Integrated strobe drivers in machine vision lights: what they change

Pulsing a machine vision light takes two hardware decisions: what makes the pulse, and what limits the current behind it. You can put both jobs in an external light controller, or buy a light with the strobe driver built in. Overdrive-class lights build it in: second-generation units pulse at up to 2000 strobes per second while their own protection circuit holds every LED colour inside its safe current envelope. This page explains what the integrated driver replaces, how the per-colour protection sets the pulse current, which numbers decide the choice — strobe rate, pulse width, duty cycle — and when a separate controller still wins. A worked example for a 20,000 bottles-per-hour filler shows the arithmetic to run before ordering either way.

What an integrated strobe driver replaces

A driver regulates LED current and switches it on and off; a controller is the external box that does the same for one or more lights and adds sequencing and interfaces. The practical distinction: the driver lives inside the light, the controller outside it. A light with an integrated strobe driver carries its switching regulator, pulse timer and protection logic behind two connections — a DC supply feed and a trigger input.

A light-plus-controller chain needs more parts: the controller, the cable and adapter pair that joins the two, and the trigger line into the controller. The integrated light takes the trigger straight in, usually from the camera's opto-isolated trigger output or a PLC transistor card. You get fewer connector pairs, a shorter trigger run, and no controller channel spent on the light — a four-channel box drives four lights, and an integrated light consumes none of them.

What external controllers still add is coverage: one box that sequences several lights, stores intensity digitally per channel, and swaps out without touching the lights. Why a machine vision lighting controller earns its place lists those jobs; the rest of this page covers when the integrated driver covers them anyway.

What the protection circuit sets: one current limit per LED colour

White, red, green, blue, cyan, UV and IR LED dice tolerate different peak currents, so a single fixed limit would strangle some colours and cook others. The integrated driver stores a limit per wavelength, and its microprocessor watches every strobe and clamps the pulse so the conducting colour stays inside its envelope. That is what makes overdrive routine rather than experimental: drive the LED at several times its continuous current for a short window, keep the average low, and output rises accordingly — makers of these light families quote up to five times a standard high-current light.

One eye-safety note that belongs here: UV and IR strobes are invisible, so shield the station and fit the enclosure window before the first test pulses.

Two formulas govern the envelope:

A 10 A peak pulse at 10% duty averages 1 A — that trade is the whole game. The envelope itself is fixed inside the light: nothing wired on the trigger side raises it, because the driver clamps the pulse instead of passing it through. If the image is still short of light at the limit, the fix is a bigger light or faster optics, not more current. The manual overdrive arithmetic for ordinary controller-driven lights is in LED overdrive and current control, and pulsing extends LED life too — see why strobe mode makes lights last longer.

The interface: supply, trigger and intensity

An integrated-driver light still asks two things of you: power and a trigger. Most ring and spot lights in the shop run on 24 V DC, and the integrated driver takes the same kind of feed — size the supply on the light's rated input current.

The trigger input comes in both transistor flavours: second-generation integrated-driver lights expose NPN and PNP strobe inputs, so one light accepts either a sinking signal from an NPN sensor card or a sourcing pulse from a PNP output. The trigger-high time is the light pulse: set the pulse width there, keep it above the light's minimum, and leave headroom for the optical rise time, which eats a few microseconds at the start of every pulse. Triggering from a robot or PLC is the same wiring problem — trigger the strobe from robot I/O shows the polarity mapping.

Intensity is set once and held: a 0-10 VDC analog input or an onboard potentiometer on second-generation lights. The driver turns that setting into pulse current inside the envelope, so brightness at commissioning repeats shot after shot.

Strobe rate: what the 2000 SPS ceiling limits

Second-generation integrated-driver lights are specified at 2000 SPS (strobes per second) — one pulse every 500 µs, because 1 / 2000 = 0.0005 s. The rate you need follows from the line:

At one shot per part, a 2000 SPS driver covers 7.2 million parts per hour (2000 × 3600), so on discrete parts the ceiling almost never binds. It starts to matter in multi-shot work: stacked pulses per part for multi-exposure images, several strobed lights on one station, or continuous strobing on a web. Ceiling and duty cycle then arrive together — at 2000 SPS a 200 µs pulse already sits at 40% duty (2000 × 0.0002 s = 0.4), and combinations past the LED envelope get clamped. Overdrive, duty cycle and timing for strobed lights works the duty side in detail.

