Trigger strobe lighting from robot IO digital outputs

Robot cells and PLC lines often need more light than a camera's built-in illumination, and the natural place to switch it is the digital output that sits beside the camera node in the program. Done right, that output fires a short, bright, overdriven flash per image instead of leaving the lights cooking at full power. Done wrong, it switches a head on and off with millisecond slop, no overdrive and no timing margin. Here are the wiring, the timing numbers and the controller settings that turn a discrete robot or PLC output into a dependable strobe trigger — and the cases where the camera should fire the light instead.

Can you trigger strobe lighting from a robot or PLC digital output?

Yes — connect the output to the strobe input of a light controller and let the controller shape the flash: its trigger delay and pulse width resolve in microseconds, so the robot output only has to arrive somewhere near the exposure. The discrete output requests the flash; the controller times it.

In the robot program you set the output just before the camera node and clear it after, with a wait node wherever the sequence needs the light armed before capture. That is gating, and it is where discrete IO belongs — the robot decides when inspection happens. The microsecond decisions happen inside the controller, which answers the input edge with a fixed-width pulse into the LED. That split, robot logic arms and controller times, is the pattern that makes discrete-output strobing production-stable.

Why can't the digital output switch the light itself?

Two hard limits: a discrete output switches tens of milliamperes to at most a few amps with 1-10 ms of update jitter, while an overdriven light pulls several amps and needs a pulse timed to tens of microseconds.

Current comes first. Transistor outputs on robot controllers and PLC cards commonly source 0.5 A per point, 2 A on the generous units. An overdriven head draws multiples of its continuous current — 3-5 A pulses are ordinary — which the output will current-limit or refuse. Controllers are built for that league: a 4-channel unit like the LS-CTL-0322 carries an 8 A total output.

Then timing. A PLC scan runs 1-10 ms, and a robot controller refreshes its IO once per servo cycle, every 2-8 ms on common machines. Strobed exposures run 100 µs to a few milliseconds, so the jitter in the trigger path is already larger than the event being timed. Relay outputs make it worse — 5-10 ms of actuation plus contact wear — so reserve relays for mains loads and read up on relay selection and wiring for PLC-switched lighting before you put one near a vision cell.

What does a light controller strobe input add?

Four things the discrete output cannot do: microsecond trigger delay, microsecond pulse width, a driver that overdrives the LEDs past their continuous current, and duty-cycle protection that trips when you pulse too often.

Most digital controllers also offer one-shot mode: the input edge starts a pulse whose width is set on the controller, not in the robot program. This is the feature that rescues robot IO. The output can be late and slow by milliseconds; the controller re-times the flash from the edge and emits exactly the width you asked for. Delay shifts the flash later in fine steps — 10 µs resolution is common, over ranges of 0-100 ms or more — and pulse width sets flash energy and motion blur in one number.

A workhorse for this pattern is the LS-CTL-0070: a 4-channel, 200 W digital control box with a strobe feature and an RS232 link for remote set-up.

The load can be any 24 V head — bars from the bar lights family for linear fields and print, or ring lights for even light around the lens axis. Typical pairings from the lights finder: the LS-BAR-0400 high-luminance bar light at 312 x 30 mm, and the LS-RNG-0100 ring light, 120 mm in diameter with a 60° beam, 620 nm red, 24 V.

How do you wire the robot output to the strobe input?

Source 24 V from a PNP (sourcing) output into the controller's strobe input, tie the robot's 0 V to the input common, and you are done: the 5-10 mA an opto-isolated input draws is trivial for any output.

If the output is NPN (sinking), select the input's sinking mode where the controller offers it, or add a pull-up of 2.2-4.7 kΩ to 24 V and invert the logic. Keep the strobe signal as a twisted pair with its return, shielded and grounded at one end, and route it away from the LED power leads — pulsed currents of several amps induce edges of their own into parallel wiring.

Two grounding rules save an afternoon. The input common must share a reference with the driving output; the opto-isolator is what lets you keep the controller's power ground separate. And never run the strobe wire through a connector that also carries the light's 24 V supply. In one-shot mode the input needs only a clean edge — a 1 ms output pulse still produces the full controller-timed flash.

How do you line the strobe up with the camera exposure?

Make the pulse longer than the exposure plus twice the IO jitter, so the exposure sits inside it anywhere it lands: with 10 ms of jitter and a 1 ms exposure, 2 × 10 + 1 = 21 ms of pulse absorbs all the slop, and below about 1 ms of pulse, move to a camera strobe output.

The chain, from program to photons: the robot sets the output, and the level appears at the terminal after up to one IO update of 2-8 ms, plus network jitter if the command arrives over a fieldbus — triggering digital outputs over PROFINET covers that hop. The controller then adds its trigger delay, and the LED needs its optical rise time to reach full output; measuring usable strobe rise time shows how to capture that once. The exposure runs somewhere in the middle of all of it.

