Melco S100 Ethernet Switch Measurements

[plissken]

7 hours ago, R1200CL said:

Yes, i watched the video. I can unplug here as well. It will play quite short. 
I suppose the point with that video is only to show that there is a buffer present ?
 

So that’s why I’m asking where is the buffer ?
I also have two Cisco here. One is SMG 300, and the other is 8 port 2960G which I haven’t been able to log into yet. (But I think I know why). 
 

Can either of those two switches be set to extend the buffers ?

My point, and it should be obvious, is that when a switch delivers data ahead of play back and you can literally pull the plug, according to the leakage current argument that a difference should be audible.

 

The second argument about phase noise is superdad either trying to put one over on people that don't know any better or it is a staggering lack of technical understanding. 

 

Phase noise can only happen during data tx.

[plissken]

29 minutes ago, Superdad said:

 

Says the network jockey playing with data analyzer bits to the chip engineer (John Swenson) of 31 years who actually designed the power networks Ethernet switch chips and PHYs and understands and measures what happens at the lowest levels.

 

Ground-plane noise is real, caused by both leakage currents and clock-threshold jitter. Impact of such transfers through the PHYs and all other chips, ultimately affecting what goes on downstream--buffers or not.

 

Clock Threshhold jitter is only happening on actively sending interfaces. The quicker the interface the smaller the window. There is occasional management frames on unused lines.

 

So your jitter argument is borderline criminal.

 

Name me the computer based setup that I can hear the difference with the cable plugged and unplugged during play back please.

[plissken]

30 minutes ago, R1200CL said:

Again, where is the buffer (that will remove phase noise, “threshold jitter,”) located ?

 

 

Same place it was last time I told you. It hasn't moved. It's also in the video I posted.

 

Does phase noise/jitter exist on a DSD or PCM file that is saved to your hard drive?

[plissken]

33 minutes ago, R1200CL said:

Where is the buffer ?

Video only telling your plugs in and out cables. And music still plays. We dont know actually the setup. We don’t know if you’re pinging the server, or something else. And for sure we don’t know where that buffer is located. 
 

There is no need to be arrogant. 
 

I’m asking so I understand if I also can find a buffer somewhere. 
 

Also will this buffer magic work for streaming services ?

And if not, why ?

 

In the video the playback software is JRiver. It has a setting for up to 1GB of buffer. Tidal will buffer entire tracks. I don't know about other streaming services.

 

Sorry I thought it was readily apparent that JRiver was the application that was playing back the music without a break even with the cabling unplugged (the continuous ping).

[plissken]

4 minutes ago, R1200CL said:

Then we must understand that buffers only applies to those that plays directly from PC to DAC via USB, with SW allowing you to set a buffer. 
 

I must be wrong?

Not correct. High end streamers like Lumin can have several minutes. Low end units may have just a few seconds.

 

Since PCM has been around awhile I would hope and expect purpose built streamers to increase in buffer size since Moore's law is in effect: That is computing transistor count doubles every 18 months. While PCM data rates have stayed the same over the past ~30 years.

[plissken]

7 hours ago, Blackmorec said:

So the improvement is nothing to do with the data coming from the network and more to do with whatever the network is injecting into the system.

 

And this where I don't think you can cash that check...

[plissken]

2 minutes ago, R1200CL said:

This is interesting, cause let’s say you have a 5 min buffer, you would also like to be able to at lest operate stop/play from that buffer. 
This will in most cases require a network connection. 
So if you’re pulling the plug, you can’t do it in that way. 

 

You sure can with Tidal and JRiver...

[plissken]

1 minute ago, R1200CL said:

I suppose @plissken can educate us if QOS and similar technology in managed switches will help.

There must be a reason why ip-phones gets a priority, so why not audio ?

 

QoS and DSCP are traffic markup technologies for priority traffic (delay sensitive, non-bursty) normally real-time. Audio playback isn't real-time so it's delay insensitive, and bursty in nature.

