Fiber vs. Ethernet sonic differences

[TRHH]

Hi Everyone,
This is my first post. I was stunned to hear difference between my DAC connected via laptop, then a raspberry pi based streamer on LPS, then a cheap cat5 and finally a decent cat8 cable. After recently realizing that inputs can make significant differences, I have been going down the rabbit hole of fiber ever since. I was about to pull the trigger on cheap FMCs and cheap Single Mode fiber cable on Aliexpress before I read this thread.

I understand the "CISCO SFP-10G-AOC" (5Meters) sounds great. Which FMC box should I use on either ends?
Can anyone please suggest me a cheap model? Should I try a cheap one from aliexpress or buy one used - just to see if it works well? (If it works well then I could upgrade later to a FMC with superior femto clock.) Also are fiber cables always attached to this cisco model? I found a used one on ebay.

Hi

I use LevelOne GVT-2000



Torben

 

[audiobomber]

10G SFP+ Direct Attach Kable (DAC) consists of a twinax copper cable terminated with SFP+ connectors at both ends. DAC cables can be divided into two categories: passive DACs and active DACs. Both passive and active DAC cables can transmit electrical signals directly over copper cables. The active DAC cables have electrical components inside the transceivers for signal amplification.

But unlike an SFP module, there is no optical fiber conversion.

"DAC cables (Direct Attach Cable) are nothing more than two twin-axial copper cables (one for each direction) terminated to a near-empty circuit board that has SFP edge contacts. Since the signals connected to an SFP cage are just differential pairs (fiber-mode SGMII in the case of Ethernet devices), a DAC can be used for very short runs to join two pieces of gear. No PHYs, no clocks, no voltages needed at all (unless it is the Active variant of DAC--used for longer than 7-meter runs--in which case they contain a driver chip that runs needs a little voltage. But a Direct Attach Cable does not offer ANY galvanic or common-mode noise isolation between components."

Optical Network Configurations

audiophilestyle.com

 

[audiobomber]

I had been using Finisar FTLF1318P3BTL transceivers in both my oMD v2 and ER. I purchased a pair of FTLX1475D3BTL dual modes for a shootout. I found the 1475 pair to be kind of mushy sounding, much preferred the punchier 1318s. However, I found using both together was the ideal solution.

Initially I put the 1310 in the oMD with the 1475 in the ER, but the ER got quite hot, so I swapped them around. The 1475 runs cool in the oMD, and I didn't note any difference in sound quality, so that is what I am using; 1475 in the oMD to the 1318 in the ER. This combo was a clear winner for me, with the warmth of the 1475 pair, the transient snap of the 1318s, with smoother highs and better weight than the 1318s alone.

Cheap tweak, give it a try.

 

[Republicoftexas69]

I had been using Finisar FTLF1318P3BTL transceivers in both my oMD v2 and ER. I purchased a pair of FTLX1475D3BTL dual modes for a shootout. I found the 1475 pair to be kind of mushy sounding, much preferred the punchier 1318s. However, I found using both together was the ideal solution.

Initially I put the 1310 in the oMD with the 1475 in the ER, but the ER got quite hot, so I swapped them around. The 1475 runs cool in the oMD, and I didn't note any difference in sound quality, so that is what I am using; 1475 in the oMD to the 1318 in the ER. This combo was a clear winner for me, with the warmth of the 1475 pair, the transient snap of the 1318s, with smoother highs and better weight than the 1318s alone.

Cheap tweak, give it a try.

Dan, I did this modification using the 1475 in my oMDand the 1318 in the ER and had the same results. Thank you for sharing this.

 

[Gunnar]

I had been using Finisar FTLF1318P3BTL transceivers in both my oMD v2 and ER. I purchased a pair of FTLX1475D3BTL dual modes for a shootout. I found the 1475 pair to be kind of mushy sounding, much preferred the punchier 1318s. However, I found using both together was the ideal solution.

Initially I put the 1310 in the oMD with the 1475 in the ER, but the ER got quite hot, so I swapped them around. The 1475 runs cool in the oMD, and I didn't note any difference in sound quality, so that is what I am using; 1475 in the oMD to the 1318 in the ER. This combo was a clear winner for me, with the warmth of the 1475 pair, the transient snap of the 1318s, with smoother highs and better weight than the 1318s alone.

Cheap tweak, give it a try.

I have both 1318 and 1475. Didn’t like the 1475 so they are stored. For me too dark and lacking details. More or less what you say.
Will give the combination of 1318 and 1475 a try after the holidays.
Thanks for the info
Gunnar

 

[valicu2000]

I've personally tried at least 10 different SFP transcievers, but no Finisar. From my experience the best sounding ones (FS.com, CISCO) have the following traits:
- Wavelength: 1310 nm
- Fibre Type: Singlemode
- Maximum Operating Temperature: +80 C to + 85 C (industrial grade)
- Transmission Distance: 2 km to 10 km
- Cabling: duplex
- Connector type: LC (very common in switches and media converters)

Further, the optical patch cords are not built identical, they sound different. The best sounding connectors are LC Grade B due to low insertion loss. Be careful to get that OM2 cable with yellow jacket if using SM (single mode) transcievers!

