Buffer relay between camera and strobe: when you need a PLC

You have a vision controller that raises its alarms as a binary word, a strobe beacon you want to flash on any alarm, and a quote pending for a small PLC, a cabinet and a weekend of ladder work. Before you sign it, separate the two jobs in that decision. Protection for the camera comes from galvanic isolation, and one opto-isolated solid-state relay delivers it. Logic — telling one alarm code from another — is the only thing a PLC adds. This guide gives you the sizing math for the relay, the wiring rules that hold the camera side to milliamps, and the cases where the PLC cabinet genuinely earns its space.

The decision: isolation is a relay's job, logic is a PLC's

If every alarm may fire the same flash, you need no PLC. OR the alarm bits through one diode per line into a single DC solid-state relay, and let the relay close the beacon loop. The camera output sees about 10 mA, the beacon sees its lamp current, and nothing ties the two sides together.

A PLC earns its place only when the strobe must behave differently per alarm code, when you want spare I/O for a line that will grow, or when the alarms must reach a reporting system. Even then the shape holds: the camera word lands on the PLC's isolated inputs, and the strobe is driven from a PLC output or from the trigger input of a strobe-capable light controller.

What you must never do is let the camera output switch the lamp itself. Camera I/O ports are built to switch tens of milliamps; lamp loops run hundreds of milliamps to amps. The rest of this guide puts numbers on that gap and on the relay that bridges it.

What actually protects the camera: the optical gap

A packaged solid-state relay has two sides and no copper between them. The input is an infrared LED; the output is a photo-transistor or a MOSFET pair that the LED turns on through the package. Because only light crosses the gap, a shorted lamp cable or a spike on the lamp loop cannot climb back into the camera port. That is the protection the opening case wanted from a PLC — and a PLC provides it the same way, because its input modules are opto-isolated too. The isolation lives in the optocoupler, not in the processor behind it.

Size the two sides as separate circuits:

• Camera side. Machine-vision I/O is specified in the tens of milliamps per port. Design the input loop to draw about 10 mA and any normal output drives it without sag or heat — the same budget as when one output triggers several cameras without sagging.
• Load side. A panel beacon draws a few hundred milliamps, and a four-channel inspection strobe controller lists total outputs in the 8 A class. The buffer's whole job is holding those amps off the camera side.

Two relay rules finish the protection. On the DC lamp loop, use a DC-output SSR with MOSFET outputs: AC SSRs switch with triacs, a triac turns off only at a current zero-crossing, and on DC that crossing never arrives — the relay latches on. And put a quick-break fuse in series with the loop, so a shorted cable clears the fuse instead of the relay's output stage.

For the input-side pinout — which terminal is the return, why the isolated side needs its own 0 V — see which side of a camera's opto-isolated GPIO circuit is the input, and the signal levels and pull-ups in wiring a GPIO output to a camera trigger input.

If the strobe runs on a light controller, the buffer is already built in

Distinguish the two strobes on a line. An alarm beacon warns people. An inspection strobe freezes motion for the camera. If the light you want to fire is the inspection light, the buffer already exists: a strobe-capable light controller takes a signal-level trigger input from the camera, and its own output stage carries the lamp current. The camera never sees the lamp loop, so there is no relay to add and no PLC to justify. Strobing also costs lamp life when the duty cycle runs high, so match light and controller with the numbers in LED strobe duty cycle: matching lights and controllers.

That is also the wiring behind enabling a smart camera's strobe output, and it holds when several cameras share one strobe: the trigger is a signal, the amps stay downstream. Keep the alarm beacon on its own relay — a plant beacon belongs to the alarm circuit, not to the imaging light's controller.

A four-channel digital control box with an RS-232 port and a strobe feature, sized for 200 W of 24 V lighting:

Worked example: one alarm word firing a beacon

The inputs, as the case poses them plus the values this sizing assumes:

• Alarm word: 16 binary lines at 24 V DC from the vision controller, healthy state all-low — the case asks for 16 inputs and binary-coded alarms.
• Load: one beacon strobe on that same supply, taken as 10 W here, a common panel beacon size; scale the arithmetic for your lamp.
• SSR input: a DC control input drawing 10 mA at full signal, with about 1.2 V across its input LED — typical figures for opto-isolated DC inputs.
• Camera port budget: tens of milliamps, so a 10 mA input sits comfortably inside it.

The rules, in words: Ohm's law sets the input current through one series resistor; the beacon current comes from its power rating divided by the loop voltage; the diodes only steer current and never make or break the load.

Arithmetic:

• R = (24 V - 1.2 V) / 0.010 A = 2280 Ω. The nearest standard value is 2.2 kΩ, which draws 22.8 V / 2200 Ω = 10.4 mA.
• Resistor dissipation: P = (0.0104 A)² × 2200 Ω = 0.24 W, so fit a 0.5 W part and let it run at half rating.
• Beacon current: I = 10 W / 24 V = 0.42 A. Rate the SSR output at twice that or better — twice 0.42 A is 0.83 A, so a 1 A-class part.
• One rectifier diode per line, 16 diodes steering into the single 2.2 kΩ input. Any non-zero code lights the SSR input, and each camera port carries at most 10.4 mA whether one bit or all sixteen are high, because the input current divides among the high lines.

