What 300 Metres of Cat5e Actually Delivers

John Huntington published a test on 2 November that every AV integrator who has ever been asked to "just try it" should read. He took a 1000-foot box of trueCABLE Cat5e CMR, 24 AWG solid copper, terminated it to TIA-568B keystones, and then cut it down in steps from 304 metres to 130 metres, measuring link speed, throughput, Power over Ethernet and Dante audio at every length. The instruments were a Platinum Tools Net Prowler for continuity and length, a NetAlly LinkRunner AT1500 for link and PoE, and Iperf3 between two laptops against three Netgear M4250 switches, with a baseline of 940 to 943 Mbps on short patch leads.

The curve is the story. At 304 and 303 metres there was no link at all, and still none at 289, 274, 250 or 225 metres - the testers reported measurements, the switches reported nothing. At 202 metres a laptop connected directly to a switch, but negotiated 100 Mbps and moved about 90 Mbps of traffic, and switch-to-switch refused to come up. At 199 metres PoE successfully powered a Dante adapter through the cable and the device was visible to Dante Controller, though no audio passed. The first 1 Gbps switch-to-switch link appeared at 190 metres and was unstable; forcing 100 Mbps made the link more reliable and the throughput worse. At 180 metres the link settled at 1 Gbps with roughly 900 Mbps of Iperf - and Dante produced intermittent, distorted audio. Distortion persisted at 170 and 160 metres, became occasional at 150, left controller errors at 140, and disappeared at 130.5 metres, where audio ran clean for ten minutes at 1 ms latency and Iperf measured 942 Mbps.

Three findings deserve to be quoted in design reviews. First, the failure mode is graceful and therefore dangerous: the link comes up, the switch reports 1 Gbps, and audio is distorted rather than absent - which is exactly the symptom that gets blamed on the microphone, the DSP or the firmware. Second, power and data do not fail together. PoE reached 199 metres while usable data stopped well short of that, so a device can appear to be alive and be unable to stream. Third, switch-to-switch uplinks were consistently harder to establish than device-to-switch links, which is worth remembering when a design chains a switch in a rack to a switch on a stage.

Huntington's own conclusion is the one to adopt: do not deploy copper beyond 100 metres. The standard's 100 metres is 90 metres of permanent link plus 10 metres of patch, and the extra reach he measured came from one brand of cable and connector, fresh terminations, no alien crosstalk, no intermediate joints and generous bend radii - conditions no real building offers. The surplus is the margin that absorbs a bad termination, a hot ceiling void or a tight bundle. Amtune builds equipment on the assumption that the copper is compliant: HDBaseT and category-cable extenders for the runs inside the 100-metre envelope, AV-over-IP decoders where the backbone should be fibre and the last hop copper, and signal converters to bridge what the legacy cabling can still carry. Enquiries: sales@amtune.com.