Optical Network Configurations

[plissken]

4 hours ago, barrows said:

Or, they could be counterfeit.  Do not underestimate the possibility of Chinese counterfeit products at "too good to be true" prices on eBay.

 

Or you should be able to hear the difference, no?

[plissken]

4 hours ago, jabbr said:

fs.com is a great supplier. 

 

Only issue with FS is shipping speed. I just received 280 SR LC's and it took 5 weeks. 

[plissken]

3 minutes ago, jabbr said:

Yeah but $12 for 30m lc-lc single mode?

 

The big win for single mode, aside from optics, is that the same cable can be used from 1Gbe -> 100Gbe, i.e. the cables are much cheaper and offset the increased price of optics.

 

Not my call. Existing campus cabling is all MM and they aren't going to spend $1.2 million pulling new SM. Any new fiber is SM thankfully. 

[plissken]

16 minutes ago, The Computer Audiophile said:

But my SFPs are BiDi single wire. Isn’t that the same thing on one wire?

 

SR/LR modulates TX on one strand and RX on another.

 

LC BiDi modulates Tx and Rx on each strand.

 

Many times when I'm wiring SM/MM OM3/OM4 for LC applications I have to take the cabling out of it's retaining clip and flip (crossover) the connectors and I get my link to go. 

 

I don't have to worry about this on LC BiDi. It's also way more cost effective and less error prone than MPO especially when we are doing QFSP over two strand SM cabling. 

[plissken]

Chris makes a good point to bring up LC BiDi as it simplifies everything.

 

They are $36 at FS.COM. Even considering that 10GB LC modules are ~$21, for another $30 (two modules) and only needing a single SM cable strand it makes setup for people just getting into this dead simple. 

[plissken]

12 minutes ago, lmitche said:

6 six pieces of used finisars 1310nm SPF+ are still available on Ebay at $12 each,

 

Barrows, even if they are counterfeit, they sound better than the fs.com SFPs they replace.

 

What setup are you able to hear these real-time differences on?

[plissken]

20 minutes ago, barrows said:

i would submit that such power supply perturbations are only going to have a possibility of influencing the actual sound quality (in the analog domain, loudspeakers, or headphones) IF they make their way to the USB output.  I can accept that in the case of these power supply perturbations getting to the USB output, that they may affect sound quality.  

 

Also it can't be ignored that the CPU itself is always going to be modulating power like crazy compared to a network connection. Same for all the I/O demand/load. Buffers, Interrupts, Applications.

 

My Solar Flare SFP+ NIC support two adapters. Could totally setup LACP LAG to a switch LAG and quite easily put in an FS and Finistar SFP+ module and test out in real time.

 

Knowing what I know about how non-realtime audio actually works I'm doubtful but I have asked, but have not seen, the setup that this is effecting. Still curious about that.

[plissken]

37 minutes ago, lmitche said:

That's interesting, many thanks. I was curious to know if your rationale for the SMF choice was based on something measured or observed through listening.

 

I wouldn't say that the change from MMF to SMF here is massive, but it's significant enough to justify the expense of purchasing two used SMF SFP+s, $24 for two plus $10 for 5m of fiber.. In my case these are used on a 10gbps link between a PCIE card in my server and an SPF connected USB 3 controller with two ports. The difference in clarity was immediate.

 

The SQ difference between the fs.com and finisar SFP+s  is another story and is dependent on which DAC is in use. With one, there was little difference. With the other, it had a major impact on the sound adding audible distortions to the image. That led me to hypothesize that the EMI/RFI emissions impacted the SQ on the second DAC. These distortions disappear with the Finisar SPF+s.

 

I will stop posting here now.

 

What end points? What server? What DAC? How can I attempt to recreate what you have?

 

Anyways I have some Finisar 10G LC on order.

[plissken]

21 hours ago, The Computer Audiophile said:

Or, can you recommend a good 10Gb card that will work with these modules?

 

Try https://www.ebay.com/itm/SolarFlare-SFN7002F-SFP-Dual-Port-10GbE-Flareon-PCIe-3-0-x8-Server-I-O-Adapter/303259728658?ssPageName=STRK%3AMEBIDX%3AIT&_trksid=p2057872.m2749.l2649

 

I've thrown Intel, Cisco, HPE tranceivers in them without any issue.

[plissken]

2 hours ago, The Computer Audiophile said:

Wonderful. Nobody on the Ubiquiti forum has responded to my question abut this yet. 

 

Edit: Ouch these are a little pricy. $19.95 +s&h :~)

 

But they are audiophile and correction: The shipping is free 4 day CONUS.

[plissken]

1 hour ago, charlesphoto said:

Using 10db attenuators on the send sides

You only need these for 40km optics and up.

[plissken]

3 hours ago, NYCEnglish said:

Another question: I've heard opinion that longer runs are better than shorter, any truth to this? 

 

Here is where I think that comes from: 40Km and higher optics.

 

If you use 40/80Km or greater optics you can't use them with short cables because the optics will burn out.  Single Mode and 10Km Multi-mode are fine.

[plissken]

10 hours ago, R1200CL said:

I’m interested in finding out if any of Finisar modules is using CDR. As that may reduce jitter. 

 

Jitter performance is part of the spec. The light source really doesn't matter.