I used to have 2 pcs optical fiber patch cords after the router but after replacing the one farthest from the streamer (between the router and my main switch in the house) with a DAC (direct attach cable) the sound became fuller and had an increase of speed.

 

[Republicoftexas69]

I've personally tried at least 10 different SFP transcievers, but no Finisar. From my experience the best sounding ones (FS.com, CISCO) have the following traits:
- Wavelength: 1310 nm
- Fibre Type: Singlemode
- Maximum Operating Temperature: +80 C to + 85 C (industrial grade)
- Transmission Distance: 2 km to 10 km
- Cabling: duplex
- Connector type: LC (very common in switches and media converters)

Further, the optical patch cords are not built identical, they sound different. The best sounding connectors are LC Grade B due to low insertion loss. Be careful to get that OM2 cable with yellow jacket if using SM (single mode) transcievers!

I used to have 2 pcs optical fiber patch cords after the router but after replacing the one farthest from the streamer (between the router and my main switch in the house) with a DAC (direct attach cable) the sound became fuller and had an increase of speed.

Tried the DAC cables and went back to the Finisar SFP using OC SM cable. Works best in my system and with the recommendations by Audiobomber I am very pleased, head and shoulders better than the DAC configuration in my system.

 

[audiobomber]

I have both 1318 and 1475. Didn’t like the 1475 so they are stored. For me too dark and lacking details. More or less what you say.
Will give the combination of 1318 and 1475 a try after the holidays.
Thanks for the info
Gunnar

Did you try the 1475/1318 combo Gunnar?

 

[pk_LA]

TRHH brought Grade B fiber optic cable to my attention. And Taiko Audio put me on a new track with their recommendation of the DAC cable because of its low power consumption. I did some tests that may be of interest to others. I connected the cables directly between the Server (NAS and ROON) and the Audio PC (HQPlayer) for testing.

With fiber optic cable, the goal is to achieve the lowest insertion loss and the highest return loss. The best ones are designated Grade A. To my knowledge, only Grade B is available for purchase.

In the picture below, the round UPC shape can be seen on the blue LC connector and the beveled APC shape on the green LC connector.

Now the transceivers have to fit as well. The StarTech can only UPC, the Finisar SFP+ Optical Transceiver FTLX1475D3BTL UPC and APC.

For the test, I chose a piece of music I know well Kyrie by Mercedes Sosa. The spectrogram shows the good dynamics without Loudness War. When Mercedes Sosa's voice gains sharpness and the chorus smears as mush I know something is wrong.

I did two complete runs with each of the cables and also checked for any packet loss. To make a long story short, the evaluation looks the same for all three cables. No packet loss, no degradation in latencies, a jumbo packet at long intervals, and otherwise unremarkable traffic.

No changes from the sound either. I think with UPC Grade B you are very well served with both manufacturers FS and StarTech. The connectors seem to me to be a bit higher quality than the normal product. In the long-term test of APC Grade B, the Finisar transceivers unfortunately became very hot. There were drop outs, probably because of the chip throttling in the transceiver. I can therefore not recommend APC with the Finisar transceiver.

The next test concerns two cables:

10G SFP+ Direct Attach Kable (DAC) consists of a twinax copper cable terminated with SFP+ connectors at both ends. DAC cables can be divided into two categories: passive DACs and active DACs. Both passive and active DAC cables can transmit electrical signals directly over copper cables. The active DAC cables have electrical components inside the transceivers for signal amplification.

10G SFP+ Active Optical Cable (AOC) consists of a multimode fiber optic cable terminated with SFP+ connectors at both ends and requires external power to convert the electrical signals to optical signals and vice versa.

On the left is Cisco's AOC cable; on the right is FS's 10G SFP+ passive twinax copper direct attach cable (DAC).

As expected, there were no differences in terms of data transfer rates and packet loss. Not in terms of sound either. But the temperature development was interesting. The Finisar transceivers get hot even with UPC. In comparison, the AOC cable from Cisco only gets lukewarm. There is hardly any temperature development with the DAC cable from FS.

The approach of Taiko Audio with the recommendation of a DAC cable due to the very low power consumption has something. I could not determine EMI problems with me. But the DAC cable is also short (1m) and hangs in the air. Who is unsure takes an AOC cable with fiber optics. Both are much cheaper than the usual fiber optic transceiver solutions.

My experience: While the FS.com stuff is a fraction the cost of Cisco, Juniper etc, it runs very hot comparatively and has a much higher failure rate.

 

[MarcelNL]

My experience: While the FS.com stuff is a fraction the cost of Cisco, Juniper etc, it runs very hot comparatively and has a much higher failure rate.

finisar produces gear for companies like Cisco IMHO...price and failure rates etc are linked to the quality level , the suffix BTL being the highest quality, ;lowest failure rate and highest price.