I have quite a few other higher priority experiments queued up before that.

Quite a few people have tried these switches, some compared them to the Melco switch, which has a bank of extra capacitors. The recollection of my stats is that from 10 people, 8 liked it a lot and 2 did not like it that much. Incidentally, the two people who did not like it had very high-end headphone systems (with the top of the line Abyss headphones I believe).

I work in the IT field and asked a few of our vendors if they can find more of these Buffalo switches for personal use. I got a message yesterday that one of these vendors may have 10 of them. I was excited that I might have found more of these at $200 per switch. But the price they gave me was close to $350 per switch after shipping. It's even more complicated that they are business to business only and not easy to buy for personal use. My initial idea was to buy all 10, keep 4 of them and pass 6 to you guys, on a first come first serve basis. But it turns out it may not be that easy or cheap as I thought.

I also have the little 8-port model. It's going through the burn-in process and will compare to the 16-port model. Screen Shot 2020-07-16 at 12.14.14 PM.png
 
Last edited:

austinpop

Well-Known Member
May 23, 2017
55
92
83
Austin, TX
Thanks for the excellent summary!

The recollection of my stats is that from 10 people, 8 liked it a lot and 2 did not like it that much. Incidentally, the two people who did not like it had very high-end headphone systems (with the top of the line Abyss headphones I believe).
I had the good fortune to compare Nenon's modded GS-2016 with the Melco and the eR, and I posted my findings here: https://audiophilestyle.com/forums/...dio-streaming/?do=findComment&comment=1047922.

To be fair — it's not that I "did not like" the Buffalo, it's just that I did not think it completely equaled the Melco or the eR. Still, quite excellent for a $200-400 piece of gear.
 

Brucemck2

Member Sponsor
May 10, 2010
302
23
423
Houston area

About us

  • What’s Best Forum is THE forum for high-end audio, product reviews, advice and sharing experiences on the best of everything else. A place where audiophiles and audio companies discuss existing and new audio products, music servers, music streamers and computer audio, digital to audio converters, turntables, phono stages, cartridges, reel to reel, speakers, headphones, tube amplifiers and solid state amplification. Founded in 2010 What's Best Forum invites intelligent and courteous people of all interests and backgrounds to describe and discuss the best of everything. From beginners to life-long hobbyists to industry professionals we enjoy learning about new things and meeting new people and participating in spirited debates.

Quick Navigation

User Menu

Steve Williams
Site Founder | Site Owner | Administrator
Ron Resnick
Site Co-Owner | Administrator
Julian (The Fixer)
Website Build | Marketing Managersing