A screen that shows its images only under a flashlight has a dead illumination path, not a dead display. After a lamp change, verify the kit part number and seating once, then measure the inverter's 5 to 12 V DC input: present and still dark means the drive output, absent means the fuse or supply upstream. Leave the lamp plug alone; it carries 1000 to 2000 V AC rms at strike.
You fitted a new lamp kit and the screen stayed dark, but a strong flashlight at an angle still shows the display items. That is good news for the diagnosis: the panel is running its project and driving its glass, and only the illumination path is dead. After a lamp change that points at the kit's seating, the drive circuit, or the supply behind it, and one safe measurement splits those apart. The lamp plug itself is the wrong place for a meter: on a CCFL backlight it carries 1000 to 2000 V AC rms to strike the tube. Here is the order of checks, the numbers to expect, and the point where you stop fitting lamps. If the kit is not in yet, start with the lamp exchange path on small panels.
Why is the screen visible only under a flashlight?
A faint image at certain flashlight angles means the panel's graphics path works and the illumination path is dead, so you are troubleshooting the lamp circuit, not the project or the display glass.
An LCD panel does two jobs. The liquid crystal layer shapes the image, and the lamp behind it makes the image readable. The flashlight stands in for the lamp during the test: if the items appear at an angle, the panel has power, the project is running, and the glass is being driven. What is missing is light. Repair services that work on industrial panels list this exact pair of symptoms for backlight failure: a completely dark screen on a panel that still responds to touch, and an image visible only under strong external light.
Record three things before you open anything: whether the whole screen is dark, whether any edge or corner glows, and whether the image changes as the panel runs. Those observations map onto the causes in the next section.
Why did the new lamp kit not light the screen?
A new lamp stays dark for one of four reasons: the kit is wrong for the exact model and revision, the connector is not seated, the drive circuit delivers no power, or the fault is not in the lamp at all.
Match what you recorded to one of these:
- Whole screen dark, touch still responds. The lamp circuit is open somewhere along its length: lamp, connector, inverter, or the DC supply feeding the inverter.
- Image visible only under strong light. The same fault, plus confirmation the graphics path is alive.
- Brightness fades in and out, or the image takes on a dark shade before it dies. A lamp or inverter at the end of its life, with the drive still partly working.
- A brief flash at switch-on, then darkness. The lamp strikes and the drive cuts it; backlight fault guides describe this flash-then-off behavior on LED sets.
- A dark band or dark patch, most visible on dark scenes. One lamp or one strip section has failed while the rest still lights. On a single-lamp 10-inch panel the whole screen goes dark instead.
A lamp kit fixes one line of that list. If this is the second or third lamp in the same panel, read why an HMI backlight fails again after replacement before fitting another: repeat burnout usually has a drive-side cause.
What voltage is at the lamp plug?
On a CCFL circuit the plug the lamp connects to carries roughly 1000 to 2000 V AC rms while the tube strikes, generated by the inverter from a 5 to 12 V DC input, so the lamp plug is a measurement you do not make.
Both figures come from CCFL inverter application literature, and together they set your test points. The inverter takes a low DC rail and steps it up to a high AC output; one documented example in a small display ran from about 12.8 V DC. That makes the DC input the informative side. Your meter reads it accurately, and its presence or absence splits the fault cleanly. The AC side fails both ways: 1000 to 2000 V AC rms is above the 600 V or 1000 V input rating common on handheld meters, and the inverter's output sits at a switching frequency far above the mains band those AC ranges are specified for, so a meter that survives the probe still shows a number you cannot use.
Follow two rules. Measure the inverter's DC input at the pins the service sheet names for your exact model and revision, with the panel in the operating state the sheet specifies. And never apply external voltage to the lamp plug to test a lamp: it damages the panel and answers nothing.
If the backlight is LED strips rather than a tube, the drive is a constant-current source instead of an inverter. The same split applies: the driver's DC input on your meter, its documented output as the reference.
How do you check the drive circuit safely?
Isolate the panel, verify the kit's part number and seating once, then restore power and measure the inverter's DC input; that single reading, compared against the documented rail, splits the fault between the drive and everything upstream of it.
Work in this order:
- Isolate panel power through the machine's approved shutdown procedure.
- Check the kit part number against the replacement parts list for the exact panel model and revision.
- Inspect the lamp connector: fully home, latch engaged, no bent contacts, no contamination.
- Restore power and repeat the flashlight test. A screen that lights now was a seating fault, and you are done.
- Still dark: measure the inverter's DC input at the documented pins, panel in the documented state. On this class of panel expect a 5 V or 12 V rail.
- Compare the reading and take the branch it points to.
| Reading at the inverter DC input | What it means | Next move |
|---|---|---|
| Documented rail, inside a 5% band (0.6 V on a 12 V rail) | The drive is fed; the fault is at the output, the connector, or the lamp | Stop fitting lamps; go to the exchange decision |
| Well below the rail, or 0 V | The drive is not fed; the fault is upstream | Trace the backlight fuse, then the panel supply and ground |
| Rail present, one flash at switch-on, then dark | The drive shuts down after strike | Check the connector and the lamp part number, then the exchange decision |
Recommendation: measure the inverter's DC input, never the lamp plug. If the documented rail is present and a verified, seated lamp stays dark, stop replacing lamps and move to the exchange decision: the fault is in the drive output or the backlight board. If the rail is absent, trace the backlight fuse and the 24 V DC supply and its ground before you touch the lamp again.
