There is a specific symptom that turns up on mobile plant: a camera is fitted to a forklift, reach truck or yard vehicle, and the monitor shows a band or a slowly rolling line across the picture. It may change when the machine lifts, travels or switches a heavy load.
That does not prove the camera is faulty, and it does not prove a ground loop. It is evidence that the picture is being disturbed. An unwanted reference path, a noisy power converter, electromagnetic coupling or a cable beginning to fail can produce similar results. The useful skill is separating those causes before replacing parts.
Two further faults account for most of the rest: one mechanical, arriving months later and imitating the first; one that produces a perfectly good picture of entirely the wrong thing. A mobile installation that survives has been designed against all three separately.
What the line actually is
Analogue video is a low-voltage signal carried relative to a reference conductor. If the camera, monitor, converter, shield or mounting hardware creates more than one connection between that reference and the vehicle, unwanted current can flow through the video path. A potential difference or coupled electrical noise can then appear in the picture as a bar, hum pattern or moving line.
That is the ground-loop mechanism. It does not mean every electric forklift uses its steel chassis as the traction-current return. Vehicle architectures vary, and an isolated traction system can still acquire an unintended chassis bond through accessory wiring or mounting hardware. Never infer the vehicle’s earthing arrangement from the symptom; confirm it against the machine’s service information and measurements.
Why forklifts are worse than cars
A forklift places a camera system close to contactors, motor controllers, DC converters and conductors carrying rapidly changing current. It also places the cable on a moving structure. Electrical noise and mechanical fatigue can therefore change at exactly the same moment the mast moves.
If the line intensifies when lifting, record that correlation. It is a useful diagnostic clue, not a diagnosis: the changing load may expose a reference-path problem, converter noise, electromagnetic coupling or a damaged cable moving at the same time.
There is a second complication. Mobile plant does not run at 12 V. Traction packs of 48 V nominal are common, sitting above 50 V on a full charge, and other configurations run higher again. Any camera system fitted to that vehicle needs its supply converting down — and how that conversion is done turns out to matter more than the converter’s specification.
Test the reference path before adding parts
The instinct when a picture is degraded is to fit a better camera, filter or signal booster. None of those choices proves the cause.
Start by documenting when the interference appears. Inspect the power source, converter, shield terminations, brackets and cable route. With the machine made safe and using the manufacturer’s procedure, establish whether the camera-system reference is bonded to the chassis at more than one point. A temporary known-good isolated supply or appropriate video isolator can be a diagnostic test when used by a competent technician.
If that test removes the line, design the permanent system around a controlled single reference and a genuinely galvanically isolated converter where the vehicle architecture permits it. Do not remove protective bonding, alter the traction system or assume that “floating everything” is safe. The vehicle manufacturer’s electrical requirements take priority over a camera fix.
What this looks like in practice
Where galvanic isolation is the chosen remedy, it has to be genuine. A converter with a common negative between input and output does not break that conductive path. That does not make a non-isolated converter defective; it means it performs a different job.
Mounting matters too. A camera bracket bolted to the chassis can re-establish the connection you just removed, through the camera body, if the housing is not isolated from the mount. It is entirely possible to isolate the power supply correctly and then reintroduce the fault through a bracket.
If the line changes with lift or travel, test both the electrical reference and the moving cable. If it is constant, test the supply, converter, terminations and video path. In either case, substitution with known-good components is useful only when one variable is changed at a time.
The second thing nobody quotes for
Ground isolation solves the electrical problem. It does nothing about the mechanical one, and on mobile plant the mechanical problem is what eventually takes a working system down.
A camera earns its place on a forklift by being where the driver is not. That usually means on the fork carriage itself, looking at the pallet — because at eight metres up a racking aisle the driver genuinely cannot see what the forks are doing. Which means the camera is bolted to the part of the machine that moves, and it travels the full height of the mast, repeatedly, every shift. Whatever cable connects it has to survive every one of those cycles.
Standard video cable does not. Coaxial cable is designed for fixed installation: a heavy centre conductor, frequently solid, inside a braid or foil screen. Flex it repeatedly and two things happen. The centre conductor work-hardens at the bend point and eventually fractures — usually inside the jacket, where there is nothing to see. And the screen fatigues and opens up, which does not kill the picture outright. It degrades it, in a way that looks exactly like electrical interference. You can lose a great deal of time chasing a noise fault that is actually a broken cable.
There is a plainer problem too. A pre-made screened video lead in the length a mast run needs — around thirty metres, once it has gone up the mast, over the head and back down to the cab — does not exist as an off-the-shelf flexible product. You can buy it rigid, or you can buy it short.
