Melco S100 Ethernet Switch Measurements

[plissken]

Just now, R1200CL said:

Here we go again. Not easy this. Maybe I’m missing something obvious?

Or I’m not explaining myself very well ?

 

Can I have a link to one or several SFP modules that both support a 10GB switch as well as the opticalRendu? 

 

I’m asking cause then I like to purchase that 10GB switch linked to earlier, or a similar one, and connect it to my opticalRendu and maybe even etherRegen as well as my Cisco 2960. 
 

My understanding from @plisskenis this can’t be done. But now you’re suddenly indicating otherwise. 

You highly educated guys should be able to enlighten us, instead of confusing us.

 

 

 

Now I don't feel so bad :-)

 

If you have a switch with SFP you can go up to 1GB tranceivers.

 

If you have a switch with SFP+ you can go up to 10GB.

 

If you are mixing and matching interface speeds you have to match at the lowest rate tranceiver you have in the link.

 

If you have an end point that has SFP+ and a switch that has SFP the best you are going to do is 1GB. So just use 1GB optics ($6 per). You could go with a multi-rate tranceiver but your costs go up for what gain?

[plissken]

10 minutes ago, jabbr said:

I am sure the "Optical Networking and SFP" thread has these details

 

Certainly not wanting to re-create that thread here. Suffice it to say you can use OM2/OM3/OM4/OM5 cabling with the $6 SX module.

 

OM cabling is Multi-Mode

OS cabling is Single-Mode

[plissken]

2 minutes ago, firedog said:

So they should measure some commonly sold DACs at different price levels in typical setups and show us the improved results, if they exist. 

 

I've asked for what exact setup that I can recreate to hear these magical differences.

 

So baring getting 0 reply (is it really that tough of a response?) I simply have to assume that a product with only 3 returns ever (me being one of them) have to assume it's any setup.

 

But now when it comes to ANY setup, mine in particular, I get a manufacturer that whilst they show total ignorance of my DAC's capabilities, have the technical chops to call it garbage. Hence my asking what do I need to obtain... So they can comment on what I have not being resolving enough, but they can't comment on what I should purchase to experience this 'magic'.

 

I find it surreal.

[plissken]

57 minutes ago, R1200CL said:

Just to be 100% sure.

I can’t then at the same time use a normal SFP 1G in the other end (opticalRendu/etherRegen) ?

So one end (obviously the switch)  dual rate 10/1 SFP(+) module. Other end normal SFP model like we use today in our JohnS design. 
 

A no go, og good to go ?

 

If your lowest common denominator is 1gb sfp then use throughout. That's all there is to it.  For SX use om3 or om4 multi mode

[plissken]

1 hour ago, R1200CL said:

@plissken

Are you able to answer the question with a yes or no ?

(We have understood long time that normal SFP will go into a 10GB witch). 

 

If not, please let @jabbr answer. 

 

This has been answered multiple times.

 

SFP = up to 1GB

SFP+ = up to 10GB

 

You have to have compatible (speed and media type) on both ends.

 

It's that simple.  No one here that has explained this to you is having a failure to explain it correctly. Just re-read through and it should click.

[plissken]

1 hour ago, Superdad said:

You are in the wrong thread for that...

 

Shit Alex, I've asked so many times that apparently there isn't a right thread for it.

 

1 hour ago, Superdad said:

 

I thought this thread was about the $2K+ Melco switches...

 

What I find surreal is your ongoing fixation with UpTone.

Earlier you referred to (your own misunderstanding of our claims)

 

I understand your claims 100%. From either your website or your 'white paper w/ FAQ'.

 

You've yet to point out what I've misunderstood.

1 hour ago, Superdad said:

as "borderline criminal" and claimed we are "either trying to put one over on people that don't know any better or it is a staggering lack of technical understanding." 

That's quite rich given:

a) John Swenson's decades of chip and circuit designing;

b) The thousands of extremely happy customs worldwide--for John's designs both with UpTone and for Sonore (Rendu series);

c) The likelyhood that you have never in your life designed a sophisticated digital (or analog) circuit.

 

What's quite rich is that 'decades of chip and circuit designing' and still not one simple A/B measurement of supposed improvement.

 

My $8000 to $1000 is available to any of your thousands of happy customers. You should spot one of them the $1000 and then split the $8000 ;-)

 

You're correct, I don't do IC or circuit design. I architect and engineer mission critical networks with 99.999% uptime and full redundancy.

 

[plissken]

7 hours ago, Confused said:

We can all speculate, for example, I would guess that Paul Miller knows enough about ground currents that he would not connect any other equipment to the switch using a shielded cable.  But it is just a guess, so maybe not.  That said, it would definitely have been more interesting if the test was performed with the Melco cable and with a standard UTP6, but is was not (as far as we know), so we are left guessing.

