Wrist camera and gripper wiring: fix red LED power faults

A wrist camera that burns its status LED red and finds no gripper is rarely a dead camera. It is a power chain that holds voltage at idle and collapses the moment real current flows, most often on a sagging feed or a work-hardened crimp where the cable flexes. The USB hub that still enumerates a dongle and a memory stick proves the controller's port works, and nothing about the wrist. You will measure supply-to-return at the pigtail under load, budget the chain's current with real arithmetic, and fix the wiring or the supply so the scan finds the accessory on the first pass.

What the constant red LEDs are telling you

Read the pattern before you touch a wire. Many devices flash their status LED red for a second at power-up while booting; that is normal. A LED that stays red is a device that has power yet reports a fault or is browning out. When the camera LED and both gripper LEDs go red together and stay red, the whole chain shares one problem, and that problem sits upstream of any single device.

On most wrist rigs the gripper draws its power through the accessory coupling on the camera's face, so a single feed serves camera and accessory at once. That saves cable and concentrates the failure: one sagging or broken conductor shows up as red everywhere at the same time. Dark LEDs would send you to a fuse or a dead supply instead; steady red means power arrives but the feed cannot carry the load.

Likely causes in order of probability

Ranks 1 and 2 cover most wrist faults: a feed that measures fine at idle is the one that fails at grip, and the harness at a rotating joint works harder than any other cable in the cell. Each test needs a multimeter and a pose command, not a spare camera.

Measure at the pigtail under load

Meter at the device end, under peak load, or the fault hides. Park the arm at the inspection pose first, then work down the steps.

1. Command the accessory closed and put the camera into its grab or trigger state. That is peak load; an idle reading hides the fault.
2. Meter between the 24 V pin and the 0 V pin at the camera end of the pigtail, not at the supply terminals.
3. Judge the reading against 21.6 V, which is the rail minus the 10% tolerance most industrial devices are built to. At 21.6 V or better the feed is healthy, so move to the link checks. Between 20 V and 21.6 V the chain sags under load. Below 20 V, or collapsing the instant the gripper closes, the run drops too much or the supply has hit its current limit.
4. Flex the cable 150-200 mm from the wrist entry while watching the meter. A jumping reading marks a work-hardened conductor or a failing crimp.
5. Repeat the measurement from the supply pin to cabinet ground. A full 24.0 V reading there with a low pin-to-pin value puts the loss in the run; two low readings point at the supply.

Worked example: the voltage budget on a 7 m wrist run

Put numbers on the run before you cut cable. Assume a cabinet supply holding 24.0 V DC, a 7 m run to the wrist, a power pair of 0.34 mm² copper (about 22 AWG), a camera drawing 0.4 A and a compact two-finger gripper peaking at 1.0 A, which are typical figures for this class of hardware; your own nameplates set the final sum. The total load is I = 0.4 + 1.0 = 1.4 A and the loop length is L = 2 × 7 = 14 m.

with ρ = 0.0175 Ω·mm²/m for copper at room temperature. Cable resistance: R = 0.0175 × 14 / 0.34 = 0.72 Ω, so the cable drops 1.4 × 0.72 ≈ 1.0 V. Clean crimps and the accessory pass-through add roughly 0.5 Ω, another 1.4 × 0.5 = 0.7 V. Arrival voltage: 24.0 − 1.0 − 0.7 = 22.3 V, above the floor.

Now let one crimp age and add 1.5 Ω of contact resistance. Contact drop becomes 1.4 × 2.0 = 2.8 V and arrival falls to 24.0 − 1.0 − 2.8 = 20.2 V. The chain browns out exactly when the gripper closes: LEDs go red, the scan finds nothing, and the camera looks guilty. That is the mechanism behind most of these faults.

Upsizing the pair to 0.5 mm² (about 20 AWG) cuts the cable drop to 1.4 × (0.0175 × 14 / 0.5) = 1.4 × 0.49 ≈ 0.7 V, but the same tired crimp still leaves 24.0 − 0.7 − 2.8 = 20.5 V at the pins. Bigger copper does not repair a bad crimp, so fix both.

Size the supply on peaks: 1.4 A × 1.5 = 2.1 A, so fit a 2.5 A or larger unit and protect the wrist feed with a 2 A slow-blow fuse, above the 1.4 A peak and below what the pair carries happily. A supply already at 1.9 A of a 2.0 A rating droops at grip and paints this exact picture.