Could the fault be somewhere other than the lamp?
Yes: a screen with no faint image at all can be a project or firmware problem, and a weak 24 V DC supply or a poor ground can leave a panel running with its backlight off.
Split it by what the flashlight shows. A faint image confirms the graphics path, and the lamp circuit stays the suspect. Nothing visible at any angle changes the order: reload the project first, then run the panel maker's recovery or reset tool, the two steps field reports put ahead of condemning a backlight on an industrial panel. Firmware is a real suspect too; some 10-inch panels have had dark-screen faults traced to firmware rather than the lamp, so apply any service bulletin through the maker's procedure before buying a second kit.
Then check the power feeding the panel. Field reports on dead backlights pair them with 24 V supply and ground problems. Measure the supply at the panel's own terminals with the panel running, not at the cabinet end, and check ground continuity back to the cabinet bar while you are there. A supply that sags under load leaves enough for the logic and not enough for the inverter.
If the panel reports a backlight error message at power up instead of failing silently, start with the checks for a backlight error at power up before the steps on this page.
How does the whole check run on an example 10-inch panel?
Take a 10-inch panel on a 24 V DC supply: the example reads 24.3 V at the cabinet, drops 0.2 V to 24.1 V at the panel, finds 0 V at the inverter input where 12 V belongs, and ends at an open backlight fuse.
The sequence with its numbers:
- The screen is dark, touch still responds, and the flashlight shows the runtime screens. Fault class: illumination.
- Power isolated. The kit part number matches the parts list for the panel's revision. Connector reseated, latch closed. Power restored: still dark. Seating ruled out.
- Supply under load: 24.3 V DC at the cabinet terminal, 24.1 V DC at the panel's power terminals. Drop across the run: 24.3 V minus 24.1 V = 0.2 V, which is 0.2 V divided by 24.3 V, about 0.8% of the supply. The supply and ground are healthy.
- Inverter DC input, at the pins the service sheet names, panel running: 0 V, against a specified 12 V. The drive is not fed.
- Power isolated again. The backlight fuse on the power board reads open on the continuity range. Replaced with the documented type and rating.
- Power restored. The inverter input now reads 11.9 V DC: 12 V minus 11.9 V = 0.1 V low, about 0.8%, inside the 5% band. The lamp strikes and the screen is readable from the operator position.
The fault was the fuse, not the lamp, and the kit you replaced was good. Run the arithmetic the other way: had the inverter input read 11.9 V on the first measurement with the lamp still dark, the fuse and the supply would be cleared in one step, the fault would sit at the drive output, and you would be at the exchange decision without opening the power board.
When should you exchange the panel instead of repairing it?
Exchange the panel when a verified kit, a seated connector, and a healthy inverter input still leave the screen dark, because the fault then sits in the drive output or the backlight board, and that repair belongs to a board-level service, not to the machine.
Three situations end here:
- Documented rail present, lamp dark after one reseating. The inverter output, the lamp connector, or the board is at fault. A board-level shop will follow a dark-display diagnostic path for legacy operator panels of the kind its own write-ups publish; otherwise exchange the panel.
- A second lamp has died in the same panel. Repeat burnout points at the drive circuit; fix the drive, or the third lamp dies too.
- Kits and boards for the model are scarce. Pick the replacement by testing startup and runtime first when you choose an HMI panel, and keep shift information in front of operators meanwhile with shop-floor signage that plays from a memory card or USB stick, offline, with no subscription.
Record what you measured before the panel leaves the machine: the supply reading at the terminals, the inverter input reading, and the kit part number. A repair service will ask for exactly those three.
Questions
Can I measure the lamp plug with my handheld multimeter?
No. A CCFL lamp plug carries 1000 to 2000 V AC rms at strike, above the 600 V or 1000 V input rating common on handheld meters, and the output frequency sits outside their AC accuracy band. A near-zero reading there means nothing anyway: the strike pulse is brief and the probe loads the circuit. Use the inverter's 5 to 12 V DC input instead.
The screen faded in and out for weeks before going dark. Is that the same fault?
The fade tells you the drive was feeding the lamp right up to the failure, so the lamp or the inverter had reached the end of its life. That history matters after the repair: a screen that stays dark right after a correct tube change points at the connector or the fuse, not at another lamp.
How do I know the replacement kit fits my panel?
Match the part number on the kit to the replacement parts list for the exact panel model and revision, and check the connector style before you open the panel. Kits across one maker's sizes look alike; a lamp for a different screen size can seat in the connector and still fail to strike, which reads exactly like the drive fault you are chasing.
The inverter input reads the right voltage but the lamp stays dark. What now?
The fault is on the output side: the inverter itself, the lamp connector, or the lamp assembly. With the kit verified and seated once, further lamp changes are wasted work; the choice is a board-level repair service or a panel exchange. Repeat lamp deaths on the same panel point back at the drive circuit as the underlying cause.
Planning information, not legal or engineering advice. Rules differ by place: check codes and permits with your local authority and your licensed trades.