Optical Network Configurations

[JJarego]

1 minute ago, d_elm said:

I used to use D-LINK MC100CM or MC200CM, preferred 100mbps, and now use a single Netgear FS105 v2.  Fibre introduces jitter so use a short length of media.  I do not think it is clear any more that fibre is best.  The Netgear FS10x and GS10x have magnetics that provide good isolation and @JohnSwenson has found these switches to be special although he has not tested every switch.

 

Thanks @d_elm! ... really? do you prefer the result with this netgear switch? are you using a "plain vanilla" ethernet cable?

[d_elm]

1 hour ago, JJarego said:

 

Thanks @d_elm! ... really? do you prefer the result with this netgear switch? are you using a "plain vanilla" ethernet cable?

I can say the Netgear switch is as good as the fibre and is a simpler setup with one box instead of two, although this is not important to me.  I plan to have the switch moded by SOtM for better power regulation and an external clock. I am powering the FS105 v2 at 7V with an LPS -1.  I purchased that used on eBay recently.  The current FS105 is v3 and requires more than 7V.  The current GS105 can be powered with 7V.

Good power for a switch or  downstream FMC is very important.

 

Ethernet into the switch is 6 feet of plain CAT6 from a wall plate.  Upstream from that I have 70 feet of plain CAT5 and two switches.  Ethernet out of the switch is two feet of Audioquest Pearl with a JSSG faraday shield.

 

I have a wireless phone base about two feet from the back of my stereo cabinet.  Last week I received a Cornet ED88T meter that tells me the 2GHz radiation from the phone is high.  I find aluminum foil cuts the radiation to a low level so I will be hanging a piece of framed foil (stuff I purchased for a radiant barrier) behind the cabinet.

[JJarego]

Thank you @d_elm ... still, I think I must give the optical connection a try ... 

Can someone help with the "shopping basket"?

 

Thanks!  

[jabbr]

@JJarego Happy New Year. Looks like those are all compatible and you should be good to go. If you want you can power the FMC with either a LPS or a battery — I avoid cheap SMPS in my audio room. 

 

Personally I use the Intel x520 card which is available used on eBay for reasonable prices. You could entirely avoid the FMC using a switch with both SFP and RJ /copper ports — there are many ways to do it depending on where your network is located and how it is configured. 

 

Would there be an SQ difference? I’d say first get the Optical network up and running and if you hear a big SQ improvement, it might be worth swapping SFPs, switches etc. 

 

My own setup is with 10g switch (I tend to think these already have good oscillators etc, although low phase error at 10g does  not ensure lower phase error at 1g). I also have a single mode SFP connection between switch and audio PC but that’s because the single mode has better electronics. I haven’t done detailed SQ comparisons — they’d be minimal at best in my estimation — so I get reasonably good stuff and don’t sweat it. 

[JJarego]

Happy 2018, @jabbr!

 

Thank you for the sensible advice! 

 

Cheers,

João

[tboooe]

14 minutes ago, JJarego said:

Happy 2018, @jabbr!

 

Thank you for the sensible advice! 

 

Cheers,

João

Just to add my experience...I have a 2 pc HQP+NAA setup.  In my HQP PC I also use the intel x520 dual port fiber nic.  I tried adding a Cisco switch with both copper and SFP ports between my wireless bridge and HQP PC but didnt think it added anything.  Also, recently I removed the downstream FMC and this resulted in a very big increase in sound.  Finally, I bridged the network on the HQP PC.  Here is my setup now:

 

Wireless bridge >> CAT6 >> FMC >> Fiber >> HQP PC >> CAT6 >> NAA PC

 

Next I am going to try adding a switch between the HQP and NAA PC with the John Swenson grounding trick to see if it helps.

[sdolezalek]

48 minutes ago, tboooe said:

Here is my setup now:

Wireless bridge >> CAT6 >> FMC >> Fiber >> HQP PC >> CAT6 >> NAA PC

 

 

Why wouldn't you put the fiber between the HQP PC and the NAA PC?  Isn't that where you are trying to get noise insulation?  

[tboooe]

5 minutes ago, sdolezalek said:

 

Why wouldn't you put the fiber between the HQP PC and the NAA PC?  Isn't that where you are trying to get noise insulation?  

Absolutely legit question...the problem is my NAA PC does not have the capability for me to add a fiber NIC.  That is why I tried using an FMC there first.  Based on my experience, having an FMC between was more detrimental than just running straight CAT6.

[d_elm]

5 minutes ago, tboooe said:

Absolutely legit question...the problem is my NAA PC does not have the capability for me to add a fiber NIC.  That is why I tried using an FMC there first.  Based on my experience, having an FMC between was more detrimental than just running straight CAT6.

My experience with FMCs was that good power ( eg LPS-1) is needed for the downstream FMC so that the noise injected by FMC circuit was less than the noise removed by the conversion to light by the upstream FMC.  Still room for improvement in an FMC with switching regulators and a noisy clock.

