The run that breaks the budget is never the interesting one. It is the gate post at the far end of a horse paddock, the pump house, the workshop at the back of a lot, or the far corner of a self-storage site — always just far enough that the cable will not reach, and always discovered after the cable has been ordered.
The instinct is to look for something that boosts the signal. That instinct is roughly right and specifically wrong, because the 100 metre limit is not a signal-strength problem and the devices sold to solve it are not interchangeable. Power over Ethernet rides the data cabling itself — the 4-pair amendment describes raising device power "by utilizing all four pairs in the specified structured wiring plant" — which means anything that reaches further has to deal with a data link and a power feed at the same time, and most of the cheap fixes only address one of them.
That is the source of the most expensive mistake in this area. A device sold as a way to "extend PoE" may be a power injector that adds electricity to a run that is already legal in length, and nothing more. It will not buy a single additional metre, because the data link was never extended.
What follows sorts the options by what they actually do rather than by what they are called, gives the distance and bandwidth each one really delivers, works through what happens when you chain them, and ends with a table keyed to distance and camera count.
The Short Answer, in One Paragraph
Only a device that regenerates the data signal can extend a run. An inline PoE extender or repeater adds up to 100 m per side and chains to roughly four units or 500 m, but each hop costs about 5 W and the last hop will only carry a base-class camera. A powered switch at the far end adds one clean 100 m hop if there is power there. Fibre is the only option that removes the limit entirely. And a midspan injector extends nothing at all — it adds power to a run that is already within 100 m.
What the Hundred Metre Limit Actually Is
Understanding why the limit exists makes the options obvious, and it also explains why some of them fail in ways the packaging does not mention.
Where the number comes from
The 100 m figure is a worst-case performance limit rather than an arbitrary rule or a physical wall. It was derived from several characteristics acting together: insertion loss, propagation delay and delay skew, the cable's nominal velocity of propagation, operating frequency, DC resistance, and temperature.
The insertion loss figures are the clearest illustration of how little margin there is. For a full 100 m channel, the maximum insertion loss is 24 dB for Cat5e and 21.3 dB for Cat6 — and lower is better. That is the entire allowance for a compliant link.
Where the run is measured from
The 100 m is a channel, not a single length of cable. It is a permanent link of up to 90 m plus up to 10 m of patch cords, divided between the two ends — typically 5 m at each end. Measured against the termination points rather than the box, a run that "looks like" 95 m can be well past the channel limit once the two patch leads are counted.
The rule is defined jointly by IEEE 802.3, which sets the physical and link layer behaviour, and the ANSI/TIA-568 cabling standard, which defines the channel structure. Both are worth knowing by name because installer arguments about this number usually end when someone checks which one applies.
Why some cable legitimately reaches further
Distance is also affected by the cable itself and by the conditions it is installed in. The standard's own guidance notes that smaller-gauge conductors with more resistance, or an ambient temperature above 20 °C, can require a length de-rating. In the other direction, heavier-gauge cable can be run further: one vendor's extended-distance product is rated for 10 Mbps to 185 m, 100 Mbps to 150 m, and 1 Gbps to at least 100 m.
Note the pattern in that rating, because it repeats across every option in this article. As the run gets longer, the speed drops. Distance and bandwidth trade against each other, and any product claiming full speed at extreme distance is worth reading twice.
The Options, Sorted Correctly
The market names for these devices overlap badly, so the reliable way to sort them is by a single question: does the device regenerate the data signal?
| Option | Regenerates data? | Distance it adds | Typical bandwidth | Needs local power |
|---|---|---|---|---|
| Inline extender / repeater | Yes | Up to 100 m per side; chain to about 500 m | 10/100 Mbps, gigabit on some models | No — runs from the PoE input |
| Midspan power injector | No | None | Unchanged | Sometimes, at the midpoint |
| Powered switch at the far end | Yes | One new 100 m hop per switch | 10/100/1000 Mbps or higher | Yes — AC or PoE passthrough |
| Fibre with media converters | Yes | 550 m multimode; 20 km single-mode and beyond | Gigabit | Yes, at each converter |
| Long-reach / VDSL extender pair | Yes | 820 m to about 1,200 m depending on model | 100 Mbps at moderate reach, 10 Mbps at maximum | No |
Inline extenders and repeaters
An inline extender sits in the cable path between the switch and the camera. It receives the data and the power, re-times and regenerates the data signal, and re-drives it out the far side while passing the remaining power through. Each side of the connection can span up to 100 m, so one unit roughly doubles the reach.
