Network optimization for serious streamers

I actually found the cheap iFi X was better than the Uptone JS-2 on the modem/router/wifi hardware. For streamers it was the opposite, the JS-2 was better. This surprised me because a linear power supply always tends to improve things.

But I had never tried lps upgrades on network switch type hardware before.

@fredrik222 You have now said your piece multiple times - saying the same things over and over. Now go away. People like you feel a need to SPAM (yes its SPAMING) these conversations again and again with the same thing. We don't care or go into your threads on ASR doing this. It's just your opinion, nothing more.

I wish a manufacturer designed an in house wireless streamer. With the air gap it could theoretically solve all these network noise problems.

But because the only choice is off the shelf wireless solutions they all sound worse than hard wired.

There are a few manufacturers who now eschew the use of metal housed RJ45 connectors as used on many high end ethernet cables (such as Nordost), claiming the metal connector touches the ground, and this then exposes the external jacketing to a high frequency skin effect, reducing performance.

Its all black magic but there may be something to this.

It seems the best solution is to wire your house with Ethernet ports, then turn wireless off permanently in the modem/router no?

@singingg The contiguous ground is what passes noise down the chain from one component to the next. For the purposes of audiophile streaming this is one ‘performance benefit’ you don’t want.

@sns I got all setup to do a subjective A-B test to see how much gain there is to be had disabling wireless on the modem router.

When I sat down to do the 2nd half I found I could not connect to my Lindemann streamer with my iPhone app. Cos wireless was disabled. Idiot. =)).

As suggested here I will try the TP-Link Omada as a router with separate wireless access point.

@singingg Its not a choice between shielded vs unshielded cables. Its whether the ground is contiguous. Some of the cable manufacturers are floating the shields, floating at one end, not using metal RJ45 connectors, e.t.c

Of course doing these things means they are not meeting Cat7/8 spec, but they claim its results in better network audio streaming performance.

@singingg Drain ground wire only on one end, so as not to pass noise. Melco Audio specialise in audiophile switches, so have hardware experience in this.

@singingg I think its so new there is no distribution yet. Melco does have an older model cable that does the same thing. But remember this is just an entry level product.

There is now a huge number of boutique ethernet cable builders doing weird and wonderful things like this to kill noise, with prices into the stratosphere.

Ok I got a TP-Link wireless access point today. My setup is a Lindemann Limetree Bridge II streamer - spdif out into Teddy Pardo DAC (with alps blue velvet volume control option).

When turning the wireless off in the modem/router there was a noticeable jump in performance. Previously when just using the normal modem/router, hard wired ethernet into the Lindemann was better but it was subtle- I could live with the wireless connection and not have a 30 foot cable snaking down my hallway (in-wall wiring is coming soon tho). But now with the wireless turned off in the modem, and a wireless access point used instead, I cannot go back.

Kudos to @sns for working this one. I must have read a hundred pages of audiophile switch investigations, with these guys spending 5-10k on switch/clock setups, and I cannot think of a single instance where any attention was paid to the upstream cable modem. Sure put an LPS on it, but that’s about it. The wireless has a bad effect.

And all this just using a standard residential TP-Link modem/router. I have a TP-Link Omada wired router (ER605) on order. Wonder if this will be better.

@soix You have to be on a PC and log into the admin console of the modem/router by typing its IP address into your web browser.

Default IP address will vary by manufacturer, but for TP-Link its often...

192.168.1.1 or 192.168.1.100

Then you will need the username/password which are often admin/admin or its printed on the sticker underside of the device.
Once in the admin console, find the Wireless settings and disable both 2.4ghz and 5ghz wireless. Modern devices have these two different wireless modes and they can be enabled/disabled individually.

I’ve not heard a good result from fibre yet. The conversion from photons to electrons is very critical and can itself cause noise. Different Fibre SFP modules from different manufacturers sound different. In theory they should all sound the same if fibre eliminated these issues.

I am just wary of these fibre converters because I remember back in the day Optical was supposedly superior to spdif because of its electrical isolation. Yet I never heard an optical output sounding better than a spdif, and the difference widened and magnified when quality digital cabling was used.