[sdolezalek]

 

46 minutes ago, d_elm said:

My experience with FMCs was that good power ( eg LPS-1) is needed for the downstream FMC so that the noise injected by FMC circuit was less than the noise removed by the conversion to light by the upstream FMC.  Still room for improvement in an FMC with switching regulators and a noisy clock.

 

These are good points.  I guess one of the key questions is whether the noise actually gets into the data or whether all of the noise we should be concerned with is noise that travels in the cabling or its shielding and somehow gets to the DAC that way, affecting its ability to process the otherwise clean data signal.  In analog cables it is easier to see how noise can become part of the signal, in data cables less so. Nonetheless, I agree that using a good LPS with the downstream FMC  is important, but that doesn't directly address where we are trying to eliminate noise. 

 

Presumably, if noise can "enter" the digital bitstream then once it is there you can't get rid of it, no matter how "clean" your downstream wiring and components are.  But if "bits are bits," then presumably all the noise we need to deal with is to prevent stuff from traveling from one component to the next in the wiring/shielding, which would be one reason to use Cat 6a cabling from the FMC to DAC.  

[d_elm]

11 minutes ago, sdolezalek said:

 

 

These are good points.  I guess one of the key questions is whether the noise actually gets into the data or whether all of the noise we should be concerned with is noise that travels in the cabling or its shielding and somehow gets to the DAC that way, affecting its ability to process the otherwise clean data signal.  In analog cables it is easier to see how noise can become part of the signal, in data cables less so. Nonetheless, I agree that using a good LPS with the downstream FMC  is important, but that doesn't directly address where we are trying to eliminate noise. 

 

Presumably, if noise can "enter" the digital bitstream then once it is there you can't get rid of it, no matter how "clean" your downstream wiring and components are.  But if "bits are bits," then presumably all the noise we need to deal with is to prevent stuff from traveling from one component to the next in the wiring/shielding, which would be one reason to use Cat 6a cabling from the FMC to DAC.  

Agree, we do know  "bits are bits", as long as a digital stream's eye pattern and receiving clock are good enough to recognize the bits correctly,  and the noise introduced by a component can get into the next downstream component.   In addition to CAT6, or whatever, into the DAC I suggest installing a JSSG faraday shield.  Use 0.5 inch shielding and slieving as it will slip over the ethernet connectors.

[jabbr]

As @JJarego is planning, a direct SFP connection on the rendering PC/NAA is best to avoid these questions about copper Ethernet grounding, noise etc. 

 

If your DAC has a Linux driver, then the Solid-Run Clearfog Base is a great little choice. 

 

Otherwise PCIe SFP(+) card.

[jabbr]

Lumin just announced the X1 DAC with fiberoptic Ethernet (SFP) input:

 

https://audiobacon.net/2018/05/10/the-new-flagship-lumin-x1-fresh-output-stage-psu-and-optical-networking/

 

Roon Ready

Crystek 957 clocks

Dual ESS

Altera Cyclone IV FPGA

 

Good to see.

[Ralf11]

we need something that is not in the $12k price range...

 

good to see that it can clock a femto tho

[Superdad]

 

Just now, Ralf11 said:

we need something that is not in the $12k price range...

 

Think it is going to be more like $14-15K...

[Ralf11]

even worse - divide by 10 to get my attention

 

or 8 maybe...

[jabbr]

5 hours ago, Superdad said:

 

 

Think it is going to be more like $14-15K...

There are many price points — I’ve suggested the ClearFog feeding USB at the low but excellent end — it would be interesting to compare this + ISO Regen vs one of the Rendu family using copper Ethernet. Still at the lower end particularly if you own a USB DAC.

[Superdad]

35 minutes ago, jabbr said:

There are many price points...

 

Oh sure.  Did not mean to imply that optical Ethernet needs to be costly.  I just thought it was funny that you of all people were bringing up such an expensive unit.  Of course I know you were just using as an example that things are headed in the direction that you have been a proponent of for a long time.

The smart manufacturers read CA, so who knows, maybe Lumin got the idea from you!  

[Ralf11]

46 minutes ago, jabbr said:

There are many price points — I’ve suggested the ClearFog feeding USB at the low but excellent end — it would be interesting to compare this + ISO Regen vs one of the Rendu family using copper Ethernet. Still at the lower end particularly if you own a USB DAC.

 

if I understand this correctly, you get 2 units or a base and the other one, run fiber using the SFP port and then... ??  do you need to program it/them - I see it has a uSD slot

 

that looks like a ~$300-$500 solution - if I'm looking at the right web page for the manf. - $144 each + power supplies?

 

I'm not ready to move away from my simple Apple TV3 system - yet, and not willing to rip the ceiling out again (I do have CAT6 which I ran when I did).  But a cheap, simple opto-isolator for the end of CAT6 wiring might be a direction if I get away from Redbook (and Apple doesn't) or if my testing shows significant SQ gains over my current setup.

