Ethernet to Fiber media converters


(Johan N) #42

Just had a look at one of my Roon endpoints, the one tiny my main listening room
It’s been connected via an FMC the last day with Roon playing many hours.
The connection is
Roon Core -> 10GEth Switch -> 1GEth Netgear GS110 TP with SFP -> TP-LINK MC220L -> Roon Bridge

All except the Roon Bridge have been up more than 30 days.
None of the devices where I can check have had any packet loss in layer 2.


#43

Just curious as I also own a GS110TP… what model 10gbe switch are you using?


(Johan N) #44

I have two Netgear GS110EMX and 2 Asus XG-U2008.
My home network is a tiny bit more complex than the description above. :slight_smile:


(Sean) #45

Yes, as mentioned this is what these experts told me (in private) and they THINK it’s due to the ethernet to optical to ethernet conversion process. If not retries they think it could be a signal integrity and/or jitter/phase noise thing where the same thinking applies - if the FMC is directly connected to a DAC internal card, the latter is working harder - again due to the optical conversion process.

I have to re-emphasise again if it wasn’t clear - nobody is reporting anything as fact. Just guessing what the technical mechanisms could be, to explain some people’s observations.

You won’t find me disagreeing/arguing with a network engineer (you) :slight_smile:

I’m not an expert but I do ask the experts and sometimes they disagree among themselves - that’s OK too.

Do you have any theories? Increased jitter due to optical conversion, creating ‘more work’ inside the DAC? SI?


(Sean) #46

When discussing retries (not dropped bits) do you mean packet loss or frame loss?


(Sean) #47

Also, if we pretend for a small moment that adding stock standard non-audiophile FMC’s directly before a networked DAC does cause the DAC’s internal network card to work a little harder (and assume we don’t know the exact mechanisms)… the bigger picture thing is some networked DACs will design for/around this better than others, with better isolation internally of the network card, maybe EM/RFI shielding internally too. So this will be a factor in different people’s experiences when adding FMC’s.

Just in case anyone reading this thinks I’m making any huge sweeping generalisations.

But we are in the tinkering section here, so all crazy theories are welcome.


(Johan N) #48

There were (and still are) no packets dropped (& no packets with errors) in layer 2.
This implicitly means there were no retries.


#49

It’s tricky and I cannot really isolate what’s what. With my BDP-1, I have both USB ports for local playback with MPD and RAAT with ethernet.

When I’m playing music off a flash drive with MPD, I can plug in and out an ethernet cable and the sound doesn’t change. Furthermore, I can load up the Manic Moose page, which is the interface page for the BDP-1 and can see on the ethernet port that it’s constantly active. It doesn’t affect the sound one bit. Zero difference in sound between no ethernet cable plugged in, ethernet cable plugged in but inactive, and ethernet cable plugged in and constantly transmitting. If the ethernet cable was bringing in noise, you’d expect something to change. Yet it doesn’t.

However, like a few other BDP-1/2 users, with MPD playback, I do notice slight differences in WAV and FLAC playback off the same flash drive. I’ve even done this blind before by setting up a shuffle playlist.

Here’s another thing. The BDP-1 has multiple USB ports. So I can have a portable 2 TB WD hard drive plugged in along with a flash drive all at the same time running off the bus power. The WD is a spinning drive and consumes more power. The flash drives consumes very little in comparison. They are both plugged in simultaneously and active, and so any effect of their power consumption or noise injected would be cumulative. However, I can go back and forth playing the same WAV file from both the hard drive and flash drive and notice a difference. The flash drive just sounds a bit tighter and cleaner. There’s couple of people who have noticed the same thing regarding playback from different drives. However, in my case I’m doing it with all of them plugged in at the same time. You might expect a difference when compared individually, but not when they are all running at the same time off the bus power. I can be playing music off the flash drive and unplug the portable drive. Nothing. No change.

Absolutely mind boggling. Common sense went out the window. Things that you assumed would make a difference, make absolutely no difference. Things that you assumed didn’t made a difference, sometimes do. Honestly, it’d be much easier to think this was all inside the head.


(Sean) #50

Noted and not to rule increased retries completely out but as mentioned (a couple times) the increased workload due to retries is only one theory. The others being increased phase noise (jitter) and lower signal integrity, all due to the optical conversion process.

Again, if not already clear to others reading this, we are deep in the weeds of guessing but it can be fun.

And again (for anyone new reading this) the assumption is always bit perfect playback (we are never discussing/assuming dropped bits).

Personally I only hear (subjective) improvements with using battery powered FMC’s with my microRendu, versus normal copper ethernet. There is a very noticeable bass increase especially. The sound quality is noticeably different, but some may not like that difference I guess. Lot’s of variables like personal preference and system gear/implementation at play.

But it’s an unwieldy connection/setup that I only pull out on weekends for super critical listening but I really do love the SQ improvement.

What are you powering your MC200L with and which Roon Bridge is connected to it? And which DAC?

If you don’t mind me asking, just out of interest only.


(Peter Lie) #51

I have two hypothesis to explain this. Although both music files are bit perfect (after FLAC decoding), on-the-fly FLAC decoding requires more CPU workload, which leads to:

  1. Higher power draw from CPU affects power supply to DAC circuitry, e.g. tiny voltage drop and/or power supply noise
  2. Increased EMI / RFI affects the DAC circuitry

A drive consumes more power from accessing than being idle. Your drive draws power from the USB bus. Therefore, the power consumption fluctuates, which again may affect the power supply to DAC circuitry. A spinning drive also introduces mechanical movement, and vibrations are bad for some types of components such as capacitors.