Decision for this chain: 0.5 mm² pair, crimps re-terminated, 2.5 A supply minimum, 2 A slow-blow fuse. That gives 22.6 V at the pins at grip, over the floor with margin.

Why the working USB dongle misleads you

The hub, the license dongle and the memory stick all draw their power from the controller's USB port. Enumerating them proves the controller's USB stack and the hub chain work; it says nothing about the wrist feed, which powers the camera and the accessory instead.

The camera's data side references its local ground. USB and Ethernet signalling assume device ground and controller ground sit within a few hundred millivolts of each other. A return conductor carrying 1.4 A through a couple of ohms of contact resistance lifts the device ground by volts, pushes the signalling out of its common-mode range, and the camera goes silent on the bus while a hub-powered dongle on the same controller enumerates happily. Chase the return before you condemn a working camera.

If pin-to-pin voltage holds at 22 V or better through repeated grip cycles yet the camera still will not appear, the fault has moved to the link. The USB link checks after a cable swap sort host-side causes, and isolating USB cable and hardware faults covers the dropouts that mimic them. On a networked wrist camera, the GigE camera on a robot wrist guide handles addressing and image retrieval.

Fixes that survive the wrist

Fix the power path first. Re-terminate every crimp on the run, or better, replace the harness with cable rated for continuous flex. The LS-CBL-0300 is a 7 m, 8-core M8 male-to-open power and I/O cable built for drag-chain use; route the power pair through its spare cores and land the open ends on a fused terminal block in the cabinet.

Keep the image link flex-rated as well. A 3 m GigE cable with screw locks, high-flex and drag-chain rated, or a 3 m USB 3.0 cable of the same construction, covers most wrist runs with a service loop to spare. Static-rated cable belongs at fixed cameras, not across a moving joint: its jacket and cores work-harden and crack within months of chain duty.

If the camera's pass-through still sags at grip after the rewiring, split the feeds. Give the accessory its own fused pair from the cabinet distribution and keep only the control lines through the camera; that separates the two brownout mechanisms so the next fault announces which half it lives in. Mind the shared references when you do, because the camera GPIO wiring guide shows how to land trigger and control grounds without creating a loop.

If you are rebuilding the cabinet side at the same time, a vision controller keeps the camera links, the accessory wiring and the supply in one enclosure. The MV-VC2140-128G40-1T pairs an N97 processor with 8 GB memory, a 128 GB SSD and a 1 TB HDD, and carries 2 USB 3 ports and 4 built-in light channels, which is enough port count to keep the wrist camera on its own link instead of sharing a hub.

When the power checks out

If pin-to-pin voltage holds at 22 V or better through repeated grip cycles, the flex test stays flat and the LEDs still burn red, stop chasing the feed. Re-seat the accessory coupling on the camera's face first: its spring-pin contacts carry power and the gripper's control lines, and a half-seated collar mimics every symptom on this page. Then check that the controller-side integration package matches the device version, and that the device sits on the link your program expects.

If the harness ever shorted the rail onto a data line, port damage joins the suspect list, and the no-image-after-short isolation covers the checks that follow. A powered, link-clean camera that still delivers no pixels belongs to a different tree again: the blank-image pipeline probe walks each stage from connector to pixels.

The power decision in one line: healthy feed, clean crimps, supply at 1.5 times the peak, 21.6 V minimum at the pins at grip. Everything past that point lives in the link or the software.

Prevention checklist

Commission the wrist once, correctly, and the red-LED fault stops coming back.

Wiring

• Flex-rated, drag-chain cable for every conductor that crosses a joint; static-rated cable stays at fixed cameras.
• Power pair sized for the loop: 0.5 mm² copper keeps a 1.4 A, 14 m loop under 0.7 V of drop.
• Crimped contacts at both ends; solder concentrates fatigue where the strands meet the joint.
• Strain relief where the harness enters the wrist, with a service loop.

Supply

• One fused feed per accessory chain: a 2 A slow-blow above the 1.4 A example peak, below the conductor rating.
• Supply rated at least 1.5 times the sum of the peaks.
• A single-point ground reference; returns daisy-chained through housings create the offset faults in the first place.

Verification

• Record arrival voltage at the worst pose during commissioning and repeat it at every service visit.
• Flex the harness under load at commissioning and at each annual check.
• Inspect crimps and continuity yearly; flex-rated means longer life, not immortal.

Spec the whole chain in one pass — smart cameras, vision controllers and cables — or start at the configurator's cable step below. If a wiring sketch needs a second opinion, the contact page is the way in.