 

This is my point: The lack of details about how all this was benched is THE problem.

 

 

[plissken]

5 hours ago, R1200CL said:

Still you or @jabbrhas not been able to produce a link to a working 10/1 auto SPF+ or SFP28 module. 
Preferable working together with that nice 4 port switch.

 

I most certainly have:

 

Cisco 2390 for 4 SFP+

I listed a 1GBe FS.COM SX fiber module. If your switch is only 1GBe why bother with anything else?

 

Here's the dual rate 1/10 https://www.fs.com/products/29899.html

[plissken]

1 minute ago, jabbr said:

read the datasheets, the one you quote clearly states that its multimode (SX), LX is singlemode

 

Short hand:

 

If it's S it is multimode cable

If it's L it is single mode cable

 

S is for Short Reach

L is for Long Reach

E is for Extended Reach

[plissken]

5 hours ago, R1200CL said:

Can you explain the technical requirements for this test. (Before I start a new tread about it)

 

Is it all about the etherRegen, or about a switch won’t matter in general?

 

The setup is pretty much what I layed out in the video I posted. This is predicated on the fact that these audiophile switches (Bonn, Melco, UpTone, etc) are supposed to improve everything from streamer to computer + DAC combo.

 

So what is required in my proposal:

 

1. I'll supply a Celeron 3150 based system with a dual port NIC capable of LACP LAG into a switch. I have several sub $100 switches capable: TP-Link, Cisco SG-200-8, Cisco 2390, Cisco 2960, Aruba 2530.

 

The switch will be configured for LACP and one port in the bundle will feed  port A of the NIC directly, the other will feed the audiophile switch ingress. The audiophile switch egress will go direct to port B of the NIC.

 

Playback application will be JRiver. This will feed participant supplied USB cable and what ever is down stream.

 

Cabling will be long enough to host the managed switch out of site of participant where A or B cable can be swapped.

 

I'll even do one better to stack the deck in participants favor: The generic switch side cable will be 3X the length of the audiophile switch cable.

 

As far as the evaluation: I have the idea of participant picks 3 tracks. I'll generate 7 potential swaps and randomly selected time.

 

Participant has two requirements: They need to be able to pick the time of the swap and they need to state if it's A or B.

 

I'm also open to other suggestions.  I think a few hours of the participant actively being able to manage the entire testing rig and evaluate is perfectly acceptable.

 

If they bow out or can't reliably pick then I get $1000, if they can evaluate 3 of their favorite tracks on a setup of their choosing they get $8000. Travel expenses paid by incorrect party.

[plissken]

 

51 minutes ago, MarkusBarkus said:

@plissken I don't design testing procedures professionally, so evaluate my comments and "file" accordingly, but I wonder if you use seven swaps, do you encourage a guess, because the seven "structure" strongly sets up a win/lose dichotomy.  
 

Of course, I recognize that is your point; however a tie is also a very instructional finding to the community. 

 

Additionally, the requirement to choose "when" the change occurs, seems like a "second-pass" leg of the test. A two part test.

 

My thinking is, some tracks might confuse the issue because of the composition of the music played. e.g. maybe some effects are more obvious at certain frequencies, types of music/time signatures, etc.

 

A "when" the change occurs requirement would make me want (if I had to take this test, which I would not do even with house money because I can't always hear such things anyway) to provide a track that was just a tone, for example. 
 

This brings up a corollary point: any three tracks? Even a continuous tone, or frequency sweep of some kind? Cheers, fella...

 

My attempt at structuring this is taking the input of the claims made. The claims are typically of a nature of 'immediacy' and 'apparent'.

 

This would be done in the persons own, intimately aware of it, setup. From their USB cable on back. *I don't have an issue doing this on a streamer if it supports LACP thus ensuring the network interface on said streamer will reconnect on the fly and start receiving data automatically.

 

This would be done with tracks of their own choosing. I also don't have an issue if we take three tracks, or a snippet of, and make a change between the playback. So you could even take a 30 second portion of a track and have a potential change before playback. So that is you could pick three 30 second snips, and for 7 iterations you would listen and mark down what sounded better. And did it match with a $60 switch or a $640 or $2000 switch? That's 21 samples. In this scenario I'm willing to take out the guesswork of when it happened. I think that's more than fair given the claims of stark improvement. That type of claim should be detected on the fly w/o the participant knowing if a change was made. Maybe a mix. two rounds of a known change interval and one round of random change.