 

Take for example Vocera PTT devices that are used in hospitals. These are a puck that nurse and MD's wear that they can push a button and for instance ask for a crash team. This has to go over the intercom and it has to go now. So I setup IGMP at L2, PIM Sparse across the entire route to the Vocera server (this sets up the Reverse Forwarding Path) for the needed multi-cast address.

 

Next on the ingress/egress routed interfaces we set Diff-Serv code to EF and ToS to 5 to put these packet into queues that have preferential treatment by Routers and Switches. You have to do this from stem to stern. As soon as you hit a router that you don't control it's game over.

[plissken]

1 minute ago, R1200CL said:

You test has only a value in some very specific environment. And as you know there is in general accepted that a PC is the worst endpoint. With expensive exceptions. 

Maybe why you couldn’t hear the benefits of the etherRegen?

 

No I don't know that it's generally accepted that a PC is the worst end-point. In several ways I think it's the best:

 

1. I can do 10/25/50/80/100GB connections

2. I can easily do fiber

3. I can buffer entire albums

4. I can use SoX and convolution

5. I drive a 50" display with mine

6. I can do MCH Audio

 

[plissken]

1 minute ago, R1200CL said:

Nice. So we don’t need managed switches (for whatever they offers) when it comes to audio. 
 

Thanks for clarifying that. 

 

Never did for audio playback. You can still benefit from managed switches and you can get them for not a lot. My Cisco 2360 with four 10GB SFP+ was $60.

 

My IoT and security sit in their own vlans and are ACL'd off from the rest of the network. My Wireless has Mangement VLAN, House Data Vlan, Guest Vlan. My file server sits in it's own Vlan with appropriate ACL.

[plissken]

Also don't forget about wireless. It addresses all the 'Audiophile Switch' Gremlins.

 

My WiFi is two TP-Link AC1350's with the TP-Link Omada controller for 802.11k/r/v. I routinely get ~38MB/s. Way more than needed for PCM 24/192.

 

Things are $56 a pop and PoE. I put one in the laundry closet and another in an office closet.

[plissken]

3 hours ago, The Computer Audiophile said:

I'm late to this party, but for many of the measurement nuts, it means something is better engineered because it has better numbers so it must be better. 

 

I look at it as a difference that was measurable at the output of a DAC, which the objective crowd says is impossible. We can see here that it actually is possible. 

 

If the objective crowd was really into science and discovery, they'd try to determine why/how this happened. 

As miska pointed out it's a shielded cat6. So no I don't agree with your assessment. Miska is on point. 

[plissken]

13 minutes ago, The Computer Audiophile said:

Which part do you disagree with? 

That a switch is supposed to fix a chassis tied shield. You fix that with the correct cable. That's another point I've made ad nauseum.

 

What happens on the melco if you use utp and the .4 goes away?

[plissken]

10GB modules only work at 10GB. So you can't have a 10GB tranceiver on one end and 1GB on the other.

 

SFP only supports 1GB

SFP+ supports 1GB and 10GB modules

SFP56 supports 1/10/28/56GB modules

 

[plissken]

9 minutes ago, jabbr said:

Many newer modules are multi rate ie the Intel 10/1Gbe

 

Most SFP28 modules do 25/10/5/1 

 

The vendor modules I've been working with have been single rate only. I know there are dual rate 1/10GBe but they aren't as cost effective and we don't have business use cases that support the expenditure. If we are going 10GB or what have you we just standardize on it.

[plissken]

3 hours ago, Confused said:

I note the augments of some that what is measured is not audible, but I remain curious of what is actually happening, specifically what is causing what effect. 

 

Miska pointed out that they are measuring with shielded cable. This is most definitely not recommended. The UTP cabling is common mode noise rejection by design. The shield is not and it's also tying two network devices together over their ground plane. With such devices potentially plugged into different wall outlets on different circuits or even different buildings given the 100 yard length spec you get into oddities of current flow.

