Optical Network Configurations

[jabbr]

Hi Jabbr,

could you please tell us more about it ?

Thks

 

I have started to implement the rollerballs and inner tube based isolation techniques that Barry Diament's has extensively discussed on the "vibration isolation" thread as well as on his website. Until I get alum 7075 bowls made I am using concave glass lenses and bearings made of chrome steel, silicon nitride, silicon carbide and tungsten carbide 12mm and 1/2"

 

This vibration isolation seems to improve the sound not only of my tube gear but digital gear as well including my optical Ethernet -- I also have the same switch as you.

 

Wondering why this would help? Most likely suspect is the piezoelectric crystal. If vibration isolation helps it makes sense, or at least raises the possibility that this is the same mechanism by which TCXO or OCXO clocks help.

 

The OEM version of the Diablo is $115 so worth a hack. I've asked Paul Pang to look into this himself. If he isn't interested in doing the mid himself I may break out my soldering iron

[tgb]

Wondering why this would help? Most likely suspect is the piezoelectric crystal. If vibration isolation helps it makes sense, or at least raises the possibility that this is the same mechanism by which TCXO or OCXO clocks help.

Hi Jabbr,

You still can think about causes & why ?...

Honestly, after 1 week tricky tests with my optical bridge to the streamer (based on 2 MC110CS plugged on the RJ45 port of the SG5412F), I've been disappointed due to harshness...

then I followed your advice & received today the TrendNet FMC, nearly the only one on the market to enable auto-negotiation 100/1000 <= critical point when you have a streamer with 100mb interface !!!

and tonight... this TrendNet brings music to its highest level.

We are so close to the "master tape", or something like that, that is really amazing !

Thanks to you again & again for having created that thread !

Great job you did !

And thanks to that "global village" to exist ! Without it... booo.... I can't imagine... :-)

 

EDIT : a few people here in France, following my thread on the french audiophile forum ordered directy the OEM version of the Diablo rather than our SG5412F. It may be the right choice.

As someone said, maybe by end 2015 & years to come, I will test sound of switches rather than DAC & amps !!!! Great time & great fun !

[jabbr]

Hi Jabbr,

You still can think about causes & why ?...

Honestly, after 1 week tricky tests with my optical bridge to the streamer (based on 2 MC110CS plugged on the RJ45 port of the SG5412F), I've been disappointed due to harshness...

then I followed your advice & received today the TrendNet FMC, nearly the only one on the market to enable auto-negotiation 100/1000 <= critical point when you have a streamer with 100mb interface !!!

and tonight... this TrendNet brings music to its highest level.

We are so close to the "master tape", or something like that, that is really amazing !

Thanks to you again & again for having created that thread !

Great job you did !

And thanks to that "global village" to exist ! Without it... booo.... I can't imagine... :-)

 

EDIT : a few people here in France, following my thread on the french audiophile forum ordered directy the OEM version of the Diablo rather than our SG5412F. It may be the right choice.

As someone said, maybe by end 2015 & years to come, I will test sound of switches rather than DAC & amps !!!! Great time & great fun !

 

You and the french audiophile forum members have done a great part in early adoption and testing of various optical systems. I certainly have learned a great deal from your experiences. It is very important for many people to test various configurations so we all gain the knowledge of what works and what doesn't and can gain an overall sense of what sounds the best!

 

I am trying to understand the factors beyond mere digital signal transmission that can affect sound. Groundplane noise,RF/EMI are all types of noise that optical networking can isolate -- this is clear. There is also (much) literature in the optical networking area regarding SI, and "jitter" in the fiberoptic signal, not for audio purposes but as network speeds are being pushed past 100Gbe (!!) these technical issues gain more importance.

 

Multimode fiber (aqua cables) works great. Singlemode (which is more expensive) is even better, particularly for long cables. This has to do with a single laser beam traveling down a single cable (less blur ~ jitter). If one has an optical system in place, it is very easy to swap out 2 SFPs at each end of a cable to switch between multimode and singlemode. Since the SFPs are relatively cheap (particularly multimode), it is clearly more important for people to first experience and become convinced of the improvement before tweaking with singlemode to get an even, but very incrementally, better experience.

 

Back to the clock. If we look at the fiberoptic switch as being very equivalent to an Uptone USB regen, then having the most stable clock is an improvement because the receiving transceiver has less "work" to do to capture and resync the signal. Particularly if the clocks are identical between the switch and computer network card (NIC). John Swenson has written about SI and noise. Implementing vibration isolation of the equipment can isolate the piezoelectric crystals which are clearly vibration sensitive (they vibrate!!).

