Router and Ethernet Switch suggestions

[plissken]

T.I.'s paper on Radiated Emissions of 10/100 networking makes it abundantly clear what is the singular Achilles heal of network equipment:

 

The single ended power supply.

[plissken]

Ordered one yesterday. Will report back once I have played with it a few days.

 

WHY would you spend $400 Euro when you could spend WAY less and go fiber, and even less and go Wireless.

 

This makes no sense.

[plissken]

and my guess it is partly due to the QoS feature of D-Link.

 

Since QOS is a traffic prioritization method, and it sound's like it comes pre-configured, how can someone know what type of traffic QOS is enabled for? How does the manufacturer know what protocols and services you are using so they can configure QOS?

 

You would be better off with a layer three managed switch, and setup your audio playback chain in its own VLAN where no other traffic makes it from other non-music devices instead of attempting to QOS your entire environment.

[plissken]

They don´t! QoS on D-link switches just have the highest priority on the last 2 ports. Soundwise it is actually a big difference with my 5 port switch using port 2 vs 5 as output. If this is due to QoS is as I wrote just a guess...but it is a fact that it sounds better on the last port using Tidal as source. Go figure!

 

Why not get a $100 switch and enable QoS then? I hope you see the point I'm trying to make.

 

If QOS sounds better, then VLAN's are going to go one step further since you have established a smaller broadcast domain in addition to not having to have the Switches CPU engaged for traffic shaping and prioritization.

 

VLANs are almost CPU load neutral while QOS isn't and the CPU in the Dlink isn't anything to write home about.

[plissken]

Thanks pilissken! I will check out those VLANs. Any pointer were to start that journey? Will they work with Aries Mini?

 

I am over the moon with externally grounded and battery powered D-Link switch, Cat 6A U/UTP cables, router with galvanicly isolated linear PSU and TP Link RE450 (wireless adapter) connected to a isolation transformer with floating secondary. Sounds much better than a battery powered FMC chain. If you mean that VLAN could top this up I am all ears! ��

 

I'm not saying VLANs can top it. What I'm saying is someone that is selling an 'Audiophile grade switch' and touting QOS when Vlans are advantageous and readily available most likely doesn't know what they are talking about.

 

The reason I say this is because QOS is generally a tic mark. Vlans take some understanding to implement but they have all the advantages I've listed of a dedicated broadcast domain and minimal CPU overhead.

 

Take a managed switch and create a VLAN with 192.168.1.64/28 (255.255.255.240) bit sub-net mask would give you a broadcast domain only 16 hosts big. So the audio only portion is 192.168.1.65-78 (64 is your base network and 79 is your broadcast address).

 

Create another VLAN with 192.168.1.0/26 (255.255.255.192) with 1-62 usable. 0 is your base network and 63 is your broadcast.

 

A good router will need to support static routing. Just double check in case. Anything going on in the 192.168.1.0 network (broadcasts, chatty protocols, etc...) are stopped at the boundary of the VLAN unless you have SPECIFICALLY setup a route between them.

 

As an example: Had a business call me up with a failing POS system (basically slowed to a crawl). They were blaming the POS vendor and the vendor checked things on their end and everything was fine. Then I noticed that the diagnostic light was going solid amber at different points.

 

I would reset the router and things would be ok and then back to random lock ups. I turned on sys logging and the network was getting saturated with broadcast traffic from the DVR system. So I unplug the DVR Ethernet, I log into the switch, create a /29 and two /28 networks. Put the POS system into one of the /28's, put the DVR into another /28, and put everything else into the /29 and setup routes from the good /28 and the good /29 into the router.

 

So now I have everything up and running, I plug the DVR into the 2nd /28 and the router is stable even though the DVR is still problematic. Long story short is that the DVR had been hacked and turned into a Zombie, it was effectively doing a Denial of Service by hitting the router. My subnetting the switch and killing the route for the 2nd /28 meant the router wasn't having to respond to the broadcast and I could effectively port mirror and troubleshoot while the rest of the network functioned @ full wire speed. QOS wouldn't have helped.

[plissken]

Thanks @pilissken for that detailed info regarding VLANs. I have setup my TP Link RE450 (as wireless adapter) with 5ghz only, static IP and DHCP off. It is only communicating with the D-Link unmanaged switch via short Cat 6a U/UTP cables (50+30cm). Is this completely different to VLAN? I am not a network guru. I just know that this approach sounds awesome, even though the galvanicly isolated ac mains connections to all network devices was the real brake through.

 

Your switch is unmanaged and therefore most likely incapable of doing VLAN. Your router may be able to do VLAN. Is your RE450 your Internet router?

[plissken]

No, I have a Netgear CG3700EMR router (service providers choise) which is communicating via wireless ac (5Ghz only) to the TP Link 1750AC RE450 wireless extender (in wireless adapter mode). Cat6a U/UTP cables to D-Link DSG-105 to Auralic Aries Mini.