[plissken]

2 hours ago, firedog said:

 

The main takeaway I have from those measurements is that I'd rather get an item like the Arcam or the Mytek that doesn't need "improvement", as opposed to the Lumin, that needs "help". 

 

Keep in mind a $10 UTP cable may make all the difference for the Lumin.

[plissken]

2 hours ago, R1200CL said:

Does there even exist a consensus about how to measure a switch?

Is there a standard for it, and if so, will measurements fully cover what we’re after ?

 

It's obvious: Your ears. Ears are the gold standard poster child for many members.

 

I'm going to up my offer to $8000 to someones $1000 and travel expense paid for by the the person that's ends up incorrect. Of course this will have be after a vaccine is in wide distribution.

[plissken]

8 minutes ago, R1200CL said:

Can you link me to a SFP+ or a SFP28 that’s fits in and work with cisco 2960, etherRegen and opticalRendu ? I’m asking, cause if not to expensive, I would like to try. 

 

The 2960 is 1GB only and the FS.COM module for SX OM2 LC MM Fiber (Orange cable) is $6.

 

OM2 at 1GB is good for 550M. At 10GBe it's good for 82M. So you have future proofing built in.

[plissken]

5 minutes ago, R1200CL said:

That is not what I asked for. I’m discussing implementation of 10GB. (In order to eliminate jitter).  
I taught that was clear. 

 

I hope @jabbr will answer. 
 

 

I just frikkin told you that the 10GB modules WON'T work on the 2960.

 

THE 2960 IS 1GB ONLY.

 

You however can get the Cisco 2360 for ~$60 at Ebay (in the U.S.) that does 48 GBe copper and 4 SFP+ (1 or 10GB).

[plissken]

1 hour ago, R1200CL said:

 @jabbr says different. Let’s be patient. As he stats those modules can also give 1GB. 
 

 

No he hasn't said different. I don't need to be patient either as I do this for my full time job and just installed $2.4 Million in gear the first half of the year.

[plissken]

59 minutes ago, ASRMichael said:

Being patient is being respectful, it has nothing to do with you knowing because it’s you job or not. Maybe simple explanation how you knew the answer would have helped & kept things calm. 

 

Fair enough...

[plissken]

7 minutes ago, jabbr said:

 

I have no idea what "John's present optical design is", but consider that a 10Gbe switch can operate at 1Gbe, and realize that its the same circuit, clocks, PSU etc.

 

 

The opticalRendu, the clearfog that I've been using since ?2016 and any endpoint including PCs with NICs, can have the benefit of a 10Gbe network -- which works with 1Gbe endpoints just fine using either 1Gbe or 10/1 Gbe modules.

 

 

No doubt if you become a customer of Cisco or Mellanox etc, they will provide compliance testing. @plissken do you insist prior to a major install? I'm fine with trusting Intel, Cisco, Mellanox, HPE, Dell etc ...

 

 

In the medical field you can't install something that the vendor hasn't supplied a conformance/compliance certificate.

 

FS.COM modules are even FDA certified.

[plissken]

12 minutes ago, jabbr said:

No doubt if you become a customer of Cisco or Mellanox etc, they will provide compliance testing. @plissken do you insist prior to a major install? I'm fine with trusting Intel, Cisco, Mellanox, HPE, Dell etc ...

 

 

The GetWell system uses Trendnet. I've seen Konica Imaging install NetGear.

 

Pyxis is also using NetGear from what I recall.

[plissken]

1 minute ago, jabbr said:

 

The take home message is that if you use a 10Gbe+ compliant switch, you can be assured that the stressed eye pattern conformance testing has been done. That's the take home message for switch measurements.

 

Going further, though, suppose you wanted to see the actuall stressed eye pattern tests ... what would it take to get those?

 

Here is what FS.COM gives out:

https://community.fs.com/blog/what-kinds-of-testing-are-needed-for-transceivers.html

 

Honestly never looked. So busy it's just run and gun for me and trust that they have the required certifications.