 

So this is the basis behind what I am doing. So far, the improvements I've made have audible results and I believe other people are hearing these as well. I would never trust only my own listening because perhaps I am biased. This is why it remains essential for everyone to post their own results, and yours in particular have been most helpful (the most helpful results are where there is a problem that has been resolved)!

[tgb]

Implementing vibration isolation of the equipment can isolate the piezoelectric crystals which are clearly vibration sensitive (they vibrate!!).

So this is the basis behind what I am doing. So far, the improvements I've made have audible results and I believe other people are hearing these as well.

Hi Jabbr,

do you mean you put the FMCs and switch on rubber feet with heavy weight on it, to reduce vibration.

Is it what you've done ? leads to good improvement ?

I would not be surprised because it is what I did with my streamer with nice improvement, slightly better but better anyway.

Rgds

[jabbr]

Hi Jabbr,

do you mean you put the FMCs and switch on rubber feet with heavy weight on it, to reduce vibration.

Is it what you've done ? leads to good improvement ?

I would not be surprised because it is what I did with my streamer with nice improvement, slightly better but better anyway.

Rgds

 

No, the aluminum bowls and ball bearing, inner tube technique. I've actually been using concave glass lenses until the alum bowls are machined. This is much much better than rubber feet.

 

This post gives the essential info: http://www.computeraudiophile.com/f8-general-forum/equipment-isolation-and-vibration-damping-24805/index2.html#post436931

[best]

Hello

 

Am a novice, so before buying wanted to confirm: If my renderer is 100 base then I cannot use a 1000 base fmc? Will it not "downsample" automatically to 100 base?

Thanks

 

best

[tgb]

Hello

Am a novice, so before buying wanted to confirm: If my renderer is 100 base then I cannot use a 1000 base fmc? Will it not "downsample" automatically to 100 base?

Hi Best,

To negotiate (the right word instead of "downsample") between the switch at 1000Mb and your renderer at 100Mb :

- you need this FMC, the TrendNet TFC-1000MGA, as Jabbr gave the link previously : TRENDnet | Products | TFC-1000MGA | 100/1000BASE-T to SFP Media Converter

- you'll put in its SFP slot a 1000Mb SFP, like the ones in the switch.

This FMC has a auto-negotiation function to manage the difference in speed between your optical network based on 1Gb, and your renderer and its Ethernet interface that work at 100Mb, so on its RJ45 you'll have the 100Mb speed to discuss nicely with the renderer.

I received it last week... and it works perfectly. Put it on a power bank of course, not on its SMPS.

 

 

Warning / my experience : I have a SG5412F switch. It has RJ45 ports. If you use its RJ45 because they do auto-negotiation and so you can plug in your 100Mb renderer => it sounds harsh !

My tests show that none of RJ45 ports of optical switches must be used, only SFP ports, otherwise it sounds harsh. Because any device plugged on a RJ45 port will "pollute" the switch.

So, if auto-negotiation is required, the solution is this TrendNet FMC ; and no harshness anymore... and... you'll say "waaaah ! what's that sound ?!" :-)

Rgds

[tgb]

Hi,

FYI,

as said before, I relayed this thread on the french audiophile forum (at least the best for computer audio ; over there Phile is my name) => here :

Réseau Optique : Musique dématérialisée

From the 1st post to page 8 it was more or less about testing of my TP-Link SG5412F & brainstorming about that technology applied to audio.

But people look at that subject like a weird stuff...

 

So, I posted yesteday something to try give a clearer idea about optical network with a shopping list (because from scratch people might be confused with all these new references & specifications). It is on page 8 on this thread,

here => Réseau Optique : Musique dématérialisée - Page 8

 

It is in French... hope Google will translate it nicely to your language.

If "experts" find some mistakes (I'm not an expert at all), do not hesitate to let me know & I'll correct it.

I'm writing a more complete stuff for newbies about it, but it is going slowly due to lack of time.

If Google does not translate properly, I can make a English version on pdf for instance with links to photos & slide, just ask.

Hope it helps

Rgds

 

@ Jabbr,

thanks for your post about vibrations issue.

I've some "special feet" under my switch and it sounds very promissing (dead cheap feet of course, not audiophile ones but funny one !). I should complete these tests by end of next weekend.