Worked example: strobe rate and duty cycle on a 20,000 bottles-per-hour filler

Inputs: a filler running 20,000 bottles per hour (a typical mid-size line), one camera and one light per inspection point, one shot per bottle, and a 200 µs light pulse chosen to freeze bottle motion — the blur arithmetic is in how strobe pulse width freezes motion blur. The driver ceiling is the 2000 SPS printed for second-generation integrated-driver lights.

Rule in words: convert line rate to inspections per second, compare with the strobe ceiling, then multiply by pulse width for duty cycle.

Arithmetic:

• Line rate: 20,000 bottles/h / 3600 s/h = 5.56 bottles/s.
• Required strobe rate: 5.56 SPS at one shot per bottle.
• Fraction of ceiling: 5.56 / 2000 = 0.0028, about 0.3% — a headroom factor of roughly 360.
• Duty cycle at the inspection rate: 5.56 × 0.0002 s = 0.0011, about 0.11%.
• Even five stacked shots per bottle: 5 × 5.56 = 27.8 SPS, 1.4% of the ceiling.

Result: the driver idles at about 0.3% of its strobe ceiling and 0.11% duty, so average LED power sits near a thousandth of peak.

Decision: on this line the driver spec is not a selection criterion — pick the light by field of view, working distance and emission angle. The binding constraints are pulse width against motion blur and the optics themselves; the 2000 SPS ceiling leaves the strobe side unambiguous.

Integrated driver or separate controller: how to choose

The single-light station — one camera, one ring or spot, overdrive wanted, machine at 24 V — is the integrated driver's home ground: shortest trigger run, no channel spent, nothing to configure beyond intensity. Ring lights and spot lights cover that geometry, and the drive decision rides on top of the optics.

A 120 mm ring at 60° in 620 nm gives strong colour contrast on many printed and plastic parts and runs from a 24 V feed, while a compact 4 W spot reaches small fields inside tight housings:

Reach for a separate controller once the station grows past one light: staggered pulses for a lighting sequence, one intensity policy across many stations, digital control from the cell, or a spare box that covers any light in the plant. Shop controllers run from 1 to 16 channels and many carry a strobe feature; a four-channel digital control box with 200 W total output, an RS-232 port and strobe support runs a typical four-light station:

For large line, bar or multi-array lights, size the constant-current class instead: high-power constant-current units run two channels at 240 W maximum output per channel, 320 W total:

On webs and line scan cameras the controller question has its own answer — see what lighting controllers fix on line scan cameras.

Checks before you order

• Strobe ceiling of the exact model: 2000 SPS is the second-generation integrated-driver figure, so confirm the number printed for the light you buy.
• Shots per part times parts per second against that ceiling, with margin for reject re-shots.
• Trigger polarity on both ends: sourcing (PNP) output into the PNP input, sinking (NPN) into the NPN input.
• Pulse width: at or above the light's minimum, short enough for the motion-blur budget, with rate × width inside the printed duty limit.
• Intensity path: 0-10 VDC from a DAC or controller output, or the onboard potentiometer — decide who sets it and who can change it later.
• Supply sized on the light's rated input current, not on its continuous-mode output.
• Multi-light stations: count channels and check sequencing support on the controller side.

Where the usual answer is wrong

The first mistake runs one way: a controller is not automatically the safer choice. For a single overdriven light, the integrated driver is fewer parts, fewer connector pairs and one less failure mode, and its per-colour protection is built for exactly the overdrive you are about to ask of it.

The mirror mistake assumes integrated means no controller ever. The moment two or more lights must pulse in a timed order, or brightness must be held against LED ageing by feedback, a separate controller is the right tool — closed loop lighting control shows the feedback version. And when the trigger line blinks but the image comes back black, the fault is timing overlap, not hardware: strobe on but image black walks the check.

Decide in this order: optics first, then strobe rate and duty against the numbers above, then drive side. The light finder and the controller finder filter both halves, the configurator walks the pairing, and the lighting guides carry the optics side in depth. If the station sits between the two columns, send the loop sketch through the contact page — expert support comes with every order.