 

You'd have to hit 17 of 21 correct.

 

If you want to pick frequency sweeps, multi-tone, single tone, pink noise, as your ears only track that's fine by me.

Again it will be ears only and single blind.

 

[plissken]

1 hour ago, R1200CL said:

Also no buffers

 

Not possible. They are just a critical as every other component. Won't work otherwise.

[plissken]

7 minutes ago, R1200CL said:

Is this you (almost) admitting the etherRegen highly likely will benefit Roon users or users with player SW that can’t have buffers ? 

 

You may not know it, but you are making a very large assumption.

[plissken]

1 hour ago, jabbr said:

 

That's fine. I have *never* claimed that 10Gbe fiber "sounds  better" than 1Gbe fiber. What I am saying is that *all* the compliant 10Gbe switches and NICs have been designed such that end to end jitter *is not* transmitted across links. Nor do any of the fiber devices transmit common mode noise (e.g leakage current) which is an electrical phenomenon.

 

The user whether professional or home, can be assured that the compliance sticker for 10Gbe indicates that the device has been tested for and does not pass along upstream jitter. Its a non issue. It may not be an issue for your 1Gbe devices but I haven't seen any testing.

 

For the the longest time I always glossed over your fiber thread because 1GBe copper is the standard in the home. But I have to give you credit: All that the end point manufacturers have to do is start implementing SFP+ and any mythical issues are just solved.

 

Thanks for beating the drum as long as you have on this.

[plissken]

On 9/29/2020 at 1:40 PM, Miska said:

You are still left with the higher level protocol things though

 

Left with those even if doing dial up SLIPP

[plissken]

4 hours ago, Miska said:

Only critical clock is DAC's sample clock.

 

According to @superdad you aren't correct. That the network switch phase noise and both low and high impedance leakage currents affect a DAC's clock. Apparently even when nothing is being transmitted. So all the time.

[plissken]

15 minutes ago, R1200CL said:

Would same apply for RAAT ?

 

(I’m using SonicTransporter with HQPlayer embedded). 

 

I think they agree with Miska:

 

"Audio devices must own the audio clock. Many other protocols get this wrong, including AirPlay. It's not possible for two clocks to agree perfectly. Letting the DAC control the pace of streaming removes the need for a clock-drift-compensation mechanism that is bound to increase cost, decrease sound quality, or both."

[plissken]

2 hours ago, R1200CL said:

Somehow I feel we’re drifting away for the intention (as described in John’s white paper) with better clocks for eliminating jitter in switches transferred down the chain 😀

 

The problem IS the white paper. IMO John knows better being a designer with 31 years of experience. Most switches are Store and Forward by default. That is the frame is completely reconstructed and then forwarded on. There are exceptions to this as there are cut-through and modified-cut through. Again those are the exceptions.

 

Look up 802.1p sometime.

[plissken]

2 hours ago, R1200CL said:

If that Mikrotik is powered by an LPS 1.2, ground loops shouldn’t be an issue. 

 

Ground loops aren't going to be an issue if using fiber anyways. Or WiFi.

[plissken]

42 minutes ago, jabbr said:

What do you mean "almost equal solution"? I have not seen measurements to demonstrate whether the EtherRegen can handle the stressed eye pattern test required of all compliant 10Gbe switches.

 

LOL. I got a kick out of that post... Some how the eR or other audiophile switches are somehow the gold standard.

 

I've got $8000 just burning a hole in my pocket to give away.

[plissken]

2 hours ago, R1200CL said:

Should I understand an eye pattern diagrams is the measurement you looking for when it comes to the etherRegen?

 

That's what @jabbr is looking for. I'm taking people at their word about their ears only claims.

[plissken]

1 hour ago, R1200CL said:

. Can I manipulate these buffers in any way you describe in my switches ?

2. It may be a bit more complicated as discussed here. Also buffers has jitter. 

 

Depends on the version of the 2960 and what IOS platform. Most likely you can do 802.1p. So what you can do is manipulate frame priority que.

 

Buffers don't have jitter. The read in and read out are where jitter happens.

 

But we are talking about buffer on the endpoint and not the Switch.

[plissken]

18 minutes ago, ASRMichael said:

I don’t know how your Router works but my router has switch interface for VLAN’s, inter-VLAN routing i think it’s called. Using this will give you access to all VLAN’s from any device you want to use. So your PC can be on VLAN 2 accessing VLAN 3. You have access but no broadcasting. 

 

Correct. It's called subinterfaces or multi-armed router. It's all hairpin traffic that L2 trunks up to the router and then IVR back down. If it's a good firewall then it can typically apply L2-L7 controls and inspection.