Rgds

[jabbr]

tgb,

 

That's great. Please keep us informed of people's results. Regarding the "Diablo" vs. Chinese version. I don't believe the "Chinese version" is a copy, rather the real thing, that Diablo has imported and silkscreened their name onto. That is to say Diablo imports and resells, and they do a great job and have great products, and this type of business is important because it gives people a place to purchase products from Amazon, or directly in the USA. For people who have a hard time purchasing from the USA, the AliExpress site has been a good one for me to work with, and I personally have this product (labelled "GT-Tech"). It was packaged well and is built solidly. When I get around to reclocking this switch, I will be less worried about wrecking a $115 switch than a $300 switch

 

Regarding vibration isolation: there is much that it inexpensive and home grown solutions. Using an 18" bicycle inner tube underneath a 20x20" plywood platform, on top of which the aluminum bowls, then 1/2" or 12mm ball bearings and then 18x18" marble tiles is all very inexpensive. The aluminum bowls are the most specialized piece and if these can be made for $10-25 per bowl, the price is not too bad. Very inexpensively people can use a concave glass lens (35 - 50mm diameter) which is much cheaper but not as good -- still better than nothing Barry Diament has posted extensively on this subject, and I am basically following his advice. I am sure what he uses bests most other so-called audiophile platforms. He has so graciously given his vibration isolation techniques that I encourage people to at the very least purchase some of his music (which is of course outstanding in its own right): Soundkeeper Recordings

[jabbr]

Here are some "unboxing" pictures of the fiberoptic switch:

 

Note supply could be either 7.5 or 5V 2A i.e. Anker battery can supply (or LPS).

 

Circuit board. Arrows point to clocks. Red arrow to clock which looks to me to be associated with SFP switch. Yellow looks like associated with RJ-45.

 

[gldgate]

A quick e-mail to emphasize what tgb said in regards to FMC power. The SMP's that come with the units (I am using TRENDnet) are pretty awful. I noticed an immediate improvement when I put the SMP's and Ethernet switch (new Paul Pang) on battery.

[jabbr]

An update on my ASRock Q1900M NAA build. First motherboard arrived DOA, so have new one and installed Mellanox card. Working but I'm having trouble iSCSI booting Windows via Mellanox Connectx-2 card -- looks like I might need FlexBoot 3.4 to do this but card has 3.3.500 and unsure 3.4 supported on Connectx-2. if anyone has any suggestions on doing this I'm all ears -- otherwise switching to Intel x520 NIC

[Ben-M]

An update on my ASRock Q1900M NAA build. First motherboard arrived DOA, so have new one and installed Mellanox card. Working but I'm having trouble iSCSI booting Windows via Mellanox Connectx-2 card -- looks like I might need FlexBoot 3.4 to do this but card has 3.3.500 and unsure 3.4 supported on Connectx-2. if anyone has any suggestions on doing this I'm all ears -- otherwise switching to Intel x520 NIC

 

Thanks for posting your pictures of the switch and clocks. Can you remember/see what frequency they were? As I'd mentioned about, I have 2 PPA TCXO mobo clock kits that I basically can't using on my micron sized clocks on the Intel boards. But I'd be willing to give them a shot on one of these switches.

 

And I'm very interested in hearing how the Intel x520 DA2 NIC works with network boot either way. To me it looks like a great card, and on eBay the prices aren't so outlandish. I've also got Xeon's in my Control and Audio PCs, so the Intel NICs and the right PCIe port would allow me direct to cache communication.

 

Very interesting thread. I'm excited to have my part-time audiophile brother in-law over again after I get my system back together and implement more of these "enhancements". Last time I scared him away, next time he might call the psych-ward on the spot...

[YashN]

Musique dématérialisée

 

This reminds me of the recent use of 'milieu aquatique profond et standardisé' instead of 'piscine'

 

Cool forum! Might hang out there sometime.

[YashN]

No, the aluminum bowls and ball bearing, inner tube technique. I've actually been using concave glass lenses until the alum bowls are machined. This is much much better than rubber feet.

 

This post gives the essential info: http://www.computeraudiophile.com/f8-general-forum/equipment-isolation-and-vibration-damping-24805/index2.html#post436931

 

Yup, excellent idea. This can be used under your power conditioning/distribution unit too. Once I finalise my own AC Filter box, I'm sure isolating it.

 

Following the various thread on network isolation from time to time as for the moment, I have yet to find a good little device to use as an NAA.

 

Kudos for exploring all this.

[thisisvv]

I have ordered the switch. Now the wait will begin.

[jabbr]

Yup, excellent idea. This can be used under your power conditioning/distribution unit too. Once I finalise my own AC Filter box, I'm sure isolating it.

