For years there’s been talk of optical fibre superseding copper, and that pulling fibre is a good way to “future proof†installations. However, history has shown us that even fibre can also be at risk of obsolescence, and that there’s plenty of life left in copper. After all, even a humble CAT5e cable can support gigabit Ethernet and HDBaseT. But there’s now a shift occurring, fuelled predominantly by our hunger for bigger and better video formats which seriously challenge the bandwidth limits of copper.
The embracing of fibre by technology integrators can be split into three broad categories;
- Those whom are already pulling, terminating and using fibre,
- Those pulling raw cable but leaving it unterminated as an optional upgrade path, or
- Those who aren’t yet working with fibre (other than TOSlink, of course) but may be curious to know more.
If you’re an integrator, which describes you? If you’re at number 3, you may be wondering what you need to do to step up to position 2, or even 1. What type of fibre cable should you pull? And how many? And if you’re already working with fibre, is the cable you’ve pulled into various sites going to be compatible with the new emerging technologies?
To best understand the landscape of optical fibre, it’s useful to look back at its evolution, and also to understand things like optical modes and wavelengths. This culminates in new standards like OM5, and new technologies to enable doing more with less; developments which go a long way to assuring far better longevity from today’s fibre cable and integration projects. Getting super technical at times is unavoidable as we delve into this, so bear with me…
The need for speed
Fibre is sometimes described as being superior because it operates at light speed. Have you heard that one? Well, yeah it sounds impressive, but this is a misnomer as it’s not about literal speed of delivery.
When describing a cable, the term speed — eg; High Speed HDMI — actually refers to the relative time interval of each digital bit in the stream. The more bits per second, the shorter the time interval of each bit. Aka; faster. But the actual delivery of the signal through the cable isn’t any faster; there’s just more bits. Data load is the real issue here, and let’s face it, we’re rather insatiable on that front!
What do we really need fibre for, anyway? CAT6 cable will support 10 Gigabit Ethernet (GbE) to 55m, and CAT6a boasts up to 100m. So, for typical residential and light commercial networking, copper is fine. However, there are two primary catalysts for fibre in the home;
- HDMI 2.1 with uncompressed signalling to 48Gbps, and with compression can support video formats which would otherwise command up to 160Gbps, and
- Network switch uplinks. Whether 1GbE or 10GbE, connecting multiple switches requires uplinks in-between to ensure the system remains non-blocking.
Fibre evolution
Back in the late 80’s, network integrators were pulling Fibre Distributed Data Interface (FDDI), a type of multi-mode fibre (MMF) cable rated to 100Mbps. Ironically, they did so with the intention to “future-proof” installations against the predicted demise of copper, yet it’s that old fibre that’s now obsolete, and copper is more prevalent than ever. Go figure!
MMF disperses the light into different modal wavelengths through the core of the fibre, which is for the most part determined by the cladding surrounding the fibre core. Another tech term here is “wavelength”, which is essentially the colour of the light. Light sources — LED or laser — operate on wavelengths from 850 nanometres (nm), being the more economical end of the spectrum, up to around 1610nm. These sit in the lower IR range, invisible to human vision, as seen in Figure 1.
Figure 1: The electromagnetic spectrum showing the wavelengths used in fibre as being just above the human visible light range. Image source: © CEDIA
By 1989, fibre technology had developed to better focus the signal, leading to the new Optical Multimode type 1 (OM1) cable. OM1 is the same size as FDDI cable with a 62.5μm core and 125μm cladding (designated as 62.5/125). OM1 was optimised for 850nm, while retaining the 1300nm wavelength for backwards compatibility.
Then came OM2 which went to a smaller 50μm core, focusing the light more to increase its bandwidth capability at 850nm. However, FDDI, OM1 and for the most part OM2 are all now redundant as they used LED as the light source, giving way to the superiority of lasers.
Modern fibre
The current specs for multi-mode fibre are OM3 and OM4, both of which were introduced for 10GbE, and can also support up to 100GbE on a single wavelength. They’re both the same physical configuration as OM2 (50/125), but are optimised for laser at 850nm. The difference between them is bandwidth, OM4 supporting more than twice that of OM3. In practice this means OM4 can go further, but in “shorter” runs (typically <300m — hey it’s all relative!) the performance is identical.
Single-mode fibre (SMF) is the high-calibre configuration, boasting very high bandwidth and lower attenuation than MMF. SMF has a much smaller core, finer than a human hair at 8-10μm, focusing the light into a single, high precision wavelength which is then capable of much greater distances. Where MMF is good for hundreds of metres, SMF can go kilometres. There are two types — Optical Single-mode type 1 (OS1), which is typically indoor rated and more tolerant to bending, and OS2 which is most commonly for outdoors with up to five times the distance capability.
