Strobe light on but camera image is black: fix the timing

You trigger the camera, the bar light flashes, and the frame comes back black. The flash is real: the eye integrates any visible flash, while the sensor only collects light during its exposure window, so a pulse that fires before or after the exposure adds nothing to the image. This guide works the fault in order for anyone running an external strobed light with a smart camera or a PC-based camera: proving the pulse is missing the exposure, sizing the pulse, why two trigger signals set at the same instant still race each other, and how to drive camera and light from one trigger with a light controller.

Why does the light visibly flash while the image stays black?

Because your eye integrates light over roughly a tenth of a second, but the sensor integrates only during the exposure window: a 1 ms flash outside a 2 ms exposure looks bright to you and contributes zero photons to the frame. The eye sums every pulse longer than a few microseconds into one bright impression; the sensor sums nothing outside its window. A pulse that ends half a millisecond before the exposure opens produces a frame identical to lights-off, and a pulse covering only part of the exposure produces a frame at exactly that fraction of full brightness.

The eye cannot time a millisecond event, so watching the light tells you it fires, never when it fires. Every test below replaces the eye with the camera, the only observer that matters.

How do I prove the pulse is missing the exposure?

Set the light to continuous-on and capture once: if the image is normal, the optics, gain and camera are fine and timing is the fault; a ten-times longer pulse that suddenly produces an image confirms it. Work through three tests in this order, because each one separates a different cause:

1. Continuous test. Run the light continuous-on and trigger the camera. A normal image means everything except timing works. A black image with the light solidly on ends the timing hunt — work the exposure path instead, under "When is strobe timing not the fault?" below.
2. Stretch test. Return to pulsed mode and multiply the pulse width by ten. An image that appears means the pulse was too short or too early; a frame that stays black at ten times the width means the pulse lands far from the exposure — a delay problem, not a width problem.
3. Fraction test. On a global-shutter camera, mean gray level scales with overlap. Compare the pulsed image against the continuous-on reference: 45 gray levels against 180 gives 45 / 180 = 25% overlap, so a 2 ms exposure is lit for about 0.5 ms.

How wide should the strobe pulse be?

Set the pulse at least as long as the exposure plus the worst-case skew between the two trigger paths — pulse ≥ exposure + skew + LED rise time — and round up 10-20% while commissioning, tightening once the image is stable. For bring-up, type in twice the exposure plus the measured camera delay, then narrow it.

The skew term is the gap between the moment the light turns on and the moment the exposure opens. The rise term is the LED's optical turn-on, tens of microseconds on a driven bar light. If the part moves, pulse width sets motion blur exactly as exposure does, so the moving-part limits in exposure time vs shutter speed apply to the pulse as well.

Worked example: 2 ms exposure at 20 triggers per second

A station fires 20 parts per second with a 2.0 ms camera exposure. The delay sweep described two sections below measured 3.0 ms from trigger edge to exposure start; the controller's stated input-to-output response is 0.1 ms; the bar's optical rise time is about 50 µs.

• Light turns on at 0.1 ms; the exposure runs from 3.0 ms to 5.0 ms.
• Brute-force width: 5.0 − 0.1 = 4.9 ms, plus rise margin, so set 5.0 ms.
• Duty at full rate: 5.0 ms × 20 /s = 100 ms of light per second = 10% duty.
• Tighter: delay the pulse start 2.9 ms so it turns on at 3.0 ms, width 2.2 ms, covering to 5.2 ms. Duty = 2.2 × 20 = 44 ms/s ≈ 4.4%.

The delayed pulse lights the part for 2.2 ms where the image needs 2.0 ms, so brightness reaches full level while the LED runs at less than half the duty of the brute-force setting. The same overlap arithmetic governs freezing motion with a strobe, and selecting a strobe light for servo timing verification applies it to drive and cam timing.

Why did two separate trigger signals fail?

Two outputs set TRUE by the same condition still switch one controller cycle apart, and a single PLC scan of a few milliseconds is enough to move a 1-2 ms pulse completely outside the exposure. Each bool is written at its own point in the program, outputs refresh on the I/O cycle after that, and the two paths accumulate skew independently — the budget table in the next section puts two bools at up to two scan cycles.

A shared high-accuracy clock helps only when both devices schedule their actions on that clock. Two software outputs do not: they fire on scan time. The fix is structural, not a bigger number — one physical trigger event fans out to both devices, or the camera drives the light directly.

What delay sits between the trigger and the exposure?

Budget it in three numbers: the light path (controller input-to-output around 0.1 ms plus tens of microseconds of LED rise) and the camera path (tens to hundreds of microseconds on a frame-triggered machine-vision camera, milliseconds on a smart camera pipeline). The table is the working budget; measure your own camera number rather than assuming it.

