Can't I just fit a spare module myself?

You can, and for a fast turnaround at a show you should. Be aware that many screens hold per module calibration data, so a spare from a different batch often shows as a brighter or cooler patch until the wall is recalibrated. Keep spares from the same production run with the screen wherever you can.

Why does the dead cluster come back after an hour of running?

That is a thermal fault, almost always a marginal solder joint or a driver chip that fails as it warms up. It will pass a cold bench test and drop out mid show. We run suspect modules hot on a soak rig at full white so the fault shows itself before the module goes back to you.

Do you need the receiving card and hub board as well as the module?

It helps a lot. Sending one full cabinet, or a module plus a known good receiving card and cable, lets us prove where the fault sits rather than guessing. If you can only send loose modules, tell us the screen make, pitch and scan rate so we can drive them correctly.

Is a dead cluster the same as a dead pixel?

No. A dead pixel is one lamp package that has failed on its own, and we replace it with hot air. A dead cluster is a group of pixels sharing a driver, a scan line or a supply, which means the fault is in the circuit feeding them rather than in the lamps themselves.

The screen is still in warranty. Should I open the cabinet?

Check with your supplier first, because opening a sealed cabinet can end the warranty on some products. We can inspect and write up a fault report without carrying out rework if you need evidence for a claim. Once the warranty route is exhausted, component level repair is usually the cheaper path.

A square block of pixels is dead on one LED module while the rest of the screen looks normal

LED module dead cluster: what's failed and what we can fix

A block of pixels out on one module is almost never the LEDs. How to read the shape of the dark area, what we can rework, and when a module is genuinely scrap.

Usually: Driver IC, scan switch or cracked track, not the LEDs
Check first: Swap the module to a known good position
Fixable: Yes on most SMD modules, limited on potted GOB
Send in or on site: One module, send in. Whole wall, we come to you

One pixel, a cluster, a whole module. The same panel, and three different jobs.

Nine times out of ten the LEDs in a dead cluster are perfectly healthy. They're just not being fed. An LED module dead cluster (a block of pixels out while the rest of the module lights normally) points at one driver chip, one scan line, or a broken track under the mask. Lamps don't fail together in a tidy rectangle. Something upstream of them has.

So before you write the module off, look hard at the shape of the dark area. The outline tells you which circuit has gone.

Reading the shape of an LED module dead cluster

A module is a grid. Column drivers feed the lamps along one axis, scan switches connect rows along the other, and every failure mode draws its own outline on the face of the panel.

A solid rectangle with all three colours out, edges dead straight and aligned to the pixel grid, is the pattern we see most. That's one driver chip, or the supply rail feeding it. If the same block is dark in one colour only and the other two look right, the fault is inside a single bank of that driver or in the track feeding it.

Full horizontal lines, evenly spaced down the module, are a scan side problem: a row switch transistor or the shift register driving it. Look at the neighbouring modules before you touch anything. If those lines carry on across the cabinet, the module is innocent and your fault is in the hub board or the receiving card.

Two more outlines worth knowing. A block whose edge lines up exactly with the ribbon header is usually a bent pin or a dry joint on that connector, not silicon. An irregular dark patch that ignores the grid, often with tide marks or green crust on the back, is moisture and corrosion.

Dim rather than dark is a different animal. A cluster that lights but sits well below its neighbours points at current setting on the driver or a partial short across a lamp, and those are worth catching early because they cook.

Four outlines a dead area draws. The shape names the circuit: a straight edged block is a driver, even rows are the scan side, a block against the header is the connector, and a patch that ignores the grid is water.

Five checks to run on an LED module dead cluster

Work through these in order. Most people find the answer by step two.

1.Drive a plain test pattern. Full white, then solid red, green and blue from the processor, at low brightness. Content lies. A flat field shows you the true edges of the dark area and whether one colour bank is missing.
2.Swap the module to a known good position. This is the whole diagnosis in one move. If the dark block travels with the module, the module is faulty. If it stays put, blame the hub board, the receiving card or the ribbon.
3.Reseat power and data. Pull the ribbon and the power flying lead, check every pin for bend or blackening, and refit. Do it with the cabinet off (hot plugging a module is a good way to turn a cheap repair into an expensive one).
4.Inspect the back of the board. Torch at a low angle. You're looking for scorch marks around one chip, lifted or bulged capacitors, white or green corrosion at the headers, and solder that has gone dull and cracked around the connector legs.
5.Press the face gently. If the cluster flickers into life under light thumb pressure, you have a cracked joint or a hairline crack in the PCB. Note where you pressed and stop pressing.

One more thing worth doing before the module leaves site. Note the cabinet, the position and what the screen was doing when it went, because a module that died during load in behaves differently from one that quietly stopped after two seasons in a truck.

What we can repair, and what's honestly scrap

Most of what arrives here goes back working. Driver ICs, scan transistors, shift registers, decoupling caps, ribbon headers and cracked tracks are all standard rework, done with hot air and a microscope to the usual electronics rework practice that IPC publishes. Individual lamp packages come off and go back on the same way. Receiving cards and hub boards we repair as well, and we test them against the screen's own controller family where we have it.

The honest limits are these. Potted and glue on board modules resist component level work, because the resin that makes them tough also buries the parts we need to reach, so on those we're often limited to the connector side. Boards that have been soaked and then powered up wet can be cleaned and returned to service, but corrosion keeps eating tracks under the solder mask and a proportion come back. We'll tell you which of your batch we rate and which we don't.

Lamp supply is the other constraint. Fine pitch packages come in colour bins, and if the original bin isn't available the repaired area reads as a slightly different white against its neighbours. We check that before we start, not after. Calibration matters too: many screens store per module correction data, so a module that has had a large group of lamps replaced usually wants recalibrating with the rest of the wall before anyone will call it invisible. Manufacturer control platforms handle that in their own software, for example the calibration tools published by NovaStar.

One fault we see a lot on returning tour stock: the module is fine, the ribbon is fine, and the receiving card has a dead output. It's worth ruling that out with step two above, because a cabinet full of good modules can look catastrophic when one card has failed.

Getting an LED module dead cluster to the bench

A single dark block on one module is a send in job. Pull the module, pack it face protected, and get it to us through our send in repair route with the cabinet type and pitch noted. Sending the receiving card and ribbon from the same cabinet lets us test the module driven the way it actually runs, which is worth the extra weight.

When the problem is spread across a wall, a fixed install in a venue, or a rig you can't strip, we come out instead. On site service suits screens that are flown, bonded to a structure, or booked for a show at the weekend. Either way, the detail of what we do to LED product sits on our LED screen repair page, and there are examples of finished jobs in our work.

For rental houses the pattern that works best is batching. Set failed modules aside through the season, mark the cabinet and position on each, and send them in together during a quiet week. You get a stock of tested spares back rather than a shelf of unknowns. If that volume is regular, a retainer is worth a conversation, and you can start that through contact.

Keeping the next module alive

Heat and damp do most of the killing. Modules stored in a damp truck and then run at full white will fail faster than anything else you own, so let cold stock warm up before you power it and dry anything that has been out in weather.

Dress the ribbons so they aren't strained when the cabinet closes. Keep magnets clean and seated properly, because a module sitting proud runs hotter at one corner. And handle boards by the edges with a wrist strap on the bench: static damage rarely kills a driver outright, it just weakens it so the LED module dead cluster turns up three shows later.

Check the easy things first. It's very often the ribbon.

Technician concentrating on soldering iron rework of an LED module at the workshop bench

Technician soldering an LED module at the CPS workshop bench, extraction hose overhead