The other difference with SMF over MMF is that employ higher wavelengths, with two “windows” specified, centring on 1310nm and 1550nm. The lasers and resulting devices to work with single-mode are generally a lot more expensive than for multi-mode, in addition to the added expense of the cable itself. As such, multi-mode is the way to go in residential and even most on-campus commercial installations.
Multiplexing – the emerging “new normal”
Each fibre is designed as a single optical lane driven by a single laser, which only a few years back peaked at about 10Gbps. For example, for full-duplex (bi-directional) 10GbE Ethernet, two parallel fibres were required; one for 10Gbps downstream, the second for 10Gbps upstream. It then follows that for 40GbE, 8 fibres were needed — 4x 10Gbps down, 4x 10Gbps up. But 100GbE is already rolling out at the enterprise level, and they’re starting to talk about the next generation up to 400GbE! Even with the latest 25Gbps lasers, a whopping 32 fibres would be needed. There’s got to be a better way.
Enter Wavelength Division Multiplexing (WDM). This amazing technology buys us a whole lot of expansion capability as we’ve typically only been rolling out fibre in pairs. So, what is WDM and how does it work?
Figure 2: Traditional single laser per fibre vs WDM with multiple lasers creating multiple channels over one fibre. Image source: © CEDIA
WDM employs multiple lasers (transmitters) and/or sensors (receivers) at different wavelengths crammed into transceiver modules, to create discrete “channels” each on its own wavelength over a single fibre. There are two types of WDM at the enterprise level, where such technologies tend to incubate;
- Course WDM — potentially up to 16 channels with 20nm spacing. It’s mainly for use with OS1 or OS2 fibres, with 8 channels centred on the preferred 1550nm wavelength (1470-1610nm range), then the optional extra 8 centred at the lower 1310nm wavelength (1270-1450 range).
- Dense WDM — also for use with OS1/2, DWDM has been demonstrated to support anywhere from 40 up to a whopping 88 channels with tiny 0.4-0.8nm spacing in the 1530-1565nm range. There’s already talk of going to 128 channels, each with a theoretical capacity of 100Gbps. You do the maths.
Makes the head spin, doesn’t it? The problem is that it does nothing for us in local networking as this technology is prohibitively expensive. In response, there’s a new WDM tier, being at lower wavelengths and designed especially for use with MMF cable. It’s called short wavelength division multiplexing (SWDM). Where OM3 and OM4 fibres are optimised for a single wavelength at 850nm, SWDM defines a window of 850-953nm, targeting four wavelengths with around 30nm spacing.
In application, SWDM with 10Gbps lasers and one duplex fibre pair could achieve 40GbE, or 100GbE if stepping up to the flagship 25Gbps lasers. There’s even an alliance to promote adoption and interoperability of this technology, called SWDM Alliance (see swdm.org). Members include the likes of Dell and Corning, and CEDIA member Inneos, LLC.
As a use-case example, SWDM could be applied to achieve the full capabilities of HDMI 2.1 on one pair of fibres. But we’re already seeing some innovative companies developing CWDM solutions for use with multimode fibre, massively increasing the bandwidth while reducing the number of fibre strands to just a single one. This could get us HDMI 2.1 over a single MMF, with a simple, economical installation. It’s literally more for less, so SWDM may not even be necessary.
.jpg?sfvrsn=1 "wavelength division multiplexing (WDM)")
Figure 3: Showing the use of multiple lasers in wavelength division multiplexing (WDM) Image source: © CEDIA
OM5 fibre
Yep, there’s a new OM cable in town. OM5 fibre, also known as Wide-band Multimode Fibre (WBMMF), has been ratified internationally under standards including TIA492AAAE, ISO/IEC11801-3 and ANSI/TIA-568.3-D. It’s essentially OM4 which has been optimised for SWDM.
But does that mean that installations with OM3 or OM4 are now redundant? Thankfully, no. OM3 and OM4 will still support SWDM, and they’re already being used for CWDM. But as they’re not necessarily optimised to do so, bandwidth and range may be more limited compared to using a new OM5 cable. But that might mean only a few hundred metres instead of several hundred, so still plenty for even larger scale residential applications.
Next steps and best practices
As a result of these emerging technologies, CEDIA recommended practices for optical fibre cabling are at this stage unchanged — minimum 2 strands (one duplex) of OM3 fibres, or optionally OM4 for longer runs to each location in star configuration, along with CATx etc as you’ve already been running. Where OM5 is available it might prove more scalable in time, but unless you’re talking super long runs then we’ve barely scratched the surface of what we can push down OM3.
Either way, fibre is inevitable.
David Meyer is director of technical curriculum at CEDIA