Oscilloscope check: lighting pulse timing and camera exposure overlap

A strobe that fires at the wrong moment does not look like an electrical fault. The image comes back dark, half dark, or bright on one frame and dim on the next, and the instinct is to re-tune the vision algorithm. If you run strobed lighting — short pulses for motion control, sequenced channels for multi-view inspection — the timing between the controller output and the camera shutter is a physical measurement, and a two-channel oscilloscope settles it in minutes. This guide shows you where to probe safely, how to trigger on the exposure, and how to turn the two edges into delay and width settings you can write into the controller and trust.

What does a strobe timing fault look like in the image?

A strobe timing fault shows up in the image as a black frame, a dark strip along one edge, or brightness that jumps frame to frame, and a two-channel oscilloscope separates the three in one capture by putting the exposure and the light pulse on one time axis.

The three symptoms point at three different faults. A black frame while the light visibly flashes means the pulse landed outside the exposure window, or camera and controller never agreed on the trigger at all; the strobe light on but camera image is black guide walks that diagnosis. A dark strip along one edge means the pulse clipped the shutter at one end, which is the black bars in camera images failure mode. Brightness that wanders frame to frame usually means the strobe covers only part of the exposure and room light fills the remainder — the ambient light variation in pick and place vision guide treats exactly that as a repeatability problem. Guessing between the three costs an afternoon; one scope capture ends the argument.

What do you need on the bench to measure it?

You need a two-channel oscilloscope with a 20 MHz bandwidth limit, two 10x passive probes, access to the camera's exposure-active output, and the production controller, cable and light on the bench — no fast scope is required, because the edges you are timing are microseconds wide.

• A two-channel oscilloscope, with the 20 MHz bandwidth limit switched on for both channels. It strips the switching hash from the trace so the edges you cursor are the edges that matter.
• Two 10x passive probes. Match each channel's probe attenuation setting to 10x before you read a voltage.
• The camera's exposure-active output: a spare terminal or line that indicates the shutter is open. Most machine-vision cameras provide one; the trigger section below covers the fallback if yours does not.
• A multimeter, to identify the controller's 0 V terminal before any clip goes near it.
• The production controller, light and cabling. A substitute channel measures the substitute, not your line.

Where does the probe go, and where must the ground clip never go?

Probe the channel single-ended: ground clip on the controller's 0 V terminal, tip on the switched output feeding the light, which reads a negative-going pulse from about 24 V toward 0 V on a constant-voltage channel — and never clip the scope ground to the light terminal itself.

The reasoning is the wiring. The light sits between the supply positive rail and the channel's switched terminal; when the controller fires, it pulls that terminal down and current flows through the LEDs. Probed against 0 V, the terminal rests near 24 V and drops toward 0 V for the pulse, so the falling and rising crossings of the midpoint are your timing landmarks. Some channels switch the high side and give a positive-going pulse; the edge times matter, not the polarity.

To measure across the light itself you need a true differential probe, which reads the difference between two live nodes without an earth reference. If the probe point lives in an adapter lead or extension, confirm the connector pinout and current rating first; the light controller to light cable adapters guide covers what those leads have to carry, and the lighting accessories range holds the spare leads and breakout parts.

How do you trigger the scope from the camera exposure?

Trigger the oscilloscope from the camera's exposure-active output on channel 1 — edge trigger, normal mode, rising edge — with the light pulse on channel 2, so every capture starts at the same instant and the delay between the two edges reads directly in microseconds.

Route the exposure-active signal, optocoupler or line driver as your camera provides, into channel 1 and the strobe probe into channel 2. Set the threshold at the middle of the signal swing and the timebase to 100 µs/div, which gives a 1 ms screen: a 500 µs exposure and its strobe pulse fit with room around both edges. What the two traces show is precisely the relationship the trigger call versus exposure time guide explains — the trigger arrives, the shutter opens, and only the overlap lights the part.

If the camera has no exposure output, trigger the scope on the strobe pulse itself and take the overlap evidence from the image instead: steady brightness across ten frames, no dark strip, no flicker, means the pulse sits inside the shutter.

How much dead time do you leave between the edges?

Plan 20-30 µs of dead time at each edge — exposure opening to strobe rise, strobe fall to exposure closing — enough to absorb the few microseconds a driver needs to switch, the camera's shutter settling, and the small offset between the exposure signal and true integration.

Three offsets eat into a nominal alignment, and none of them is stable to the single microsecond. The driver takes a few microseconds to build current, and the optical output lags the electrical edge by roughly that much again; the LED strobe rise time guide shows how to measure the optical edge directly. The camera needs a moment to open the shutter cleanly, and the exposure-active signal leads or lags true integration by a few microseconds depending on the sensor family. A rolling shutter adds one more constraint: rows start sequentially, so the pulse has to sit inside the whole window, not the average of it. The settings then follow from two subtractions:

Here t_exposure is the measured exposure width in seconds, t_dead_open and t_dead_close the chosen margins, and v the part speed in metres per second.

