Trigger multiple cameras from one opto-isolated output without sag

One master camera firing several slaves through its opto-isolated trigger output is the standard way to synchronize a multi-camera rig without extra timing hardware. The classic failure: one slave triggers, the rest never see the edge. The cause is rarely the camera. It is the pull-up resistor on the shared trigger line, sized as if it fed a single input. This guide gives you the loading math, the wiring that completes the opto circuit, a worked sizing for three slaves at 5 V, and the two measurements that confirm the line before you record.

What you need

You need one master camera with an opto-isolated output, one opto-isolated input per slave, and a small sync box or terminal strip holding one pull-up resistor and one common return. Most area scan cameras with a GPIO connector qualify, and the camera finder filters bodies by interface, resolution and shutter. Almost any GigE or USB3 area scan camera works as a slave; the MV-CU020-90GC, for instance, is a 2.5 MP GigE camera with a global shutter.

Run the trigger, its return and camera power on a power and I/O cable broken out to a terminal block:

The accessories finder lists longer and shorter leads if the rig sits further from the box. If you are unsure what has to reach the camera connector at all, read whether you need a camera GPIO cable for image acquisition. Add a multimeter for the checks in the steps, and a scope if one is at hand.

Why one slave fires and three do not

An opto-isolated trigger input is an LED behind the camera connector. It fires when the current through it reaches the activation figure in that camera's technical reference, and common optocoupler and camera input sheets specify that figure anywhere from about 1 mA to 10 mA. The master camera's opto output is a transistor that switches the shared line, and a single pull-up resistor feeds that line. Whatever current the resistor can deliver is shared by every input attached to it.

The evidence shows up on a multimeter. In one real four-camera rig, the master's output sat at 3.4 V with nothing connected, 1.84 V with one slave attached, 1.76 V with two and 1.68 V with three. Each opto input is a diode load dragging the node down to its conduction voltage, and each added input takes another share of a fixed current. One input received everything the pull-up could give. Three inputs each received a third of it, below every activation figure in the 1-10 mA band.

That collapse is load-line behaviour, the same mechanism behind an analog output that sags under load. The cure is not a better camera. It is a resistor sized for the whole load, plus a return path that actually exists. For the alignment side of a shared-trigger rig, multi-camera exposure synchronization with a shared trigger picks up where the wiring leaves off.

The circuit that works

Feed a 5 V rail in the sync box through a single pull-up resistor to the shared trigger line. The line goes to the input pin of every slave's opto pair. The return pin of every pair, the return pin of the master's opto output and the rail's ground all join at one common point. In the idle state the pull-up feeds every input LED its working current. When the master fires, its output transistor pulls the line down to the common, the LED current collapses and each slave registers the falling edge.

One reference fix matters as much as the resistor. If you derive the rail from the camera's own ground-referenced auxiliary supply, connect that supply's ground to the opto common as well. The isolation in an opto pair separates the camera's internals from this external circuit; it does not excuse the external circuit from having a complete loop.

Size the pull-up: the formula

V_rail is the pull-up supply, V_in is the line voltage with every input attached (the LEDs' conduction voltage, which you measure once), N is the number of slave inputs and I_in is the activation current from the input's technical reference. The first equation picks the resistor. The second checks the master's transistor, which must sink V_rail/R and still pull the line to a valid low. Aim for roughly twice the stated activation current per input when the specification is thin. If the low level will not fall, raise R and re-check the high state with the first equation.

Worked example: one master driving three slave cameras at 5 V

The inputs, and where each comes from:

• Three slave inputs: a four-camera rig with one master.
• V_rail = 5.0 V DC: the sync box supply.
• V_in = 1.68 V: measured on the rig with all three inputs attached; call it 1.7 V for the design.
• I_in = 2.2 mA per input: the working activation figure for these inputs.