 

Following the various thread on network isolation from time to time as for the moment, I have yet to find a good little device to use as an NAA.

 

Kudos for exploring all this.

 

Kudos to you for OPing a whole number of these threads that are getting people thinking about these issues.

 

I've been playing around with a whole number of NAA type machines, and there are tradeoffs in every direction. I don't think there is one best way. Fiber does need a PCI-e slot though.

[YashN]

Kudos to you for OPing a whole number of these threads that are getting people thinking about these issues.

 

Thanks a lot jabbr. The contributions in these threads by great people like you are invaluable.

[gldgate]

Just cross-posted this on JPLAY board. Am attempting to move from optical bridge to larger scale optical network with switch and PCIe card. I'm not really that technical but really liked the results of installing original optical bridge. This is my next step. Hopefully, things will not blow up on me. Not that long ago I was having a problem even understanding the thread (FMC's, SFP's, Single/Multi Mode, LC's) and now I'm diving head first into the optical network audio unknown. Wish me luck.

 

I have just ordered an SFP PCIe card for my audio computer and one more FMC/SFP combo (All TrendNet). Will see if I can implement a fiber optic audio network. If everything works as planned, Comcast router, Control PC, Audio PC and two NAS drives will all be connected to fiber switch and I will be able to remove copper ethernet switch from network. Also, with the SFP PCIE card, every signal the Audio PC receives will go from optical switch directly into computer without the need for RJ45 cable. All FMC's and optical switch will be powered via battery. Audio PC is on LPS. Network Fiber is Single Mode. All connections into switch will be fiber (will not use the one RJ45 connection it has). Hopefully, I will be able to report results (If I don't screw something up) next weekend. The purpose of this is to try and get as clean a signal as possible into Audio PC. An optical network should remove all traces of EMI/RFI.

 

While I have purchased some audiophile networking gear from Paul Pang (switch, NIC card, RJ45 cable), recent testing shows I always prefer my system when an optical bridge is added into the mix. Will see how the move is from bridge to a larger scale optical network.

 

For those that are curious, here are links to the equipment I will be attempting to use:

 

FMC's:http://www.amazon.co...ailpage_o04_s00(4 - one for Comcast router, one for Control PC, one each for Synology and Qnap NAS)

 

SFP: http://www.amazon.co...ailpage_o04_s00(10 - combination of TrendNet, Cisco and TP-link for FMC's and Fiber PCIe)

 

Optical Switch: http://www.aliexpres...2289862298.html(1 - Most likely Diablo OEM version. Brought to my attention by jabbr on Computer Audiophile Forum

 

SFP PCIe card: http://www.amazon.co...ailpage_o03_s00(1 - for Audio PC)

 

Battery: http://www.amazon.co...ailpage_o00_s00(5 - switch plus 4 FMC's)

 

Ethernet Cable (RJ45): http://www.bluejeans...ables/index.htm(cat 6a for connecting devices to FMC's)

[Moseng]

I have successfully used a gigabit ethernet connection between a Netgear Switch (1000base-T) to a TP-Link MC220L fiber media converter (FMC) over LC-LC Duplex OM3 fiber to MC220L to DN2800MT motherboard (NAA).

 

SFPs successfully used with MC220L:

Finisar FTLX8571D3BCL (supposed to be 10g but works ...)

GLX-SX-MM-TO (Cisco compatible labeled "iNet")

 

Congratulation from me for your successful entry.

[justubes]

tgb,

 

That's great. Please keep us informed of people's results. Regarding the "Diablo" vs. Chinese version. I don't believe the "Chinese version" is a copy, rather the real thing, that Diablo has imported and silkscreened their name onto. That is to say Diablo imports and resells, and they do a great job and have great products, and this type of business is important because it gives people a place to purchase products from Amazon, or directly in the USA. For people who have a hard time purchasing from the USA, the AliExpress site has been a good one for me to work with, and I personally have this product (labelled "GT-Tech"). It was packaged well and is built solidly. When I get around to reclocking this switch, I will be less worried about wrecking a $115 switch than a $300 switch

 

Hi Jabbr,

 

Was wondering of the benefit of reclocking the switch as you have mentioned.

 

I was thinking that basically in a internet setup, any switch, optical converter, NAS even from the modem (if rendering internet music) will have jitter being introduced.

 

Has anyone actually done this to modify to a low phase noise VCXO or XO with good results?

 

In a way, switches act also as a de-jitter device but again not immune to Jitter.

