Multi frame depth camera trigger pulse width: the math

You trigger a depth camera with a hardware line, ask it for three frames, and the sensor's documentation tells you to hold the line high for 160 ms. That number is not arbitrary and not negotiable: it is a fixed setup time plus one frame period per frame. This explainer breaks the arithmetic open — where the setup time and the 33 ms come from, what a burst really costs in frames per second, and why the setup time returns on every burst. By the end you can size the pulse for any frame count and frame rate, budget multi-camera sequences, and know when triggered burst capture is the wrong tool.

The answer: setup time plus N frame periods

A hardware-triggered depth camera that captures N frames per burst needs the line high for the whole burst:

With the two numbers used throughout — a 60 ms setup time and a depth engine fixed at 30 fps — the frame period is 1 s / 30 = 33.3 ms, which the sensor family's documentation rounds to 33 ms. Three frames therefore need:

T_on = T_setup + 3 × 33.3 ms = 160 ms

And yes, that setup time lands on every burst. Each rising edge restarts the whole sequence: the projector switches on, the scan settles, the receiver locks to the sensor's own illumination, and only frames integrated after that lock count toward your N. Nothing stays warm between bursts. That recurring cost sets the effective frame rate of the whole scheme, and it is why the mode is specified for static scenes rather than video.

The setup figure and the fixed 30 fps belong to one scanning depth sensor family; if yours prints different figures, substitute them in the same formula and every number below scales with them.

What three frames really cost

The burst itself runs at 3 frames / 0.160 s = 18.75 fps — call it 19 fps, already a third below the sensor's free-run rate. The instructive numbers appear when you change N:

• One frame per trigger: T_on = T_setup + 33.3 ms = 93.3 ms, so the ceiling is 1 / 0.0933 s = 10.7 fps. Triggering every frame, the way you would with an area camera, hands two thirds of the rate to setup.
• Enforced off time: with 40 ms of guaranteed low between bursts the cycle is 200 ms, and the rate falls to 3 / 0.200 s = 15.0 fps. Dead time comes straight off the top.
• Overhead share: at three frames, setup is 37.5% of the high time.

Back-to-back means the line drops only long enough to re-arm. The ceiling stays at the sensor's own 30 fps however large N grows: setup amortizes, it never disappears.

Why setup recurs: the gate switches the projector

On this class of sensor the external gate does more than time an exposure — it switches the active illuminator. A scanning depth projector must start its light source, settle its mirror and reference the receiver to its own projected light before the first valid frame; that start-up is the setup time. Free-run mode never switches the projector off, so it pays the cost once at power-up. The triggered mode exists so the projector can be off between bursts, and the reason is interference: two depth projectors facing the same scene corrupt each other's patterns, and depth quality collapses where their fields overlap.

The working pattern is sequential bursts — camera A's 160 ms, then camera B's. Two cameras at three frames each take 2 × 160 ms = 320 ms per synchronized set, so each delivers 3 / 0.320 s = 9.4 fps. Four cameras take 640 ms and deliver 4.7 fps each. The master/slave wiring, arming order and delay settings for that chain are a separate subject, covered in multi depth camera synchronization: slave mode and trigger setup; the numbers above are the budget you feed into it.

Some time-of-flight cameras avoid the collision with microsecond-spaced illumination windows and a programmable camera-to-camera offset. A fixed-rate scanning depth sensor exposes no such control, so plan the sequential budget rather than waiting for a firmware feature.

One safety note, because it belongs with trigger timing: depth projectors are active infrared sources, some laser-based — respect the laser class on the camera label, never stare into the projector, and leave any interleaving the vendor built into the driver switched on.

For the camera side, the MV-DT01WMU is our time-of-flight sensing camera, and the 3D sensors category carries laser profilers and stereo structured-light cameras when a line of geometry or a single snapshot serves better than frames of depth.

Edge quality and the first-burst quirk

Two behaviors bite teams moving from area-camera triggering:

• Dirty edges. Ringing or bounce on the line can register as extra rising edges, and a marginal gate can clip a frame mid-integration. Scope the line at the camera end, and drive it from a source with a programmed width and a defined edge rather than a hand-built monostable. Ringing shows in the image too — diagnosing gray-level jumps on a pulse-triggered frame grabber walks through that failure mode.
• The first burst returns one frame less. On cameras that document multi-pulse triggering, the first trigger after arming delivers N − 1 frames; the count matches the preset from the second trigger on. Either budget T_setup + (N + 1) × T_frame of high time and drop the first frame, or discard the first burst in software.

Count the returned frames per burst during commissioning — a one-line check that catches both problems.

Driving the pulse

The source must hold a solid high level for a programmed width, on command, into the camera's opto-isolated input. A multi-channel light controller fits well because strobing LEDs is the same timing problem, and it leaves channels for the illumination you will add later.

The LS-CTL-0070 is a 4-channel, 200 W digital control box with an RS232 interface and a strobe feature, so one box sequences several cameras' bursts from a serial command stream.

For the wire into the camera's I/O connector, the LS-CBL-0300 is an M8 8-pin male-to-open cable with 8 cores, built for drag-chain use, 7 m long; land the trigger pair on the opto input and keep its return with it. Browse light controllers for other channel counts, and cables for other connector families.

Start at the controller step of the system configurator to size the pulse source together with the cameras and illumination.

When the 160 ms figure is wrong

• Different depth rate: at 15 fps the frame period is 66.7 ms and three frames need T_setup + 3 × 66.7 ms = 260 ms high. Carry the formula, not the constant.
• Exposure stretches the period: the true period is the larger of 1 / f and exposure plus readout, so long exposure settings push the rate below every number above. Depth camera frame rate drops when exposure settings cap the FPS covers the mechanism; maximum frame rate on a hardware trigger: exposure plus readout gives the arithmetic.
• The same wire means something else on area cameras: on many industrial cameras the pulse width is the exposure itself, not a frame count — why trigger pulse width sets your camera exposure time explains that contract.
• Software triggering has its own tax: depth streams requested through the SDK instead of the hardware line lose rate to mode and stream settings, as 3D camera frame rate low on software trigger shows.

The decision, case by case

• One depth camera, static scene, a few frames: use the triggered burst. Hold the line high for the setup time plus 33.3 ms per frame, and take at least three frames per burst to keep setup under 40% of the budget.
• Several depth cameras, static scene: sequence the bursts and multiply. Per-camera rate is N / (cameras × T_on): with three-frame bursts, 9.4 fps for two cameras, 6.25 fps for three.
• Continuous synchronized video of a moving scene: do not build it from facing active-illumination depth cameras. One depth camera for geometry plus synchronized 2D cameras for motion is the architecture that works — see stereo camera trigger synchronization with a shared pulse and multi-camera video capture synchronization for four cameras.
• One depth camera, continuous: run free-run with no gate, and let the depth mode and exposure set the rate.
• Several images per trigger on a 2D camera — different lighting or exposures, not depth frames — is a different pattern: multi image capture per trigger.

Checklist before you wire it

• Write T_on = T_setup + N × T_frame with your sensor's own two numbers, plus the off time your controller enforces.
• Compute the effective rate N / (T_on + T_off) and compare it with what the inspection needs.
• Count frames on the first burst after arming; expect N − 1 on cameras with the multi-pulse quirk.
• Scope the line at the camera end for ringing and clean edges.
• Match the controller output to the camera input's voltage and isolation requirements before ordering.
• Plan the camera order in multi-camera scenes so no two projectors face the same volume during a burst.

If the budget does not close, the camera finder lists other depth engines with different native rates, and the contact page reaches our applications team with your scene and rate requirement.