Because it regenerates, it genuinely extends the link. The cost is that it consumes power from the very feed it is passing along, which becomes the central problem once you chain more than one — see the next section.
One caveat worth knowing before buying: not every extender can be daisy-chained, and the ones that can usually have a published hop limit.
A powered switch at the far end
This is the option almost nobody writes about, and in buildings it is often the cheapest and most reliable answer. Drop a switch at the far end and it terminates the first run and starts a fresh one. Each segment stays inside its own 100 m allowance, and the total distance is limited only by how many segments you are willing to build.
It needs one thing the extender does not: power at the far end, either from a local supply or drawn from the upstream switch where the far-end device supports PoE passthrough. If there is already a socket in the outbuilding, this beats a chain of extenders on both cost and reliability, because it also gives you a place to plug in cameras at that location later.
Fibre with media converters
Fibre is the only option that removes the limit rather than resetting it. A converter at each end translates between electrical Ethernet and optical, and the optical span is governed by the optics rather than by the copper channel: 550 m over multimode is routine, single-mode reaches 20 km and beyond, and longer-haul modules are rated far higher still.
Two things to plan for. The converters need power at both ends, and fibre cannot carry Power over Ethernet — so the cameras at the far end still need their own PoE source, which usually means a PoE switch at that end as well. Fibre is therefore a link solution, not a camera-drop solution, and it is the right answer when a whole building or cluster of cameras sits beyond reach.
Long-reach extenders over existing cable
This family uses DSL modulation or a proprietary long-reach scheme to push Ethernet over a pair that already exists, and it is delivered as a matched pair with a transmitter at each end. It is the option for reusing installed cable that is too long or too awkward to replace.
Published ratings vary widely by model and by the speed you accept. One pair is rated to 820 m with 25 W of PoE available at 550 m and 15 W at 800 m over Cat6. A longer-reach kit is rated to 1000 m, delivering 30 W at 100 Mbps out to 300 m and 15.4 W at 10 Mbps out to 800 m. Another family reaches roughly 1,200 m with an asymmetric 200/100 Mbps backbone.
Read those numbers as a menu rather than a specification. Distance, bandwidth and delivered power are three dials on the same device, and turning one up turns the others down.
Why a midspan injector is not a distance extender
This is the misconception worth correcting properly, because it costs people a cable pull they did not need to make and a device that cannot do what the listing implied.
A midspan power injector inserts power into a cable run. It does not regenerate the data signal, and it does not reset any distance counter. The IEEE standard's own requirement for a midspan is that one inserted into a channel "shall provide continuity for the signal pairs" — continuity, not regeneration.
Since the data signal is only passed straight through, the run it sits on must still fit inside the 100 m channel. A midspan is a power solution for a long-ish run, not a distance solution for a long one. If you need to add power at 60 m because you have no PoE switch, a midspan is exactly the right device. If you need to reach 180 m, it does nothing for you.
The distinction has a practical edge, too. The same term gets applied to powered switches dropped mid-run, and those genuinely do extend distance because they regenerate. Before buying anything described as a midspan, the question to ask is whether the device regenerates the data, not what it is called.
Daisy-Chaining: Why Power and Data Fail Together
Chaining extenders looks like a clean linear answer to any distance. It is not, because both the power and the data degrade with each hop, and they reach their limits at roughly the same point.
The per-hop cost
Each inline extender consumes power from the feed it is extending. The published figure is 4 to 5 W for a gigabit unit, so the arithmetic below uses 5 W per hop as the conservative case. A typical 802.3at feed of about 25 W arriving at one extender leaves roughly 20 W available downstream — a loss of about a fifth of the budget to move the signal 100 m further.
Cable loss comes on top of that, and it grows with the total length of the run rather than with the hop count, so a long chain pays both costs at once.