The rule in words: the pull-up must deliver three times 2.2 mA at the conduction voltage. Choose the resistor:

R = (5.0 V − 1.7 V) / (3 × 2.2 mA) = 3.3 V / 6.6 mA = 500 Ω

The standard value just under it, 470 Ω, keeps the current above target. The fix on the real rig was exactly this move, to a pull-up in the 400-500 Ω band, and all three slaves then fired. Check both states with the fitted value and the measured conduction voltage:

• High state: I_total = (5.0 V − 1.68 V) / 470 Ω = 7.06 mA ≈ 7.1 mA, so each input gets 7.06 mA / 3 = 2.35 mA, above the 2.2 mA target.
• Low state: I_sink = 5.0 V / 470 Ω = 10.6 mA, which the master's output sank while holding a valid low.

For contrast, the 4.7 kΩ pull-up that logic boards fit out of habit delivers 3.3 V / 4.7 kΩ = 0.70 mA in total, or 0.23 mA per input across three: an order of magnitude short. The sweep around the working point:

The result moves little with the input's exact conduction point: at 470 Ω, inputs conducting anywhere from 1.2 V to 2.0 V receive (5.0 − 1.2)/470/3 = 2.7 mA down to (5.0 − 2.0)/470/3 = 2.1 mA each. Decision for this case: fit 470 Ω at 5 V for three 2.2 mA-class inputs, confirm the output's sink rating covers 10.6 mA, and rerun the formula whenever the rail, the camera family or the input count changes.

Wire it: the steps in order

1. Read each slave input's activation current and voltage range from its technical reference. Where the sheet is silent, design to the worked example's 2.2 mA per input.
2. Join the returns: the return pin of every slave input, the master output's return and the pull-up rail's ground land on one common. If the rail comes from a camera supply, that supply's ground joins too.
3. Fit one pull-up, sized from the formula, from the rail to the shared line. Never give each slave its own pull-up: parallel resistors halve the value you calculated, and rails that differ fight over the line.
4. Set each slave's trigger to its opto-isolated input and to the edge the master produces; these settings live in each camera's trigger configuration. The case rig armed its slaves on the falling edge, when the master pulls the line low.
5. Power up and measure the line, idle, with every slave attached. On a 5 V rail expect roughly 1.7-1.9 V, the conduction voltage. Then compute (5.0 − measured)/R and divide by the input count to get the per-input current.
6. Fire the master and measure the active level. The line must sit within a few hundred millivolts of the common while sinking the V_rail/R current, and the fall must be clean on a scope, without bounce.
7. Record a test clip and count frames. Every slave's captured frame count must equal the number of pulses the master sent; any shortfall means an input is still starving.

Common mistakes

• Pull-ups in the kilohm range, carried over from logic-board habit. They feed each input fractions of a milliamp, which is the one-slave-fires symptom exactly.
• Returns left floating. An opto pair without its return never lights, whatever the resistor value.
• Treating the idle conduction-voltage reading as a fault. The current, not the voltage, tells you whether the edge will register.
• One pull-up per slave on different rails. The line sits somewhere between the supplies, and the trigger jitter shows it.
• Fixing a weak edge by removing the resistor altogether.

When one master output is not enough

Past about four inputs, or whenever the edge looks slow at the far end of a long line, stop leaning on the master's transistor and buffer the trigger: one input, several driven outputs, each slave getting its own current. That is the budget logic behind routing one GPIO to six cameras on a serializer board, and a TTL fan-out from a microcontroller is the same arithmetic in another box.

For a two-camera rig, skip the box and trigger both USB cameras from one shared pulse. When wiring is impractical, PTP master-slave synchronization on GigE cameras aligns clocks instead of edges. More wiring-level guides sit on the systems topic page.

Checks before you order

• Count the slave inputs and collect each activation current. N and I_in come first; everything else follows from the formula.
• Check the master output's sink rating against V_rail/R, not just the edge polarity. At 5 V, three to six inputs on a 2.2 mA target need 250-500 Ω and a sink current of 10-20 mA.
• Count the cores you need: trigger line, returns, camera power, any strobe. The I/O cable needs a core for each, plus spares.
• Buy the resistor set once: standard values from 240 Ω to 1 kΩ cover this whole range, and you will tune once per camera family.

If your rig mixes camera families or the input sheets disagree, send the wiring sketch through the contact page and we will size the line with you.