 

Jitter only happens on data transfer. With 10GBe and a theoretical 1250MB/s a 50MB 16/44.1 track will transfer in 4/100ths of a second.

 

I'm now convinced we have a comprehension problem and a flat out refusal to face the realities of high speed networks and cached playback systems.

 

 

 

 

[plissken]

7 hours ago, jabbr said:

Now that the HST market has moved beyond 10Gbe, these Solarflare NICs can be picked up very cheap on eBay.

I'm running SolarFlare. The drivers are rock solid and their software suite (including remote, over the network firmware updates) is terrific.

[plissken]

14 hours ago, R1200CL said:

But, 0,0007 compared to 23 seems to make a huge difference. So I was hoping you or @plisskencould enlighten us.

 

1: I think there is a fundamental misunderstanding of 10G interface jitter vs audio clock jitter. They are wholly unrelated.

   If I transfer 1250MB 16/44.1 PCM track and buffer that on my end point all at once jitter mattered for 1 second.

   If I transfer 1250MB 16/44.1 PCM track and I buffer 1MB at a time, then jitter only mattered for 1/1250th of a second.

 

2: Ethernet, with no manipulation in place or contention, transfers at full wire right. Period. There's no are are partial rates. You will get the the highest speed of the lowest speed link in the chain (generally your ISP excepting some of you lucky bastards).

 

3: 0.0007 vs 23? What huge difference to you? To your audio?

[plissken]

8 hours ago, The Computer Audiophile said:

With audio devices one can't pick and chose what to buffer or how large/small the buffer is in the endpoint. Most manufacturers won't tell you the size of the buffer either. 

 

That depends though. I went the scalable bricks route vs all in one. So for me I have 1GB of buffer.  I have 332MB/s over 10GB fiber.

 

I think asking a manufacturer how they buffer and how much they buffer are good questions.

[plissken]

8 hours ago, The Computer Audiophile said:

The speed at which an audio endpoint can handle data is far less than wire speed, even if rated for say 1G. These are usually low power ARM devices. 

 

Given the cost of some of the streamers an embedded i86 solution would be in order. I think a $4000 should have solid 1G performance.

[plissken]

19 hours ago, R1200CL said:

Isn’t John’s white paper then more relevant.

 

Their white paper works against them however. Boiled down it makes two points:

 

1. The faster your wire speed the better (since this equates to less phase noise/jitter)

 

2. Go optical since it's a low and high impedance leakage 'moat'.

 

3. It works against them since they operate at 100MB so you have the largest amount of time of signal on wire and it's copper.

 

At least the other 'audiophile' switches I've seen give you 1GBe.

[plissken]

3 hours ago, R1200CL said:

My understanding is this is a consequence of the 10GB requirements. We would be happy with much less. Actually there is a low limit to how much speed a streamer can consume 😀

Jitter can only happen when data is x-fered. The faster the x-fer the less time you spend in the jitter-sphere.

[plissken]

3 hours ago, R1200CL said:

However, since many didn’t have that option, etherRegen served that purpose.

I really hope the future is optical.

 

This is most likely the 'many' that insist my setup sucks....

[plissken]

3 hours ago, R1200CL said:

That has been very good explained by @Superdad many times. 
You should know better. With such a comment, it seems you have an hidden agenda to put Uptone in a bad light. 

 

The only thing we have from any of the 'Audiophile' switch manufacturers is mere conjecture. I would like to assume that manufacturers took an instrumented approach to this. That is they identified and demonstrable problem and then using that data had a design approach with a before and after.

 

I do know better though. I've implemented AVB and AES64 in broadcast environments.

 

I welcome John and his 31 years of experience to any open public forum to go through this.

3 hours ago, R1200CL said:

Really ? Do an audiophile switch exist ? Aren’t they all bad copies based on John’s always willing to share his knowledge. You may have something to learn there 😉

 

I'm in a constant state of learning.

[plissken]

46 minutes ago, plissken said:

AES64

AES67

[plissken]

8 hours ago, R1200CL said:

Interesting. But since we’re not after speed, and at the same time minimum of jitter, any idea how that should be solved. (Without buffers). 
The speed is more or less decided by the music stream I would expect. 
 

 

We ARE after speed. Jitter is the variance from ideal timing of a signal. If this variance supposedly affects the DAC output you want this in as small a time window as possible.

 

The music stream is not realtime. It's already sitting somewhere in it's entirety and can 100% be transferred as fast as you can manage to local storage/buffer.

[plissken]

8 hours ago, R1200CL said:

Tidal and Qobuz, is there a way to buffer that stream ?

 

From my time with Tidal it will buffer entire tracks. I believe running Tidal through Roon changes that behavior.

[plissken]

4 minutes ago, jabbr said:

 

 

Not always. I use Roon which accepts Qobuz. Roon can upsample, but I send it to HQPlayer which upsamples and converts to SDM before sending to one of my DACs. You can use different filters and algos but keep the source audio on the NAS.

 

The source file still isn't real-time. Your on the fly manipulation is however. I know it's splitting hairs.

 

On my end point I do any convolution needed and I still can queue up 1GB at a time. This stuff is trivial for an Intel 4105 based Mainboard. I would encourage a DIY approach since you can do all of this with 4GB RAM, SSD, Silent PSU, 10GB PCIe for ~$200.