Why Shouldn't a Network Cable Run Longer Than 90 Meters, and What Happens If It Does?
TIA-568 limits the copper permanent link to 90 meters. What happens to the signal and PoE when a cable runs longer and how to handle a distant drop.
The finished goods warehouse was expanded toward the back of the property, and the camera perimeter grew with it. The installer ran the cables for the new cameras from the rack in the original warehouse, because the distance on the drawing looked reasonable. Nobody added up the climb to the ceiling cable tray, the detour around the forklift column or the drop down to each pole. At the handover test, every camera powered up and showed video. On the first hot night, with the infrared illuminators on, the cameras at the farthest dock start rebooting every few minutes. Weeks later, a customer claims a load came up short and asks for the shipping video. The recording has gaps in exactly the minutes of the loading, and the company absorbs the cost of the merchandise because it cannot prove what went out through that dock door. The certification done afterward shows those links were well over the standard's length limit.
TIA-568 limits the copper channel to 100 meters (328 feet): 90 meters (295 feet) of permanent link and up to 10 of patch cords. Beyond that, the signal loses margin, PoE loses power and the link fails intermittently.
The long link that powers up on handover day and fails once operations start
The distance limit is one of the requirements of the standard most often broken in a plant, because its effects take time to show up. On handover day, a long link usually comes up and the installer closes out the job. The failures arrive later, with the heat in the cable tray, the power draw of PoE devices or the real traffic of daily operations.
A 100-meter channel: 90 meters of permanent link and up to 10 of patch cords
TIA-568 divides the copper channel into two parts. The permanent link is the fixed cable running from the patch panel in the telecommunications room to the outlet, and it measures no more than 90 meters. The patch cords are the flexible sections at each end: the one connecting the patch panel to the switch and the one connecting the outlet to the device. Together they add up to 10 meters at most. The full channel, from port to port, therefore tops out at 100 meters.
Patch cords get their own budget because they are made with stranded conductors, which are more flexible and have more loss than the solid copper in the fixed cable. That is why the 10 meters apply only to the patch cords, and the fixed cable keeps its 90-meter cap. If an application needs longer patch cords, the permanent link gets shorter so the full channel stays within the limit.
Where the 100-meter figure comes from
The Ethernet standards over copper, from 10 Mb/s to 10 Gb/s, were designed to work over a channel of up to 100 meters. Using that distance as a reference, transceiver manufacturers calculated how much signal can be lost along the way, how much noise the receiver can tolerate and how long the pulses can take to arrive. The cabling standard and the Ethernet standard were written to match: if the channel meets TIA-568 for its category, the active equipment works with margin. Outside that framework, nobody guarantees how the link will behave.
What happens to the signal when the cable exceeds the limit
Insertion loss: the signal arrives weaker than the receiver expects
The signal weakens as it travels along the cable, and that loss grows with length, frequency and temperature. At 90 meters, a link of the right category arrives with the margin the receiver needs to tell signal from noise. A few extra meters eat up that margin. In an industrial building, summer heat in the cable tray adds to the length, and the link that worked in the morning starts piling up errors in the afternoon.
Crosstalk and noise: less signal against the same noise
With a weaker signal, noise carries more weight. Crosstalk between pairs and noise from drives and motors reach the receiver at the same strength, while the useful signal arrives diminished. The result is transmission errors: the switch discards damaged frames, devices retransmit and the network slows down with no visible cause. In a PLC's cyclic communication or a camera's continuous stream, those losses turn into disconnections or video with gaps.
Delay: when the four pairs stop arriving on time
Ethernet at 1 Gb/s and above transmits over all four pairs at once, and the receiver rebuilds the data from the four parts of the signal. Each pair has a different twist rate, so the signals arrive with a small time difference. The standard limits the total delay and that difference between pairs. A longer cable increases both until the receiver can no longer compensate for them.
Falling back to a lower speed: the failure nobody reports
When the link cannot sustain the highest speed, the devices may negotiate a lower one, for example 100 Mb/s instead of 1 Gb/s. The port light stays on and nobody files a ticket. The device simply runs slower: a high-resolution camera saturates its link and an access point delivers less capacity to the terminals. Checking the negotiated speed on each switch port is a quick way to find suspect links.