Result: the camera side is loaded at about 10 mA per line while the load side carries 0.42 A — a 40:1 step handled entirely inside the relay. Decision, with its threshold: if any non-zero code may fire the same flash, 16 diodes, one 2.2 kΩ 0.5 W resistor and a 1 A-class DC SSR replace the whole input card. The moment code A must flash differently from code B, OR-logic is blind and you are into decode territory — the next section.

When a PLC genuinely earns its place

Three cases, and only these:

1. Decode. Different codes must drive different behaviour — a beacon at 1 Hz for a jam, 4 Hz for a fault, a second output for the andon stack. That is word-compare logic, and diodes cannot do it. The structure to copy is the one in vision system diagnostics to PLC: status and alarm mapping.
2. Counts and expansion. The case asks for 16 inputs and 16 outputs with spares — a 24-in/16-out plan. That is a defensible buy if the cabinet will take the rest of the line's signals next year, and an expensive one if the beacon is the only load: five points do that job.
3. Reporting upward. When alarms must reach a line PC, a database or a shift report, the PLC is the natural collector.

Even then, keep the shape: camera word into isolated PLC inputs, strobe driven from a PLC output into the beacon or into a strobe controller's trigger input. The camera output still never touches a lamp. One practical check before you buy the box: confirm the compact range you pick is supported by the programming software you already run, because newer compact hardware sometimes needs a newer software release. The same pattern applies when the source is robot I/O rather than a vision controller — see triggering strobe lighting from robot IO digital outputs.

Choosing and fusing the solid-state relay

• Output type: DC, MOSFET-output, for the DC lamp loop. An AC triac SSR latches on DC.
• Output rating: at least twice the load current — 1 A class for the 0.42 A beacon above; scale with the lamp.
• Input: DC control inputs in the 5-20 mA band are the norm, and the 2.2 kΩ resistor above lands at 10.4 mA, mid-band. A bare optocoupler works the same way with the same math — the resistor sizing in driving an LED from a GPIO pin with a MOSFET transfers directly.
• Fuse: quick-break, in series with the loop, so cable faults clear the fuse before the relay notices.
• Off-state leakage: a few milliamps, enough to make some LED beacons glow faintly in a dark plant. A bleed resistor of a few kilo-ohms across the lamp ends it.

For larger inspection cells the same idea scales, and the controller is still the buffer:

Both take the trigger as a signal and keep the amps on their output stages. Power the beacon loop from its own 24 V supply — the power supplies category has the enclosed units — and keep that rail separate from the camera's supply, so a lamp fault cannot drag the camera down with it.

When the usual answer is wrong

• A PLC in the imaging trigger path. PLC outputs update on the scan, adding milliseconds of jitter. The imaging strobe must fire inside the exposure over a wired path measured in microseconds: camera strobe output straight to the controller trigger input. The timing budget behind this is in camera strobe output timing: exposure pulse and image timestamp.
• No buffer at all. A camera output asked to sink a lamp dies quietly — a port built for tens of milliamps handed half an amp. The symptom arrives later: a trigger that worked on the bench and sags on the line.
• A mechanical relay for frequent strobing. Contacts arc and wear at high pulse rates. Solid-state for anything that pulses often; mechanical only for slow on/off duty such as a shift light.
• OR-logic on a word whose healthy state is not all-zero. Then any bit high fires on healthy lines too. Read the code table before trusting the 16-diode trick; a non-zero idle word needs the compare, so the PLC path.

Before you wire it: the short checklist

Camera side:

• Each input loop draws about 10 mA, inside the camera port's rating. Ports switch tens of milliamps; overloads cause sag, then dead outputs.
• One diode per alarm line and a 2.2 kΩ, 0.5 W series resistor on the alarm word, per the worked example.
• The isolated loop keeps its own 0 V, never shared with the lamp return. Sharing it deletes the isolation you built.

Relay side:

• DC MOSFET-output SSR on the DC loop; no AC triac SSR, which latches.
• Output rated at twice the beacon current or better, with a quick-break fuse in series. Cable shorts then clear the fuse, not the relay.
• Bleed resistor across an LED beacon, so off-state leakage cannot glow.

Logic side:

• All-zero healthy word and one flash pattern: diode-OR is enough, no PLC.
• Per-code patterns, spare I/O or upward reporting: PLC, with the camera word on isolated inputs.
• The PLC stays out of the imaging trigger path; scan jitter is milliseconds, the exposure budget is microseconds.

Where this leaves the build

For the case this guide opened with: no PLC for the strobe itself. Sixteen diodes, one 2.2 kΩ 0.5 W resistor and a 1 A-class DC SSR fire the beacon on any alarm at 10 mA per camera port, fused and isolated. Buy the compact PLC only if the codes must decode into different patterns or the cabinet must grow, and keep it off the imaging trigger path. Pick the light side from the light controller finder, step through the trigger stage in the controller configurator, and browse the rest of the strobe guides on the lighting topic page. When the cabinet is real, the quote form turns the parts list into an order.