Worked budget for two hops
Feed from the switch (802.3at) 25.0 W
− extender 1 own consumption 5.0 W
− extender 2 own consumption 5.0 W
= power available at the camera ≈ 15 W
Fifteen watts is comfortably inside base PoE for a fixed camera, with a little margin. This is the configuration that works without drama.
Worked budget for four hops
Feed from the switch (802.3at) 25.0 W
− four extenders at 5 W each 20.0 W
= power available at the camera ≈ 5 W
Five watts will not run a fixed 4K camera with its infrared illuminators on, and it is nowhere near a PTZ or a heater-equipped model. Vendor curves published for these chains land a little better than this estimate — one chart shows about 10 W remaining at four extenders, and a specific four-unit chain is rated to deliver up to 12 W in the base PoE class at the last hop — but every published version of this arithmetic points the same way.
The practical ceiling is four units and roughly 500 m, and by that point the chain is carrying base-class power only.
Why the data link fails at the same time as the power
The power arithmetic gets all the attention, but the data side has its own ceiling and it arrives at a similar place. Every hop is a re-timing opportunity and a re-negotiation, and each one has to succeed for the link to come up at all. Some extenders also step down the negotiated speed, so a chain that works fine for a couple of low-bitrate streams may refuse to carry four high-resolution ones.
The failure is also less graceful than the power case. If a mid-run extender loses its feed, the regenerator stops driving the signal and everything downstream drops with it. Most inline units take their power from the run they are extending rather than from a local supply, so a single upstream fault takes out the whole chain rather than just the segment behind it.
What a Mid-Run Device Does to the Budget and the Enclosure
Once a device is sitting outside the building, three things change that the distance tables do not mention.
Accounting for the extender's own draw
The extender's self-consumption is part of the switch's power budget, not a separate line item. Five watts per hop across four hops is 20 W of the switch's total budget spent on equipment rather than cameras, which matters on a switch sized close to its camera load. Long-range units are far lighter — one 1000 m kit is specified at 0.4 W idle and 0.7 W under full load — so the penalty is specific to the gigabit in-line repeaters rather than to the whole category.
The passthrough class also matters. A chain fed by an 802.3bt source does not deliver 802.3bt at the far end: one long-range kit accepts up to 95 W of 802.3bt input and delivers 30 W to an 802.3at device or 15 W to an 802.3af device, expressly noting that the output is limited by distance. If the camera at the end needs more than base PoE, it needs fewer hops or a higher-input device.
Powering a mid-run device outdoors
An outdoor-rated enclosure is not optional for a mid-run module. Published long-range outdoor kits are typically specified to IP67 with an operating range of about −10 to 55 °C, and the general guidance for mid-run hardware includes impact-rated housings. Where the cable itself is exposed, running it in conduit is the usual advice, and it also protects against the sun damage that shortens cable life on exterior runs.
Surge protection and grounding
A run that leaves the building and comes back, or that spans between two structures, has a much larger exposure to induced surges than an indoor run of the same length. Protection at both ends is the standard recommendation, and it matters more here because a mid-run device gives a surge somewhere to do damage in the middle of the path rather than only at the endpoints. Grounding is not decorative on these installations: an ungrounded shield or an ungrounded device chassis can turn a small disturbance into a failure.
Choosing by Distance and Camera Count
| Distance beyond the recorder | Cameras there | Recommended approach | Why |
|---|---|---|---|
| Under 100 m | any | Direct cable | No device needed; verify the channel length including patch cords |
| 100–150 m | 1–2 | Switch extend mode, if supported, or a single extender | Cheapest fix; expect a reduced speed at the longest setting |
| 150–300 m | 1–3 | One or two inline extenders | Adds 100 m per hop; keep cameras at base PoE class |
| 300–500 m | 2–4 | Four-hop extender chain, or a powered switch at the far end | By four hops you are down to base-class power; a far-end switch is cleaner if power exists |
| Beyond 500 m | any | Fibre pair plus a PoE switch at the far end | Copper chaining is exhausted; fibre carries the link, the local switch powers the cameras |
| Existing long coax or twisted pair | 1–4 | Long-reach or VDSL extender pair | Reuses installed cable; accept reduced bandwidth at long reach |
A single camera beyond the limit
One camera at 130 m is a two-hundred-dollar problem at most, and the right answer is the smallest device that regenerates the signal — a single inline extender, or a switch port running in extend mode if yours has one. Do not over-engineer it. Keep the camera at base PoE class so the power has margin to spare.
