Blog Viewer

Category cable twist rates

By David Meyer posted 09-16-2019 22:50


A common theme for discussion at a few of the courses I instructed at the recent Integrate Expo in Melbourne, Australia last month, and CEDIA Expo in Denver last week was that of infrastructure cabling and developments in twisted pair category cable. In fact, I was talking to some manufacturers on the show floor at Expo, and there seems to be some contention as to what the standards actually say about the twist rates in category cables — twists per foot or meter  and also which cable types are best to use. So as part of the Standards team here at CEDIA, I thought I'd help to clarify. Note this is only talking from a standards point of view, and does not factor product benchmarks above standard, or vendor specific recommendations.

Cat 5e and Cat 6 were both standardized in the same document: the 2001 release TIA/EIA-568-B.2. It states that Cat 5e is to perform to 100MHz and Cat 6 to 250MHz, and specifies things like impedance, near-end crosstalk (NEXT), DC resistance, bend radius, breaking strength, and many others. As for twist rate, it simply says under Section Pair Assembly: " The pair twist lengths shall be chosen to ensure compliance with the transmission requirements of this Standard." In other words, it's left to manufacturers to innovate. It is NOT specified.

The next year in 2002, Cat 7 was specified in the standard ISO/IEC 61156-5 as a 600MHz fully shielded (S/FTP or F/FTP) cable for use with GG45 or TERA connectors only (anyone tried terminating those?). Yet even that standard does not specify twist rate. Section 2.2.6 Cable Element simply says "The cable element shall be a pair or quad adequately twisted." However Cat 7 was never ratified by the TIA, and crucially does not specify use with modular 8P8C/RJ45 connectors. 

Then a considerable seven years later the TIA came up with something even better Cat 6A. This was standardized in TIA-568-C.2, allowing for a spec as low as UTP to be used with RJ45 connectors to perform to 500MHz for 10GbE to 100m/328ft. But even this spec, under Section 5.6.2 Pair Assembly mimics the Cat 6 spec with "The pair twist lengths shall be chosen to ensure compliance with the transmission requirements of this Standard." Again it's left to manufacturers to innovate.

Even the Cat 8 standard, first published in 2015 then updated in 2018 under ANSI/TIA-568.2-D still doesn't nominate the twist rate. Cat 8 is designed for 25/40GbE over 33m/108ft of twisted pair cable, with bandwidth up to a whopping 2000MHz (2GHz) with RJ45 connectors. Incredible! However, it offers no performance benefit over Cat 6A for anything up to 10GbE, other than of course the opportunity for more profit in selling a more premium cable! Interestingly though, this latest 2018 Cat 8 standard is the first to really cater for a Modular Plug Terminated Link (MPTL), being an infrastructure run terminated to a jack (socket) at one end, and direct to plug at the other. Other standards before it are based on a jack at both ends, with patch to devices. It's good to see such pragmatism reflecting real world use cases.

So how do manufacturer's determine the best twist rate? Well, a good one will typically base twist rate differences between pairs on prime numbers to ensure the never align, thereby mitigating inter-pair crosstalk. Foil wrap, either on the overall bunch (F/UTP) or each pair (U/FTP) will usually be sufficient to mitigate alien crosstalk and typical EMI levels for LAN, HDBaseT and AV-over-IP applications. Such foil shielding will also become increasingly important for screening with PoE as power levels increase. In high noise environments, extra shielding may be required, or of course fiber which is naturally immune from EMI.

For those of you that haven't read it, I would highly recommend checking out the recent CEDIA white paper titled Infrastructure Cabling for Voice Data and Video. It's free for members, and contains more info about cabling trends and the new TIA-570-D Residential Infrastructure Cabling standard. It's important reading.