#52

Hi Peter,

Regarding the FLAC vs. WAV thing, yes that’s pretty much my guess. Some hear it with the BDP-1, others don’t. You can see live updates of the CPU % with the BDP-1 and it is higher with compressed formats, especially ALAC.

When I did the comparison between flash drive and hard drive, I did think of what you are saying and even did very rapid switches. The hard drive was spinning the whole time. I had my hand on top of the chassis.

I’m wondering if there are other paramaters that are being completely ignored here. Latency?


(Sean) #53

Interesting, I always wondered when there’d be a single box, high speed, ethernet input + optically isolated + ethernet output device. From the photo it appears to need 5Vdc input.

https://www.djmelectronics.com/ethernet-emi-rfi-filter.html

I don’t really know what is inside the box though, aside from marketing or until someone opens one up.

One thing I like about my pair of FMC’s is I can see my separately purchased Corning fiber optic cable between them, so I know for certain that they are optically isolated.


#54

Now that I think about latency, it might explain some of my other experiences. I’m not sure if it’s latency exactly or something else like signal integrity. Try to make sense of this:

Back when I was using Audirvana Plus on my Mid 2012 Retina MacBook Pro with a USB connected DAC, I tried playing the same WAV file from three different locations. All of these files would be buffered into the RAM.

  1. On the MacBook Pro’s internal SSD

  2. A portable hard drive attached directly

  3. NAS (hard drive attached to router) sending data wirelessly to the MacBook Pro.

There were no dropouts in either method and it was fast. Yet, I always preferred the sound most of #1 - music on the internal SSD.

Now you might think that the sound differed because in option 2 there is a big hard drive attached to it or in #3 there is more RF and network activity. To counter each of the methods, when I played music directly off the SSD directly, I would have a hard drive plugged in and copying files in the background, or I would be downloading other things and watching HD videos to keep the Wi-fi active. Still, having the file on the internal SSD sounded the best.

I know the music is buffered and played from RAM, but the location/protocol it’s coming from can make a difference. I’m not sure if I have the right words, but you guys can hopefully make sense of it.

EDIT: Just realized this Audirvana story actually does support my experience with the BDP-1 written above. That’s why when I was playing music off the flash drive, it made no difference whether there was an ethernet cable plugged in or not, or if it was even active. The data for the music itself was coming from the USB flash drive. This definitely has to count for something!

EDIT2: In the early days of when I had a BDP-1 and I was still playing music off the flash drives, but didn’t have a way to control BDP-1, I bought Powerline adapters to control the BDP-1’s playback. When I added the Powerline’s with USB playback, I didn’t notice any difference in sound. However, later on when I used the Powerline for Roon, it sounded like crap. The ethernet switches with stock PS sounded better than Powerline for Roon.

There’s definitely something here.


(Johan N) #55

HDPLEX, an old NUC i5 with passive cooling & Mytek Manhattan II. Note that I just set it up with FMC for this test.

As for jitter on layer 2, there are at least two layers of asynchronous packaged protocols eliminating the effect before it reaches DAC protocols.


(jon michaels) #56

Can anyone help with a recommendation for an inexpensive linear power supply for this:

https://www.blackbox.com/en-us/store/Detail.aspx/FlexPoint-Modular-Media-Converter-10BASE-T-100BASE-TX-to-100BASE-FX-Multimode-LC/LMC100A-LC-R2

I don’t want to spend $300. Under $100 would be nice.


(Larry Post) #57

I don’t know of a LPS for it but this is very likely total overkill for your needs. Plus at BlackBox prices, it’s a rip off.

This device can drive data up to 2 kilometers (62.1 mi.)!

Have you tried a $22 set from Amazon or eBay that uses 5V and a Jameco LPS wall wart for about $12? If not, it’s a cheap way to find out if it improves your Ethernet or not. For me, it did not, so back they went.


(jon michaels) #58

Had these as leftovers from an IT project so just putting them to use.

I’m just not sure how to select a LPS that would work with them, electronics not really my forte.


(Sean) #59

The photo in your link shows power input requirements of 5-32Vdc at 0.8Amp.

Before buying a linear PSU, do you have a mobile charger USB powerbank? If yes, which model exactly?

You could use a cable like below to connect the powerbank to the DAC side FMC as long as the DC plug fits. There are adapters anyway if you find the plug size.

If you don’t have a powerbank I recommend something like below. Even if you don’t like the FMC idea this is a quality one for topping up phone/ iPad charger on the go.

If you really like the FMC SQ then you can spend on a linear PSU or keep using the powerbank. It should give over 6 hours of playback.


(Larry Post) #60

Phew good. They probably will still need some number of ft of single mode fiber to prevent the transceiver from overdriving the receiver. Just like too much gain from a source into an amp.

A USB battery pack as suggested is certainly worth a try but would be on the ragged edge minimum voltage at 5V. I didn’t look at the pictures until now, the tech specs read 120V probably so they can sell you the branded PS for it.

I would get a bit higher than 5V with this cheap Jameco.

https://www.jameco.com/z/DDU120100H4480-AC-to-DC-Power-Supply-Wall-Adapter-Transformer-Single-Output-12-Volt-1-Amp-12-Watt_170245.html

They come in both 2.1mm and 2.5mm barrel sizes, in my experience 2.1mm is most common.

Amazon has them too, ships from Jameco in CA.


(jon michaels) #61

Thanks. Both very cool suggestions.