Ethernet to Fiber media converters

(Peter Lie) #102

This is why we have adopted native fiber network input. People find that it sounds better than FMC.

(Mr Fix It ) #103

@Hans_Bogaert what make and model of FMC’s are you using that only cost you $30?

(Hans Bogaert) #104

I bought them direct in China via
there are many different vendors to pick from.
My selection criteria was 1Gb, single mode and compact.
Not sure if posting a link to their website will be accepted by the forum but these are the one’s I bought.

(Hans Bogaert) #105

The fact that you confirm that there is a difference and that Lumin has therefore adopted native fiber input is great for it strengthen the observations of Magnus and myself .
Do you know why this is happening? Why is the impact of FMC on PCM so much higher than on DSD ?

(Sean) #106

Fiber optic cable blocks leakage currents and is EM/RFI immune.

With the Lumin, their engineers are able to supply the fiber input section with ultra low noise power and also isolate it from other parts of the circuit (like the analogue section).

I’ll leave the hard question for Peter :grin: I haven’t noticed this myself.

(Peter Lie) #107

This is just a speculation. DSD is 2.822MHz (or higher) whereas CD is 44.1kHz. With a higher rate DSD is less prone to jitter, while 44.1kHz is easily affected by jitter (in some DAC implementations). Before the isolation, noise/interference/leakage current/ground loop worsens the DAC jitter so PCM sounds much worse than DSD in your case. After isolation, the jitter deterioration is lessened, hence the gap between PCM and DSD is closer. If you use a linear power supply for the receiver-side FMC it may sound even better.

As for @Magnus case, I speculate that before isolation, SQ degradation by jitter is worse than SQ signature change brought by DSD conversion, but after isolation, the effect of jitter becomes less pronounced than DSD conversion, which may sound “softer” in some cases - not everybody likes that.

(Hans Bogaert) #108


I have read a few times your analyses but sorry, I don’t get it.
My brain tells me that if the bandwidth goes up then the sensitivity for jitter also goes up. ( you say it goes down)
With higher bandwidth I have less time between the clock pulses to make my decision.(1 or 0 ) while with lower bandwidth I have more time. But perhaps I’m looking at a wrong parameter ?
I will do a test where I will replace the switched mode power supply with a battery to see what that does.

Thanks for sharing your thoughts

(Peter Lie) #109

See the graphs here:

(Hans Bogaert) #110

Thanks Peter,

Had a quick look at it. Reading into this topic will keep me busy for a few days :slight_smile:

(Mr Fix It ) #111

@Hans_Bogaert Thanks for the link…I found some other options there too and will try some out … cheap enough for sure. I found one set with power supplies and a shot fibre cable as a package for like US$22 a pair.

Hope some others here have a lead to some options to play with.

(Peter Lie) #112

In case I have not said it here before, when we launched Lumin X1 we held a copper vs fiber network demo at a dealer’s shop. For fiber setups different SFP modules (multi-mode, single-mode 1310nm, single-mode 1550nm) were tried. The result is that everyone agreed that different configuration sounded different, but everyone has their own preferences (just the same as every audio debate like tube vs SS) for a specific SFP module or copper Ethernet. I think more people preferred the 1310nm module than others. There were also people who preferred copper Ethernet.

I now believe that how much benefit fiber isolation brings depends on how noisy the original network is. The dealer has an optimized network where everything on the network is powered by audiophile LPS, and the switch used was a Cisco SG112-24 Compact modified with OCXO, upgraded internal voltage regulation, and again powered by an external audiophile LPS.

In typical home networks without all this expense for power supplies, I believe fiber isolation is a more cost effective way to bring improvement for most people. (There is another vendor who is coming up with an audiophile switch that claims to be superior, but it looks it is going to be much more expensive than a pair of FMC.)

(Sean) #113

This was discussed a few posts up. They are claiming their switch will do more than what a pair of FMC’s will do (i.e. more than just blocking ground / leakage currents).

(Peter Lie) #114

I don’t doubt they’re going to make a great product. I only wonder about the price.

(Sean) #115

Noted and agreed. I only shared that because it MAY not be an apples to apples comparison to compare with a pair of cheap FMC’s (like I use). It may offer more… or maybe not. We’ll have to wait.

(Magnus) #116

We have had a discussion on a Swedish HiFi forum about related issues. Since ethernet does not contain timing/clock related information, jitter in Ethernet is not applicable (only when the data becomes PCM, DSD etc can jitter be a problem). So for ethernet, electronic noise is the main (only) issue.

With this in mind, provided you have a quiet and optimized fiber converter on the DAC side (no switched voltage regulator, run from battery etc) can it realistically be improved upon? Or am I simplifying to much?

(Sean) #117

What if the optical to copper conversion process itself is noisy?

What if John Swenson chooses components that are less noisy for this conversion process, and the improvement is audible?

Or maybe this conversion process is not noisy and the modified FMC’s (like I use) is enough?

The only way to answer this will be to listen and decide.

(Magnus) #118

Yes, the TP-Link media converter I use, even with replaced switched voltage regulators, is almost certainly not optimal for audio. But if the media converter sent zero noise down to the transport (and no bit errors obviously), is it still possible to improve it?

(Sean) #119

No, you can’t improve on zero noise :grin:

Which product / solution are you aware of with zero noise on the output? Or are you working on a Swenson competing product ! :slightly_smiling_face:

As you know, my FMC’s have only have low noise linear regulators (low noise, not “no noise” remember) but the optical-to-copper conversion component/s is what was in the stock TP-Link unit… untouched. What I asked before, is what if this optical-to-copper conversion step itself is noisy?

My previous question to you, is what if there are lower noise component/s for this optical-to-copper conversion (a little more expensive) which Uptone (or someone else) could use?

Now if you’re next question is why didn’t SOtM replace those components too, then you’re basically asking them to design and build their own FMC’s, which was outside the scope of what I asked them to do and the cost would have sky rocketed. My other question in my previous reply was would this even improve SQ much? Impossible to know without trying and comparing with the Swenson switch and deciding with your ears.

(Magnus) #120

It was a purely hypothetical question: is there more to be gained from ethernet connection than removal of noise. I don’t have any immediate plans to compete with Swenson :slight_smile:

But it does make sense that Lumin choose fiber input on their high-end streamer. Than they can control everything audio-related for ethernet, and any upstream equipment don’t matter as long as it manages to send the data.

(Sean) #121

Noted but when you ask “other than removal of noise” - that itself is no trivial task.

Removal of noise is expensive. Some upgrades like I’ve done to lower noise are not expensive but these do not remove noise completely. Removal of the remaining noise (like I’ve hinted about the optical-to-copper conversion process…) gets more and more expensive…