Making light work: optimising optical

When it comes to optimising the replay of digitally stored music (whether streamed from afar or stored locally), audiophiles fall into two camps. Ok, three. No, make that four…

1)    The Fortunates: those who tried a decent network switch (just before their streamer) once and genuinely couldn’t hear any difference. Setting aside the cheap jibes they might receive from some (“your hearing must be poor”, “your system isn’t resolving enough”), these lucky audiophiles appear blessed with an audio system which is blissfully noise-free. Perhaps their system enjoys a nice clean electricity supply devoid of the noise generated by so many electric appliances in the typical home. Or perhaps they have audio kit which seems less sensitive to noise than most (anecdotally it would appear that some Linn streamers fall into this category). Either way, the Fortunates can’t hear a difference with vs without a decent switch and you might not if you went to their home.

2)    The Digital Myopics: the bits-are-bits brigade who refuse to try a switch because they “know” it can’t make a difference. The author has worked in IT for decades and knows how easy it can be to see digital audio purely through a digital lens, conveniently but erroneously ignoring the fact that ultimately digital data hits a digital-to-ANALOGUE converter and that’s where any noise will manifest itself as distortion. The sort of distortion you only realise is there when you hear your system without it. I guess we could equally call these guys the Unfortunates as they really don’t know whether they’re hearing the full potential of their system.

3)    The Switched On: those who tried a network switch and it stayed as they liked the difference it makes. They may or may not stick with the switch they first tried, and if they started with a basic generic switch then they are likely to at least audition an audio-optimised one, but a decent switch will be part of their system whatever else changes.

4)    The EnLightened: those who argue with some justification that if you really want to kill EMI/RFI noise you should forget network switches and tap into the benefits of lovely clean light by going optical; that you should drop the copper ethernet connection because noise simply can’t be conducted over an optical cable, so rather than installing an RFI antenna and then killing the noise it picks up, you should install a fibre optic network cable which eliminates the problem at source. Sounds appealing? If only it were that simple…

This article will be of interest to the last two groups as they both “get” the RFI noise thing and seek ways to mitigate it. There are real attractions to optical in theory but one or two practical challenges which might explain why the apparently compelling logic of “going optical” hasn’t to date led to optical being the de facto standard amongst committed audiophiles.

___________________________

The optical basics

[If you’re already familiar with the terminology of fibre optic networks then you may want to skip this brief overview].

The world of fibre optic networks can be far more confusing than that of copper ethernet; whereas the latter gets little more challenging than understanding the various Cat standards and deliberating over shielded vs unshielded cables, the fibreoptic world uses different terminology and has more attributes to consider. The good news is that if you choose wisely then whoever supplies your fibre optic kit will have ensured that, as a bare minimum, the components they have assembled are compatible with each other. Here are some terms you may come across:

Small Form-factor Pluggable or SFP transceiver module: this plug-in device does the actual conversion from light to electrical or vice versa.

Fibre Optic cable: these slim cables contain plastic or glass fibres which transfer information as pulses of light. Many can do so over vast distances compared with copper. They are often supplied with end caps to protect the fibres and these must be removed before inserting the cable into the SFP.

Fibre Media Converter or FMC: a case containing circuitry which enables the conversion of these pulses to and from an electrical signal (FMC’s are almost universally designed to work both ways). They typically have at least one RJ45 (ethernet) port and one SFP port.

Single Mode and Multi-Mode: in an enterprise setting, there are advantages and disadvantages to both which we won’t go into here but what’s important is compatibility: both SFP’s and fibreoptic cables should match. If we’re using Single Mode cable then we don’t want to be using Multi-Mode SFP’s and vice versa. Mismatched setups may work over the distances of a typical home network but why take the risk? You will see abbreviations like OS2 and OM3 which indicate the fibre type, wavelengths (850nm, 1300nm 1550nm), etc but if you stick with single mode or multi mode cable and SFP’s then all shall be well. Here at Reiki Audio we prefer Single Mode for its theoretical superiority though at domestic network distances of metres rather than kilometres, we’ve yet to be able to pick out one from the other.

___________________________

So why doesn’t fibre optic dominate the hifi world? Drawing on conversations with others and observations of our own, we can see at least three factors at play which may at least in part explain why optical networking in audio is far from mainstream. This paper may well stimulate discussions out of which other factors will emerge; we’ll update it if that’s the case.

1) Network inertia

Most home networks are centred around some combination of wifi and copper ethernet cable. At some point, anyone reading this will have been pleased to install their first router and to welcome broadband into their home. If we were sufficiently early adopters, the computer and perhaps printer we attached to that very first router may well have been hard-wired to the RJ45 ports on the back of the router. Simple, cheap, fast and we’re up and running. Streaming of audio and/or video is highly unlikely to have been a thing!

If we were starting today, the chances are that we’d have mainly wifi-enabled devices and there would be few if any copper ethernet cables in our homes.

In audio specifically, through experience or through received wisdom, many audiophiles prefer to dispense with wifi and go hard-wired from router to streamer, as wifi modules within  streamers etc can generate significant RFI/EMI noise in their own right. Again, there are no hard and fast rules here and people will need to experiment in their own systems. But this broad preference for non-wifi connections leads us to the emergence of the audiophile network switch as a popular approach to mitigating the drawbacks of using wired ethernet.