If the robot triggers the camera in software too, exposure start inherits the same jitter. Trigger call versus exposure time breaks that path down, and camera trigger delay timing for offset inspection sensors covers compensating a fixed offset. Then trim by image, no instruments needed:

1. Set delay to zero and the pulse to twice the jitter budget plus the exposure plus 200 µs.
2. Capture 50 images at production rate and count dark or half-lit frames.
3. Racing flash and exposure show up as intermittent dark frames; the diagnosis and fixes are in strobe light on but camera image is black.
4. Once stable, shorten the pulse in 10-20% steps until frames lose light, then step back 30% for margin.
5. If only the first frame after start-up is dark, the light is not late — the camera's exposure is still settling.

A photodiode, or a scope on the controller's monitor output, measures the true delay in one shot. The image trim above is the no-instruments version, and it is usually enough.

How long a strobe pulse does a moving part need?

Cap motion blur at half a pixel: pulse width = (0.5 × mm per pixel) / part speed, so a part moving 200 mm/s at 0.04 mm per pixel needs a 100 µs flash.

Worked example, conveyor. An 80 mm field of view across 2048 pixels gives 80 / 2048 = 0.039 mm per pixel; call it 0.04. Half a pixel of blur is 0.02 mm. At 200 mm/s the pulse is 0.02 / 200 = 0.0001 s = 100 µs. Parts arrive at 15 per second, so the duty cycle is 100 µs × 15 = 0.0015, that is 0.15%. Even overdriven 5×, the average current is 5 × 0.15% = 0.75% of the continuous rating. The duty budget is enormous; the pulse is short only because of motion. Check your own figure with the camera resolution calculator.

When the robot holds the part still during capture, motion drops out and the pulse follows the exposure: a 2 ms exposure takes about 2.2 ms of flash, the extra 10% covering LED rise and switch skew. On webs and line scan cameras the arithmetic moves to line rate and continuous power — line scan light requirements covers that regime — and sizing the strobe pulse for barcode verification does it for codes on fast lines.

How far can you overdrive the LEDs?

Overdrive is a designed mode, not abuse, once you respect the two numbers in the light's own spec: strobe current, commonly 3× to 10× the continuous rating, and a duty-cycle cap, usually 10%. Stay under both and the head pulses for millions of cycles.

Brightness rises less than current does, but it rises a lot; dedicated high-power strobe lines claim up to about 8× the brightness of their standard counterparts in overdrive. The thermal check is average current: average current equals strobe current times duty, so 5 A pulses at the 10% cap average 0.5 A — the heat the head must shed. Long pulses also droop as driver and LED heat within the pulse itself; diagnosing LED strobe current droop shows the trace to expect. The full method is in strobe machine vision light: overdrive, duty cycle and timing and matching LED strobe duty cycle to lights and controllers.

Three controllers that span the usual strobing jobs:

Values as printed in each product's data sheet; open a part for its drawing and full specifications.

When one or two heads must be driven hard, the LS-CTL-0264 puts 2 channels and 400 W behind the pulse.

When is a camera strobe output the better trigger?

Whenever pulses must be shorter than about 1 ms — in practice, any part faster than roughly 100 mm/s at normal pixel sizes — because the camera's own strobe line follows its exposure to within microseconds and needs no trimming. Robot IO cannot promise that.

The strongest set-up combines both signals: the robot output enables the controller channel, so heads stay dark between inspections, and the camera's strobe output fires each flash, locked to the exposure. Configure the camera line as exposure-active or flash in its GenICam settings; checking GenICam line settings walks the mapping. At high part rates, respect the camera's trigger readiness too — sequenced exposure trigger timing explains the overtrigger errors you would otherwise chase.

Decision, case by case:

• The robot program owns the sequence and the part dwells in front of the lens, with pulses of 1 ms and up: robot IO into a controller one-shot is the clean answer.
• The part moves faster than about 100 mm/s, or pulses must be under 1 ms: camera strobe output into the controller, robot IO as the arm signal.
• The light simply has to be on whenever the station runs: controller continuous mode, robot output as interlock only.

What should you check before ordering?

Seven checks, in order: input logic that matches your output, delay range and resolution, pulse-width range or one-shot mode, channel current, duty protection, a spare channel, and the light's overdrive spec in writing. Each one is a floor-debug you would otherwise run in production.

• Strobe input logic matches your output. 24 V PNP is the common pairing; a sinking input forces the pull-up trick and inverted logic.
• Delay range and resolution. 0-100 ms of range in 10 µs steps covers every robot and fieldbus chain you will meet.
• Pulse-width range, or true one-shot mode. You want widths from tens of microseconds for moving parts to tens of milliseconds for dwell captures, timed by the controller.
• Channel current above the light's strobe current. Size to the overdrive figure, not the continuous draw.
• Per-channel duty-cycle protection. It saves the head the day a program bug hammers the trigger line.
• One more channel than you need. Cells grow a second head within a year.
• The light's overdrive spec in writing. Strobe current and duty cap are the two numbers the whole design rests on.

Filter the field in the light controller finder or step straight through the configurator's controller stage. If the timing chain needs a second opinion, send your camera's exposure time and your part rate through the contact page and we will size the delay and pulse width with you.