Measure the camera's number with a scope on its exposure-active output, or without instruments by sweeping the controller's pulse-start delay in 0.5 ms steps from zero and watching mean gray level: the step where brightness stops rising and levels off is the camera delay plus the controller response. The gap between the trigger call and the exposure itself is what trigger call versus exposure time unpacks from the camera side.

How do I drive camera and light from one trigger?

Split one part-present sensor or PLC output with a Y-adapter into the camera trigger input and the controller trigger input, or let the camera's exposure-sync output drive the light so the pulse sits inside the exposure by definition. The wiring is modest and the timing becomes a property of hardware rather than program order:

1. Wire one trigger source to both the camera's trigger input and a light controller trigger input; confirm both inputs accept the same voltage and edge.
2. Set the camera to hardware trigger with the 2.0 ms exposure from the example.
3. Set the controller to 5.0 ms pulse width with 2.9 ms start delay — or wire the camera's flash output to the controller input and let the exposure itself gate the light.
4. Capture ten frames and check that mean gray matches the continuous-on reference within a few percent.
5. Keep polarity and edge selection identical on both devices, and label the cable.

For camera-side wiring, do I need a camera GPIO cable for image acquisition? covers the cable, and hardware trigger source not detected: check GenICam line settings covers the line source and edge settings. When several lights must fire in sequence, how to synchronize camera emitter patterns extends the same one-event rule to patterned illumination.

Which controller settings remove the guesswork?

Program the pulse width and pulse start delay in the controller, not in the PLC: a controller set to 5.0 ms width and 2.9 ms delay emits the identical pulse on every trigger, scan after scan. Four settings do the work:

• Pulse width (ms): covers exposure plus skew, per the formula above.
• Pulse delay (ms): slides the pulse onto the exposure window.
• Trigger edge and level: matches the source signal.
• Current limit and overdrive: sets brightness, bounded by duty cycle.

Controllers with a hardware trigger input and a strobe feature hold all of this in one box:

The LS-CTL-0070 digital control box provides 4 channels, 200 W, an RS-232 interface and a strobe feature. Smaller units cover the same ground:

The LS-CTL-0487 drives 24 V lights on 2 channels at up to 240 W, with constant-current and constant-voltage outputs and an RS-232 interface. The LS-CTL-0180 is a 4-channel constant-current unit rated 120 W, sized 89 x 147 mm, for 24 V lights. Match the output to the light: standard bar lights run on constant-voltage 24 V channels, current-driven heads on constant-current channels. A bar of the type these channels drive:

When is strobe timing not the fault?

If the frame stays black with the light running continuous-on, timing is innocent and the fault sits in exposure, gain, trigger routing or the lens cap — the path camera black image: check exposure, lighting and lens cap works through in order. Three other symptoms also get blamed on strobe timing but have different causes:

• Only the first frame after start-up is dark. Exposure control and the light have not settled yet; see first camera frame dark: exposure settling or light timing?.
• Brightness drifts between identical parts. The pulse current droops as the driver settles; LED strobe current droop: diagnose pulse startup separates that from timing.
• A bright horizontal band crosses the frame. A rolling-shutter sensor reads rows while the pulse is on, so a short pulse lights only part of the frame. Widen the pulse past the full readout or use a global-shutter sensor.

For a new station, these routes in smart cameras remove the sync question in hardware:

The MV-SC3050XC-16M-ATN-H is a 5 MP vision sensor with a global shutter, so one pulse lights the whole frame. The LS-SMC-0063 is a 1.6 MP smart camera at 60 fps with a 16 mm lens and its own polarized white light, so illumination and exposure share one internal clock and there is nothing left to sync. Codes on moving parts add one constraint — the pulse freezes motion only inside the overlap — and high speed barcode verification: size the strobe pulse sizes it for reliable reads.

What should I confirm before ordering parts?

Four lines of the controller data sheet decide the purchase: a trigger input with a pulse-width setting in milliseconds, a stated input-to-output response time, a duty-cycle or overdrive limit above your trigger rate, and the output type your light needs. Check them off against the station numbers:

• Pulse width range covers exposure plus measured skew with headroom: a station at 2 ms exposure behind 3 ms camera delay needs at least a 5 ms setting.
• Response time is stated in the data sheet, not guessed — it is a term in the width formula.
• Duty limit at your trigger rate leaves room for the current you plan to run.
• Output type matches the light: constant-voltage 24 V for standard bars, constant current for current-driven heads.
• Channel count leaves a spare for the second light most stations eventually grow.

Compare candidates in the light controller finder, or start from the controller step of building a system. If the timing sketch is unusual, send it through the contact page and we will size the pulse with you.