Two to four cameras further out
At this scale the choice is between a chain and a far-end switch, and the deciding question is whether there is power at the far end. If there is — a socket in the pump house, a supply in the outbuilding — put a small PoE switch there and skip the chain entirely. It adds one clean hop, powers the cameras locally, and gives you somewhere to land a second camera later.
If there is no power and no realistic way to get it, a two-hop extender chain is the workable answer, and three or four hops is the limit before the budget starts excluding cameras.
A whole building beyond reach
When the far end needs several cameras, treat it as a network link rather than a camera drop. A fibre pair between the buildings plus a PoE switch at the far end gives you clean bandwidth, isolation from electrical differences between structures, and room to grow. It costs more up front and considerably less in frustration than a five-hundred-metre copper chain carrying four cameras.
Where a kit system removes the problem
Most of this article exists because cameras are added to buildings piece by piece. A system planned as a unit keeps the problem small: where the recorder carries the PoE ports, every camera inside the 100 m radius needs no mid-run device at all. The 8 and 16 channel PoE systems from 4COVR are built this way, which turns the question into a floor plan exercise — how many cameras can sit inside the radius — rather than an exercise in chaining repeaters.
Questions People Ask Before They Buy
What is the difference between a PoE extender and a PoE repeater?
In practice they are close to the same device and the terms are used interchangeably. The distinction that matters more is whether the unit regenerates the data signal. Anything that does can extend the link; anything that only passes power through cannot, whatever it is called on the box.
Can I just add a PoE injector to go past 100 metres?
No. A midspan injector adds power to a run; it does not regenerate the data. The IEEE requirement for a midspan inserted into a channel is that it "shall provide continuity for the signal pairs," which is a pass-through function, so the channel length is unchanged. A midspan solves a missing power source, not a missing hundred metres.
How many PoE extenders can I daisy-chain?
Four is the practical ceiling, giving roughly 500 m, and that figure is published by multiple manufacturers. The constraint is not the hop count itself but the power remaining at the end of the chain: at four hops you are down to base PoE class, which will run a fixed camera but not a PTZ or a heater-equipped model. Not every extender supports chaining at all.
Does an extender reduce the power available to the camera?
Yes, by its own consumption per hop plus the cable loss on each segment. A gigabit extender typically draws 4 to 5 W, so a 25 W feed loses about a fifth of its budget at the first hop and about four fifths by the fourth. Long-range units are far lighter, sometimes under a watt, so the penalty depends heavily on which type you chose.
Do I get gigabit speed at 500 metres through an extender chain?
Not reliably, and the honest answer is that you should plan for it not to. Standard extenders are 10/100 Mbps devices, with gigabit available on some models, and every hop both consumes power and adds a re-negotiation point. If you need guaranteed gigabit at that distance, fibre is the answer rather than copper chaining.
What happens if a mid-run extender loses power?
The link drops. An inline extender is an active regenerator that powers itself from the run it is extending, so if its feed fails it stops driving the signal and everything downstream goes offline with it. That is the argument for housing mid-run devices where they can be reached and for protecting the run with surge suppression at both ends.
Where This Leaves the Decision
The whole field reduces to one question, and it is not about signal strength. Ask whether the device regenerates the data. If it does, it can extend a run and the arithmetic is about power per hop. If it does not, it cannot extend anything, no matter how the listing is worded.
From there the choices are orderly. A single camera just past the limit wants the smallest regenerating device available. A cluster of cameras with power at the far end wants a switch there instead of a chain. A building beyond reach wants fibre. Installed cable that cannot be replaced wants a long-reach pair, accepting the bandwidth that comes with the distance.
And whatever you build, count the hops before you count the metres. Four is the practical limit, the power budget is what runs out first, and the break-even between a chain and a far-end switch arrives earlier than most people expect.