I can imagine the power supply for the fibre conversion boxes play a huge part, you probably need a good LPS to get the performance, 2x of them, and then power cables on top of that. After that outlay I think its better to just invest in a higher end switch.

Here is my theory that squares things with the network packet objectivists.

The upstream network gear has various high frequency oscillators and pcb design approach that creates noise that rides down the chain via ground, shielding and skin effect from component to component. This noise does not effect data packet transfer, ERC or any of the things the trolls are continuously ranting about.

But that noise will negatively effect the DAC and it’s analog output stage.

This explains why doing things like fibre conversion has such great reported results, because it breaks this noise chain. Also explains why LPS or proper power supply design in custom exotic switches result in sound quality gains at the end of the chain. Again, nothing here is contradicting the network data packet transfer operation, it works fine beneath these other issues. 

@jji666 You would think so, but consider how even how the most expensive DAC’s respond positively to better power conditioning, cables and vibration isolation. So they are not going to reject digital high frequency noise easily either.

When I first got into high end audio in about 2002, the industry was just starting to move away from cd players to separate DACs and transports. The exact same troll skeptic screaming and ranting was going on then in the forums - but about SPDIF cables! It was all 1's & 0's, just digital data, so no possible way any type of improved coaxial cable could make any difference, its all bit perfect. Any changes heard are all in your head, a psychoacoustic phenomena or expectation bias because you paid so much for the cable e.t.c

On an on it went but they eventually conceded SPDIF can sound different. But about 10 years later, around 2012, it started again with USB cables. Again the exact same arguments, digital data 1's & 0's, it either passes the data perfect or it doesn't, you are just a fool tricking yourself into hearing differences e.t.c

Now another 10 years later, this time network transfer. Same argument, same fury, same limited theoretical understanding of what is causing these obviously heard changes.

I can guarantee I know more about Ethernet and the related protocols than you ever will. 

Are you sure about this statement? I am in IT and manage computer networks daily and configure, setup and install commercial grade network hardware for my clients. Sure maybe you do know more than me on this subject, but your presumptuous arrogance is typical of the theoretical know nothings who worship at Amir's alter.

When someone replaced a Cat 7 with a Cat 5 cable, you spun that into 'he removed all the audiophile stuff and it then worked' then proceeded to exaggerate this and make it the whole argument. That is a perfect example of dishonestly in argumentation and shows who you really are and the obfuscation you are willing to engage in. But this is part for the course with ASR disciples and is exactly how Amir and crew operate.

@singingg Just start out with the iFi Power X. I just got 3 of them for all my 12v network hardware. I have an Uptone JS-2 and Farad Super 3 for some other gear but want to limit $ expenditure for now so stuck with the iFi.

I have discovered a new performance gain on the router, at least I think it has not been mentioned previously.

Change LAN port speed from Auto/1000M/1GB to 100M Half-Duplex.

This investigation resulted from me moving from a consumer TP-Link Archer wifi-router to a (wifi-less) TP-Link Omada ER605 VPN router. The ’upgrade’ resulted in a decrease in performance. It added a slight (very slight) amount of noise to everything. Audible in music, audible in streamed movie soundtrack content, and visible in movie visual content. This was disappointing to say the least considering the Omada should be better hardware, not worse (or at least the same as before).

So I started playing. First idea was perhaps the 5 LAN ports run by 3 banks of circuit chips were sharing the NAS and streamer on one chip bank causing issues. So I swapped connections around and nothing changed.

Then I looked into port speed configuration, changing the LAN ports from Auto-Negotiation to 1000M Full Duplex. Nothing changed.

Then I changed port configuration to 100M Half-Duplex. Immediate and noticeable improvement. Almost as big as isolating the wi-fi as discussed previously. Now this Omada router has pulled ahead of the generic Archer router. In fact now my system sounds the best it ever has, just from this software setting in the router admin console.

I do not know how this is all possible. Maybe the chips that run the LAN ports at full Gigabit speeds increase noise the higher they go, and setting them to 100M (1/10th the speed of Gigabit) reduces that noise.