[jabbr]

1 minute ago, Ralf11 said:

 

if I understand this correctly, you get 2 units or a base and the other one, run fiber using the SFP port and then... ??  do you need to program it/them - I see it has a uSD slot

 

that looks like a ~$300-$500 solution - if I'm looking at the right web page for the manf. - $144 each + power supplies?

 

I'm not ready to move away from my simple Apple TV3 system - yet, and not willing to rip the ceiling out again (I do have CAT6 which I ran when I did).  But a cheap, simple opto-isolator for the end of CAT6 wiring might be a direction if I get away from Redbook (and Apple doesn't) or if my testing shows significant SQ gains over my current setup.

 

You can do CAT6 into a switch with an SFP port and fiber out (via SFP) which does a lot.  An FMC is simply a switch with one RJ-45 and one SFP, so use that. Connect it to the ClearFog with an LC-LC cable (cheap). Use this thread as a guide.

 

I loaded Ambian: http://blog.jabresearch.com/post/2017/06/04/CLEARFOG-BASE-AS-FIBEROPTIC-NETWORK-AUDIO-ADAPTER-(NAA)

 

 

[jabbr]

21 minutes ago, Superdad said:

I just thought it was funny that you of all people were bringing up such an expensive unit.  

 

Ha ha -- I applaud you -- genuinely -- for being able to bring products to market at such an affordable price point.

 

I DIY give away stuff because, among other reasons, I can't justify charging $$$.

 

The main reason I haven't commercialized my Zynq DAC interface as a standalone product is because I can't imagine being able to charge what it would take to pull me off other non-audio projects to make it commercially viable.

 

If I were to produce a DAC with an FPGA network interface etc, I don't think the price they are charging is out of the ballpark. Maybe if there were widespread adoption and large volumes the cost of development could be spread out. This stuff isn't BS cables, it is real technology which requires real development work.

 

[Superdad]

14 minutes ago, jabbr said:

The main reason I haven't commercialized my Zynq DAC interface as a standalone product is because I can't imagine being able to charge what it would take to pull me off other non-audio projects to make it commercially viable.

 

Well you have my number and my ear... 

 

14 minutes ago, jabbr said:

This stuff isn't BS cables, it is real technology which requires real development work.

 

Yup.  You are the master of understatement. 

 

By the way, wafer-level chip scale was the package type I was trying to recall on our call last week. A couple of the smaller Altera MAX 10 models are available in that package.  Requires micro-vias for the board which not all houses can do, but the possibilities and advantages are obvious (at least to you).  Of course not an ARM/FPGA hybrid like your Zynq, but useful for other logic work.

[wklie]

7 hours ago, Superdad said:

The smart manufacturers read CA, so who knows, maybe Lumin got the idea from you!  

 

I first read about about fiber for audio purpose in a Hong Kong forum and blog in 2014:

https://translate.google.com/translate?sl=auto&tl=en&js=y&prev=_t&hl=en&ie=UTF-8&u=http%3A%2F%2Fwww.avbuzz.com%2Faudio-video%2F201405%2Fnetlink03%2F0.htm&edit-text=&act=url

 

In the same year someone from Hong Kong also made an interesting custom-made product combining fiber media converters and high quality power supply to make an all-in-one Ethernet isolation box:

https://translate.google.com/translate?hl=en&sl=auto&tl=en&u=http%3A%2F%2Fediscreation.blogspot.hk%2F2014%2F03%2Ffiber-box-diy.html

 

As for the decision to include fiber optic network in Lumin X1, it is influenced by a number of positive reports from Lumin users, as well as the CA discussions.  So yes, @jabbr and others' ideas here certainly contributed to this decision.

[wklie]

This is the optical network topology used in Lumin X1 Munich 2018 demo:

 

NAS - WiFi Router - FMC - fiber - Lumin X1 SFP

FMC: TP-Link MC210CS in Auto mode
fiber: 3M Volition Single mode LC-SC Duplex patch cord (Yellow)
SFP: TP-Link TL-SM311LS

 

I also tested a Chinese equivalence of Diablo switch to be working.  Suggestions for what other popular SFP modules, FMC or switch I should do compatibility tests with are welcome.

[zoom25]

It's good to see manufacturers giving the option for fiber inputs.

 

Since Lumin will have total control for the downstream fiber implementation, I'd now be curious to test out the hypothesis regarding noise in different domains. I'd be interested in hearing comments from the manufacturers and users whether changing the power supply of the upstream FMC can make a difference or not or if the length of cables make any difference, or on the clocks of those upstream FMC. Of course, the power supplies of those upstream FMC should be on other circuits and you could furthermore put the Lumin on a good isolation transformer to get separation, so one can't claim that it's the power supply interaction rather than the signal at hand. Batteries can be used for isolation as well if needed for true isolation.

 

If there is no difference heard, then problem solved, everything works as expected. End of discussion. However, if there are differences...