PoE: the distance limit is also a power limit
With PoE, the same cable carries data and power. The PoE standards calculate the power the switch delivers with an allowance for what is lost over a 100-meter channel, so the device receives what it needs. Every extra meter adds resistance. The voltage reaching the device drops, and a larger share of the energy stays in the cable as heat.
The effect shows up when demand rises. A camera that turns on its infrared at night, a PTZ dome that moves its motor or an access point running at full load all draw more power at that moment. If the cable is too long, the device reboots right when it is needed most, like the dock cameras in the opening. The extra heat also raises the signal loss of the cable and of the entire bundle.
Why the 90 meters get exceeded without anyone noticing
Distance on the drawing versus distance along the cable
Almost no link goes over the limit on purpose. The mistake starts with measuring a straight line on the drawing, when the cable follows the tray. In an industrial building, the real route adds the climb up to the ceiling tray, the detours around columns and panels, and the drop down to the machine or pole; with high ceilings, that adds many meters to each drop. On top of that comes the service loop left at the rack and at the outlet.
Long patch cords at panels and machines
On the production floor, the device-side patch cord is often the longest in the installation: it runs from the outlet on the column to the control panel, or travels through the inside of a machine to the PLC. Add a long patch cord at the rack, and the channel exceeds 100 meters even though the permanent link meets its 90. That is why patch cord lengths are defined during design, along with the location of each outlet.
How the real length of a link is measured
The footage markings printed on the cable jacket give a first reference during installation. The formal measurement is made during cabling certification: the tester reports the length of each link along with its insertion loss and delay. With that report, you know which links are close to the limit before you connect critical equipment to them.
How to handle a drop that sits far from the telecommunications room
When a drop is out of reach, the solution is to bring the network closer to the device. There are three proper ways to do it:
A telecommunications enclosure in the zone, with its own fiber optic link to the equipment room. From there, every outlet falls back within 90 meters.
Fiber optics all the way to the device or to an industrial switch inside the machine's control panel. Fiber preserves the signal over distances copper cannot cover and is immune to electrical noise.
A shorter route, when the long run comes from detours that can be avoided.
The shortcut to avoid is daisy-chaining small switches above ceilings or on columns to reach the distant drop. Each one adds a point of failure with no backup power, and together they turn a run of horizontal cabling into a makeshift backbone. Copper Ethernet extenders also fall outside the standard and only work as a temporary measure.
How we keep every drop within distance at TeleCloud
At TeleCloud we measure the real distance of every drop during the on-site survey, following the route the cable will take, with its climbs, drops and detours. With those measurements, we place the telecommunications rooms and enclosures so every outlet stays within 90 meters, and we run fiber optics to the zones far from the main room.
Our own staff installs Cat6 or Cat6A, in phases and during the windows we agree on with production. At closeout we certify every link and deliver the per-point report and the technical documentation. If your plant already has devices that reboot or ports that drop to a lower speed, we certify the existing links, find the ones that exceed the limit and propose how to bring the network closer to them.
If your plant has cameras that reboot at night, ports that drop to a lower speed or drops nobody has measured, check the distance before blaming the equipment. At TeleCloud we measure every real cable route, place enclosures so no link exceeds 90 meters and certify every drop. Request a diagnosis of your network
Frequently asked questions
Can a Cat6A or shielded cable run longer than 90 meters?
No. The limit of 90 meters of permanent link and 100 meters of channel applies equally to Cat5e, Cat6 and Cat6A, shielded or unshielded. A higher category sustains more speed within that distance, and the reach stays the same. For longer runs, the answer is fiber optics or a closer enclosure.
What should I do if a link in my plant already measures more than 90 meters and works?
Flag it as a risk point: it has little margin left under current conditions. Certify it, check the speed it negotiates and avoid connecting high-power PoE devices to it. In the next work window, fix it with a shorter route, an enclosure in the zone or a fiber link.
Does the 90-meter limit also apply to fiber optics?
No. Fiber has much longer reach, which depends on the fiber type, multimode or single-mode, and on the transmission speed. That is why it is the usual medium for the backbone and for bringing the network to zones far from the telecommunications room. In fiber, the design criterion is the loss budget of each link.
Do you need a network diagnosis?
The TeleCloud team evaluates your current infrastructure and proposes solutions adapted to your industrial operation.
Schedule free diagnosis