The first question about any HDMI link is how far it can go. It is the wrong first question, because the answer depends on resolution, refresh rate, colour format, cable construction and how much the far end is allowed to cost. A link that carries 1080p60 to forty metres is easy; the same distance at 4K60 with 4:4:4 is a different engineering problem with a different price.

Getting the order right — bandwidth first, then distance, then power and control — avoids the common outcome where a link is specified for a distance it can technically reach but not at the format the customer actually wants.

What HDMI is actually carrying

Two eras coexist under one connector. Legacy HDMI uses TMDS, three data pairs plus a clock, with 1.4 topping out around 10.2 Gbps and 2.0 at 18 Gbps. HDMI 2.1 introduced FRL, four lanes with no separate clock, reaching 48 Gbps.

The bandwidth number maps to a pixel format, and this is where specifications mislead. 4K60 at 4:4:4 8-bit needs about 17.8 Gbps and fits inside HDMI 2.0 with almost no margin. 4K60 at 4:2:0 halves the chroma data and needs roughly 8.9 Gbps, which is why so many devices advertise 4K60 and quietly subsample. Both are "4K60" on the box. Only one of them looks right on a text-heavy display.

For an extender, the practical consequence is that two products with the same distance rating can have very different bandwidth ceilings, and the rating is usually quoted at the easiest format the marketing department could find.

Passive copper: shorter than the marketing suggests

A passive HDMI cable is copper, and copper attenuates high frequencies. At 18 Gbps, a well-made passive cable reaches roughly five metres before the eye closes at the far receiver. At 48 Gbps, that number drops to two or three metres. Longer passive cables exist and work with some source and sink combinations, which is exactly the problem: they work until the customer pairs them with a different television.

Passive is the right answer for rack-internal connections and for short runs where both ends are fixed. It is the wrong answer for anything installed in a wall, because the failure will be intermittent and the cable will be hard to replace.

Active copper: a redriver inside the connector

An active copper cable puts a redriver or a retimer in the connector shell at each end. It equalises the channel, which extends the reach to fifteen or twenty metres at 18 Gbps, and rather less at 48 Gbps.

Active copper is directional and it draws power. That power comes from the source's +5 V line, which is specified to supply a limited current, and from the sink side. A cable that draws more than the source can provide will work on one device and fail on another, and the failure looks like a compatibility problem rather than a power problem.

The advantage over fibre is that it carries the +5 V and the DDC lines natively, so EDID and HDCP pass through without any conversion. The disadvantage is that the copper still carries the high-speed pairs, so the distance ceiling is a physics limit rather than an engineering choice.

Fibre: distance solved, everything else added

An active optical cable converts the electrical signal to light at the source end and back at the sink end. Distance becomes almost irrelevant — a hundred metres is routine, several hundred is possible with the right modules.

What you add is a set of problems that copper does not have. Each end needs power, and the sink end has no natural supply, so the design needs a local adapter, power over the cable, or a battery. The conversion adds latency, small but measurable. And the DDC, CEC and hot-plug lines still have to be carried electrically or emulated, because they are not high-speed signals and cannot simply be modulated onto the optical link.

The last point is where cheap optical extenders fail. If the EDID is not read correctly, the source falls back to a safe resolution; if the HDCP handshake is emulated badly, the display shows a black screen with no error message.

HDBaseT and IP: when the cable is already in the wall

HDBaseT carries HDMI, Ethernet, control and power over a single Cat6 cable to about a hundred metres. It is the natural choice when the installation already has structured cabling, or when the run has to go through conduit that will not accept an HDMI connector.

It is not a passive adapter. Both ends are active devices with their own PHY, their own latency, and their own compatibility matrix. Power over HDBaseT is a real advantage for displays mounted where no mains outlet exists.

IP-based distribution goes further: encode at the source, decode at each display, route through a normal network switch. The advantage is fan-out and distance without dedicated cabling; the cost is compression, latency, and a network that now has to be engineered for multicast.

The details that decide whether it ships

HDCP. Every extender in the chain is a repeater, and HDCP allows a limited number. A source, a splitter, an extender pair and a receiver is already four. Beyond that, the handshake fails and the symptom is a black screen rather than an error.

EDID. The extender has to present the display's capabilities to the source, or store a fixed EDID. Both approaches work; a design that does neither will negotiate at whatever the last device left behind.

Power at the far end. Decide early whether the receiver is locally powered, powered over the cable, or powered by a PoC injector. Retrofitting this after the enclosure is tooled is expensive.

Thermal. An active module in a sealed connector shell dissipates a watt or two with no airflow. Industrial installations in an enclosure at 60 °C ambient are where these fail.

Cable construction. For in-wall installation, plenum rating, bend radius and connector pull strength matter more than the bandwidth spec, because the cable will be installed once and never touched again.

A selection sequence

Start from the pixel format and refresh the customer actually needs, and convert it to a bandwidth number. Then the distance. Then whether the far end has power. Then whether the cable must go through conduit or around tight bends. Only then compare passive, active copper, optical and HDBaseT on price.

Most over-specification comes from quoting a distance at the easiest format, and most under-specification comes from ignoring HDCP repeater depth or far-end power. Both are decided at the quotation stage, not at installation.

If you are putting together an HDMI link for digital signage, inspection equipment or a conference room, we can work through the bandwidth, distance and power constraints against the formats you need to support. We supply interface and cable-side components for these designs, so the conversation tends to be about the link budget rather than the box specification.