Many an inquisitive audiophile may have considered replacing their copper ethernet cables with fibreoptic but been discouraged by the thought of replacing cables which are difficult to access. Or they may have been deterred by the need to investigate, evaluate, match and procure from an apparently bewildering number of options for FMC’s, SFP’s and fibreoptic cables.

2) TOSLINK experiences

Many audiophiles have experienced optical connections and the disappointment sticks in their memory; this is most often TOSLINK, the optical connection between say a CD transport and a DAC. The technology is bandwidth-limited and the early optical-electrical converter circuitry was often noisy, generating more RFI than the optical connection itself eliminated.

We suspect this memory is what causes many audiophiles to turn up their noses at the very idea of fibre optic, which is of course a missed opportunity as performance can be very different today.

3) Where theory meets practice

Let’s imagine that we’ve decided to change our network to optical. We’ve worked out the right combination of FMC’s and SFP’s and cables; we buy and install them; and then we listen.

And this, where theory meets practice, is where some audiophiles may have been disappointed that an optical connection performs about the same as ethernet with a network switch.

Fibre Media Converters and SFP modules are where the magic happens, where light and electricity can be transformed into each other. The critical FMC/SFP is at the streamer end; it receives over the fibreoptic cable a series of light pulses which carry the data representing our music, completely absent of RFI noise, and proceeds to convert these pulses into an electrical signal for onward transmission to our streamer.

But RFI can still reach our streamer. How can this be? As the incoming signal is 100% noise-free, we need to look closely at the final (output) FMC/SFP device.

Optimising Optical

Ruling out noise travelling over the optical cable leaves three potential sources of RFI noise in the output (nearest to streamer) media converter; let’s see if we can address them in turn. Unsurprisingly perhaps, two of these concerns are similar to the design principles Reiki Audio employed in designing its SuperSwitch.

Environmental noise: noise from nearby devices can reach the FMC/SFP circuitry if the shield is inadequate. The best conductors make the best shields so look for a case made of aluminium or copper (or both), and one which has the bare minimum of holes in it – ideally, the only holes should be where you need to connect wires. Devices with an external power supply typically run cooler than those with an internal PSU, so the former need fewer or no holes for ventilation.

Circuitry noise: all electronic circuits generate noise as they do their job, the trick is to identify components/circuitry which are quiet in the first place and then mitigate the modest and unavoidable noise generated. High quality components, well laid out (like discrete Ethernet Port magnetics) are the way to go, and LEDs are to be avoided here as they are in a network switch. The selective application of EMI absorber will help to mitigate remaining issues.

SFP module noise: the conversion circuitry in SFP modules gets some negative press as being, relatively speaking, noisier than other common circuits. It’s a sweeping generalisation which may be unfair but perhaps this might help to explain why some audiophiles have been disappointed and, despite initial experimentation, have ultimately given up on optical and gone back to good old copper ethernet optimised by a good not-so-old audiophile network switch. Whilst we have yet to see any comparative measurements we’d be willing to rely on in selecting SFP modules, we have tried a few different SFP’s ourselves and we keep coming back to specific models from Finisar and Cisco as our go-to SFP’s, in Single Mode operation.

We also have personal experience of friends using mismatched optical components as they have dived in without fully understanding the various components they’ve assembled and, for example, have used Single Mode SFPs with Multi Mode cable (or vice versa).

We have incorporated all the above mitigation measures into the design of our Optical Bridge, and we provide SFPs and optical cable with matching specifications. We’ve also applied additional copper shielding and RFI/EMI absorber to the SFP port with positive listening results.

Ground plane noise:  a fourth source of noise seems to be emerging in audiophile circles as a potential source of concern, along with tentative solutions but, more than many other aspects of digital audio, the problem and potential solutions seem to be highly system- and context-dependent. We continue to investigate how best to address this – or, more precisely, how best to give audiophiles the ability to address it, starting with the incorporation of grounding posts into an increasing number of our products to give owners the option to explore grounding measures. Watch this space.

Closing remarks

If you’ve got this far (thank you for persevering, your dedication to the audiophile cause is impressive), you may be looking for a conclusion, even a tentative one. Well, here are a couple as our way of saying thanks:
1) The promise of optical remains valid, it just needs the right implementation. If you’ve started experimenting, keep doing so, just be clear on the hypotheses you’re testing and be prepared to share your findings for the benefit of fellow audiophiles. There will be particular interest if you conclude that an optical setup should replace a copper ethernet setup, particularly one which had already been optimised through the use of an audiophile network switch.

2) Food for thought: there may be a place for both optical FMC/SFP and a network switch in an audio-optimised network. Installed before a switch, an optical connection would do much of the heavy lifting when it comes to noise elimination, leaving the switch to simply deal with any noise arising from the optical-electrical conversion process itself. Yes, for most of us this feels excessive, but there are audiophiles who swear by daisy-chaining/cascading multiple switches despite protestations that the better the switch, the less benefit there is in doing so! Combining two different technological approaches to noise reduction feels intuitively more attractive, though we have yet to begin to explore this ourselves.