 

I was of the view a better clock in a switch would make a good improvement, in my case where i render from Tidal, reclocking my router may be a good start. The problem would be that there is just no extra space to include a clock circuit and low noise PSU inside the router or switch.

[lmitche]

Hi Jabbr,

 

Was wondering of the benefit of reclocking the switch as you have mentioned.

 

I was thinking that basically in a internet setup, any switch, optical converter, NAS even from the modem (if rendering internet music) will have jitter being introduced.

 

Has anyone actually done this to modify to a low phase noise VCXO or XO with good results?

 

In a way, switches act also as a de-jitter device but again not immune to Jitter.

 

I was of the view a better clock in a switch would make a good improvement, in my case where i render from Tidal, reclocking my router may be a good start. The problem would be that there is just no extra space to include a clock circuit and low noise PSU inside the router or switch.

 

FYI - while I am not using a fiber switch, I do have two single mode fiber TPLINK FMCs. It seems that the FMCs sound better run at a 9 volts than 6 volts. Kinda like higher voltage into the REGEN,

[jabbr]

Hi Jabbr,

 

Was wondering of the benefit of reclocking the switch as you have mentioned.

 

I was thinking that basically in a internet setup, any switch, optical converter, NAS even from the modem (if rendering internet music) will have jitter being introduced.

 

Has anyone actually done this to modify to a low phase noise VCXO or XO with good results?

 

In a way, switches act also as a de-jitter device but again not immune to Jitter.

 

I was of the view a better clock in a switch would make a good improvement, in my case where i render from Tidal, reclocking my router may be a good start. The problem would be that there is just no extra space to include a clock circuit and low noise PSU inside the router or switch.

 

I've been looking into sources for low jitter clocks, either TCXO or VCXO. Many clocks seem to be made for DACs etc but not so many for CPUs/networks i.e. 25.0000Mhz. Some of the prices are out of range. Paul Pang has low jitter TCXO at a reasonable module cost. I've been in contact regarding reclocking the fiberoptic switch. At first he was hesitant because it has two clocks but thinks that reclocking the main optical switch clock is doable. Question is whether he can do, sourcing the switches direct from China, or I would do (when I get around to it

[justubes]

I've been looking into sources for low jitter clocks, either TCXO or VCXO. Many clocks seem to be made for DACs etc but not so many for CPUs/networks i.e. 25.0000Mhz. Some of the prices are out of range. Paul Pang has low jitter TCXO at a reasonable module cost. I've been in contact regarding reclocking the fiberoptic switch. At first he was hesitant because it has two clocks but thinks that reclocking the main optical switch clock is doable. Question is whether he can do, sourcing the switches direct from China, or I would do (when I get around to it

 

QUOTE=jabbr;452196]I've been looking into sources for low jitter clocks, either TCXO or VCXO. Many clocks seem to be made for DACs etc but not so many for CPUs/networks i.e. 25.0000Mhz. Some of the prices are out of range. Paul Pang has low jitter TCXO at a reasonable module cost. I've been in contact regarding reclocking the fiberoptic switch. At first he was hesitant because it has two clocks but thinks that reclocking the main optical switch clock is doable. Question is whether he can do, sourcing the switches direct from China, or I would do (when I get around to it

 

The xo's have to be order in minimum quantity and are not popular in audio as it is not a frwquescy used. A quick search shows newclassd clocks are offered in 25mhz but it get costly very very quickly. From what I see, if OMC switch used dual frequencies and require 2 per unit, ouch. But digital transmission is all about accuracy and jitter.

 

I have not looked much into optical media converters but replacing just 1 in a switch could be doable, but as I stream from tidal exclusively, I may need to look into the my optical point and router as well.

 

Certainly looks that no one has done it beside paul pangs switch which is interesting. A normal switch just uses a single 25mhz and have put myself in the groupbuy in this forum under the wewelltempereded thread.

 

@imitche

 

Using a higher voltage for the Fmc at 6v as provided or slightly higher should not cause any difference except for more heat produced and the upper limit of the regulators feeding the FMCs As circuits.Are the 2 psu the same, likely not.

 

There are no components using 6 v directly in the and almost certain the is fmc is preregulated down further by lower voltage regs feeding the various circuits and chips , some maybe as low as 1.5v. I would not try any voltage higher.

[jabbr]

I tried to put myself in the DIY audio group buy but got no replies -- I was unsure about some of the options in the order -- I would also like the boards built -- at those prices I'd get a handful or two of them