But a word of caution - there is also a choice to run a LAN port either at 100M Half-Duplex or 100M Full-Duplex. Maybe its not the speed reduction from 1000 to 100, but the fact there is now a choice of ’Full-Duplex vs Half-Duplex’ at the 100M speed level. I tried to compare the 100M Full-Duplex option and it crashed my network so was unable to eliminate that as being a potential factor.

To be clear, this is me running everything through the router. Maybe putting an audiophile switch between the router and my audio streamer will resolve any issue around LAN port speeds, making it all a null issue. But for now it helps to investigate these things. I note that Nordost and a few other manufacturers of recent switches have specific ports for 100M, claiming the lower speed port is lower noise and better for the audio streamer connection. Well I certainly experienced that here, and it wasn’t just audio, video content improved as well.

@sns Your change of 300 to 500mb is something different, that is the incoming internet speed of your ISP.

In my example the internet WAN port speed was still kept unchanged at 1 Gigabit. It’s all the other 4 LAN ports that were reduced in speed, the ones connecting to the NAS, the audio streamer, the wireless access point etc

A router like the Omada ER605 has the ability to configure this port by port.

I am not ready to pay big money for an audiophile switch. They seem to just be repurposed consumer switches in a fancy case with a linear power supply and sometimes clock upgrade. You could probably get the same grade of switch and add your own power supply to it.

Speaking of which.... Christiaan Punter from Hi-Fi Advice in the Netherlands is a very careful reviewer I have been following for years. A recent review of the Lejonklou Kalla network player they suggested the best sounding cheap switch is the grey colored Netgear GS108T v2 (now v3 is available).

https://www.hifi-advice.com/blog/review/digital-reviews/network-player-reviews/lejonklou-kalla-network-player/2/

Then there is the Eno and Muon ethernet filters. These seem to be getting genuinely rave reviews, even from respected people like Hans Beekhuyzen. Its possible these noise filters are doing the same job as the optical modules, not letting that high frequency noise pass, but without all the hassle.

https://www.networkacoustics.com/

Interestingly Network Acoustics had their own network switch, the Rubicon (now NLA due to supply chain issues). But on its product page they say after an exhaustive evaluation process the iFi Elite 12v was the best power supply for it, not a linear power supply, or battery supply e.t.c

https://www.networkacoustics.com/shop/rubicon-network-switch/

I mention this because I have come to a similar conclusion. I found my best LPS, the Uptone JS-2, while fantastic on streamers and other hardware, did poorly on my routers and switches. Why? Perhaps the high frequency switching chips in network hardware is too fast for an LPS to provide instant power?

@fastfreight hey nice find Fastfreight. I had not been on the JCAT site in a long time. The M12 Gold looks to be a real custom ground up design. This is what we want. The only other company I know who has built custom core switch hardware is Nordost with their just released QNET.

Both products, combined with their respective power supplies, are really getting up there in price though.

About the Muon... did you have the lower end Eno or Muon before, and if so how much better is the Muon Pro model? I too like it because its inline passive, not having to worry about yet another LPS, DC cable and AC power cable, which when done correctly literally triples the cost of any of these tweaks.

There is a lot of ethernet filters out there but the Muon seems to have the best listener reports. I was thinking to get one of these to put just before my Lindemann streamer. I want to keep the router, wireless and switch in the other room, powered on a separate line to the main system. 

@fastfreight did you end up also using the matching Network Acoustics LAN cable?

I just want to give an update on the Melco C100 ethernet cable, because what it does ties directly into network optimization.

One end of the cable has a floated ground, so its ’asymmetric’. The idea is to not pass noise downstream, instead noise can only drain one way (backwards), away from your streamer/dac e.t.c

The theory makes sense to me, so I got one in to test. Immediately out the box, with no break in (if you believe in that) it was much superior to the Nordost Heimdall 2 ethernet. You can buy 6x of the Melco for one of the Nordost cables.

A qualifier - I do not have any fancy audiophile switches yet, just a basic Netgear switch. So perhaps my network setup is noisy and the Melco is a real benefit here but in a higher end setup maybe the Nordost would be preferred.

But for now I don’t really care because for the price the C100 is an absolute bargain. I have 3x more coming to isolate the NAS/Modem/Wireless devices upstream at the router.