So stop looking for video cable and start looking for flexible cable.
Conductor class is only part of the cable specification
IEC 60228 classifies conductors by construction. Class 1 is solid, Class 2 is stranded, Class 5 is flexible and Class 6 uses finer stranding. Finer strands can reduce strain in a bending conductor, but the class alone does not certify a cable for continuous flexing or state how many mast cycles it will survive.
The reason strand count governs flex life is geometric. When a conductor bends, the outside of the bend stretches and the inside compresses. The thicker an individual strand, the further its surface sits from the centre of the bend and the more strain it absorbs for the same radius. Divide the same copper cross-section into many fine strands and each one takes a fraction of that strain — and they can move relative to one another rather than behaving as one solid conductor.
A moving-mast cable also needs a manufacturer’s dynamic-flex rating, permitted bend radius, temperature and oil resistance, suitable screening, strain relief and a route designed for the movement. A fine-stranded cable used outside those limits can still fail quickly.
The practitioner recollection behind this article is that a seven-core screened Class 6 cable was selected for this work. The surviving quotations do not record that purchase because the client sourced it directly, so it should be read as a field recollection rather than a verified product specification.
Seven cores in one jacket does more than carry the picture. Power goes out to the camera through the same jacket, so there is one cable on the mast rather than two chafing against each other. And it leaves spare cores.
The spares are the part worth copying. Cable in a flexing application does eventually fail — that is the nature of the application, not a defect in the cable. When a core goes in a single-purpose lead, the machine comes out of service and the entire run is replaced. When a core goes in a seven-core, you move the termination to a spare and the machine goes back to work. Across a fleet, that difference is the whole maintenance argument.
Routing then matters as much as specification. The cable needs a defined path with a controlled bend radius — a chain, a festoon, or a properly managed service loop — not a length of cable left to find its own way up a mast. Cable allowed to bend wherever it likes will reliably choose the tightest radius available, and fail there.
The third mistake: bringing fixed-site lens thinking to a moving machine
Ground isolation fixes the electrical fault. Flexible cable fixes the mechanical one. There is a third error that is easier to make than either, because it comes from doing fixed-site CCTV correctly for years.
On a building, the usual problem is that the target is far away. A person at a perimeter fence forty metres out occupies very few pixels, so you reach for a longer lens — 6 mm, 12 mm — to concentrate resolution onto the part of the scene that matters. That instinct is right, and on a fixed installation it is most of the job.
On mobile plant it is exactly wrong.
The camera on a fork carriage is not looking at something forty metres away. It is looking at the fork tips, the pallet, and the racking immediately around them — frequently under two metres from the lens. A long lens at that working distance shows a tight crop of one fork tine and nothing either side of it. The driver cannot see where the load actually is, which is the entire reason the camera was fitted.
What is needed is the opposite: a wide lens, in the 2.4 mm to 2.8 mm range, which takes in both tines, the pallet and enough of the surroundings to judge alignment. Six-millimetre units fitted to this kind of machine do not fail — they work perfectly, and they are useless, which is a harder fault to argue about than one that produces no picture at all.
The principle underneath is the same one that governs fixed CCTV: pixel density on target, not resolution of the scene. On a building that means narrowing the field of view to put pixels on a distant subject. At two metres, the subject already fills the frame, so the constraint inverts and the job becomes fitting enough of the working area into the shot to be useful.
Anyone specifying mobile-plant cameras from fixed-site habit will reach for the wrong end of the lens range. It is worth checking before the order goes in.
Why this belongs in a fixed installation too
The reason this is worth writing down is not only that mobile plant is difficult. It is that the lesson does not stay on the vehicle.
The diagnostic discipline transfers to fixed sites: identify the signal type, map every reference and shield bond, and measure before inserting an isolator. The component does not transfer automatically. Analogue video, power, Ethernet and screened structured cabling have different isolation and bonding rules. A device called a “ground isolator” in one part of a specification is not evidence that every patch panel or IP camera link needs the same treatment.
Durban has a lot of this work. It is the country’s busiest port city, and the warehousing, cold storage and materials-handling operations around it run significant fleets of electric plant. Any of those sites can present the same problem twice over: on the vehicles, and again across a large distributed building where camera runs cross between structures that do not share a single clean earth.
The symptom looks like a camera fault. The cause may be electrical, mechanical or both—and the measurements decide which.
This article draws on repeated mobile-plant camera work carried out in KwaZulu-Natal between 2013 and 2020. The client is not identified. The electrical explanation is offered as engineering reasoning from field experience, not as a laboratory measurement, and any specific installation should be assessed on site.