How do you turn cursor readings into controller settings?

With a 500 µs exposure and 30 µs of dead time at each edge, the strobe width you type into the controller is 440 µs, the lit window covers 88% of the exposure, and the two cursor measurements from the scope produce exactly those two numbers.

1. Confirm polarity and idle level: fire one pulse with capture stopped and note where the trace rests and which way it moves.
2. Arm the exposure trigger as above, at 100 µs/div, with both probes matched to 10x attenuation.
3. Capture ten frames from the production trigger and cursor two numbers on each: the delay from exposure rise to pulse rise, and the pulse width at its 50% crossings.
4. Compare both margins with the 20 µs rule. A first measurement on a 500 µs exposure might read 12 µs at the opening edge and 8 µs at the closing edge — too tight, and the rare dark frame this line throws is that 8 µs margin losing a jitter race.
5. Recompute and write the settings: 500 µs − 30 µs − 30 µs = 440 µs of pulse, so set the delay to 30 µs and the width to 440 µs. Re-run and confirm both cursor margins now read about 30 µs.
6. Check what the fix costs optically: 440 / 500 = 88% of the exposure now delivers light, and at a 150 mm/s belt the blur inside the lit window is 0.15 m/s × 440 µs = 0.066 mm — about one pixel on a 120 mm field spread over 2,000 pixels.

Controllers name these fields differently, so it helps to see one laid out; a strobe on-time configuration reference shows the delay, width and exposure-gating fields in one place. The light in this worked example is the LS-BAR-1047, a 290 x 30 mm, 24 V bar light — the sort of standard bar light most strobed belt lines already carry.

Why does the voltage pulse look stranger than the light?

The voltage waveform always looks worse than the light: ringing and overshoot at the edges come from wiring inductance and the LED's own impedance, while the current pulse and the optical output are much squarer, so judge timing at the 50% amplitude points and ignore the spikes.

What you see at the terminal is the voltage across a wiring harness and a semiconductor load, not the light. Cable inductance rings at every edge, the LED forward voltage curves with current, and a constant-current controller deliberately moves that terminal around to hold current steady. The current through the LEDs, and with it the optical output, follows a far squarer profile than the wobble on screen. That is why the pulse-width cursors land on the 50% crossings, and why chasing the ring with snubbers or shortened leads is work the image never notices.

When you need to see the light itself, a photodiode wired across a 10 kΩ resistor into the scope shows the optical pulse; the amplitude proves nothing absolute, but the edges confirm the optical delay against the electrical one. A sagging plateau at the start of the pulse is a different fault — supply and cable losses under load — and the strobe current droop guide covers that diagnosis separately.

Which numbers do you log so the timing holds in production?

Log four numbers before you close the cabinet — strobe delay from exposure open, strobe width, jitter over ten captures, and duty cycle — because a 440 µs pulse firing at 8 parts per second is 0.35% duty and roughly 1.8 mA average from a 500 mA peak channel.

Jitter is the spread of the delay cursor across those ten captures. A few microseconds is normal, and as long as it stays well inside the 20 µs dead time, no capture clips. Duty cycle is width times trigger rate: 0.000440 s × 8 = 0.00352, so 0.35%, and a channel peaking at 500 mA draws about 500 mA × 0.00352 ≈ 1.8 mA on average — numbers that say the driver is loafing and the LEDs barely warm on each pulse. When controller and light are mismatched on pulse current, the damage shows up as heat and droop, not timing; the strobe duty cycle matching guide works that matching in full.

One capture before sign-off: repeat the overlay after an hour of running or at the start of the next shift. If delay or width has moved, you have a thermal or trigger problem rather than a settings problem, and the lighting stability test guide is the routine for that sign-off.

Which controller and cabling suit strobe timing work?

Order a controller that documents a strobe output mode and states its per-channel current, plus a spare extension cable so the scope lands on the same terminals the machine runs, and match the channel count to the lights you will fire in one frame.

Constant-current channels repay the extra effort on strobed lines because brightness is defined per pulse rather than per terminal voltage; the light controllers range filters directly on strobe and constant-current features. A channel of the kind this article scopes is the LS-CTL-0070, a 4-channel, 200 W digital control box with RS232 and a strobe output mode.

Keep the probe point permanent: a 5 m light extension cable in the loom means the bench scope lands on the same terminals the machine runs, and the accessories finder holds the adapters for the odd connector.

The light controllers finder, the lights finder and the controller configurator are the fastest routes from these measurements to an order. If the timing measures clean and the image still misbehaves, the fault was never the strobe — send the scope captures through the contact page and we will go through them with you.