 

I searched the PDF for the 3700 and it doesn't mention VLAN.

[plissken]

Lol, what does VLANs have to with anything? In fact, wouldn't that just make a switch work harder to run a routing table?

 

 

Sent from my iPad using Computer Audiophile

 

Actually no. Compared to QoS, Vlans present almost no load for the router. The switch doesn't do any extra work what so ever. You do IVR on the router.

 

Additionally VLANs do some things you don't get with QoS. That is traffic from other subnets can't make it over unless routed there. It provides a significant level of isolation include the ability to apply Access Control Lists. So you can even firewall the Vlan.

[plissken]

The question becomes: If you can play back without any Ethernet connection what does the switch have to do with the SQ?

 

 

[plissken]

8 minutes ago, Confused said:

Logically, nothing whatsoever.  I'm wondering if the switch might afford some benefit with respect to reducing noise / EMI, but the 'reclclocking' aspects leave me somewhat baffled.

 

Noise floor can be easily measured. It's trivial actually. The switch isn't re-clocking it's simply using something like a 25.000000 Mhz clock vs a 25.0000 clock. It's clock is just MORE 25Mhz :-)

 

The other consideration is that the NIC in the computer is still 25.0000 and since the endpoints need to synch they can't do any better than the clock with the fewest zeros. 

[plissken]

1 hour ago, Confused said:

To be honest, I'd love to measure the noise floor for the Aqvox.  The trouble is it is not trivial for me, I do not have any test kit.

 

You only need your ear. With one switch, and no music playing, start to turn up the volume. You may or may not get tweeter hiss. Now if you have the volume turned up and you change out the switch and nothing has changed. Then nothing has changed. 

[plissken]

When I was playing around with Tidal on the PC what I noticed is that you could pull the plug after 5 to 6 seconds (I get 13MB/s down) and the entire song would play.

 

What benefit would an Audiophile switch provide me? If the song isn't relying on the switch and the NIC isn't transferring any data and I can unplug/plug shouldn't I be able to (or anyone else) to hear when this is happening?

[plissken]

2 hours ago, R1200CL said:

Well, I would think the data stream has already passed the switch, and the “damage” is already done ?

 

What 'damage'? If I take a Windows ISO image and transfer it with my normal switch and with an audiophile switch, burn to a DVD and install Windows on a machine, am I not going to end up with the same installation on both?

 

After the file has been transferred in full and the Ethernet cable pulled what's the difference?

[plissken]

Dave, this is going to fall on deaf ears most likely. Even though you are correct.

 

Music playback isn't real time, there is no timing data in the packet. GBe speeds can literally deliver a 16/44.1 track in a split second and the player application (Tidal and JRiver can do this) cache the entire track in RAM.

 

Some are behaving like there is some form of stored jitter in a file that took .5 seconds to transfer. Move up to 10GBe and you are now talking about a 0.05 second transfer.

 

This talk about TCXO and other tweaked out switches is pure lunacy.

[plissken]

I'd put my Cisco Catalyst 4948 up against that AQVOX any day of the week.

 

The AQVOX is a pig with lipstick. It's a POS D-link for pete's sake.

[plissken]

On 12/2/2017 at 12:17 PM, BigAlMc said:

Thanks for the contribution @EdmontonCanuck

 

All I can say is that there are many aspects of this hobby that don't make sense. I'll trust my ears over perceived wisdom as that approach has led to many improvements that don't all make sense at first glance. 

 

My friends laughed there asses off when I told them I was investing in clean electricity for my hifi. And completely pissed themselves when I told them how much. But how many of this forum would ignore power supplies now. 

 

Same thing for clocks at every stage of the chain. We're only just finding out how this stuff really works. 

 

Cheers, 

Alan 

 

I heard they can still get a command signaled to Voyager I

[plissken]

1 hour ago, Albrecht said:

Yes,

Very true, but irrelevant to the subject at hand: which is the quality of what is loaded into the buffer.

 

What can you possibly mean by this statement? 

 

Create two 256MB RAM disks, drive f:\ and drive g:\. 

 

Transfer a DSD or 24/192 PCM track with one cable to RAM disk f:\ and use another cable to RAM disk g:\

 

1. Dollars to donuts the MD5 hash will be the same

2. You will fail miserably when we use Foobar ABX to compare the identical track on f:\ and g:\

 

 

[plissken]

On 12/3/2017 at 7:49 PM, Milt99 said:

True.

But how does does the command string sound to Voyager?

Forget about potential packet loss the gamma, beta, zeta and x rays must be awfully fatiguing plus the wireless noise!!

 

Trust me, the fact that Voyager did 100% what was transmitted to it and it could report back will be lost on most subjectivists here.

[plissken]

21 hours ago, Superdad said:

 

Well unless your DAC or DDC has an Ethernet input (or unless you choose to use a renderer with S/PDIF output), USB regeneration still has its place—in the critical spot right before your DAC.  

 

Well, maybe a poorly designed DAC.