External sync for a digital camera vs analog HSYNC and VSYNC

An analog CCIR camera danced to your timing because you fed it two TTL pulses: HSYNC opened every line, VSYNC opened every field. Engineers moving to a GigE Vision or USB3 Vision camera often plan to reproduce both signals, then find one trigger input and no sync connector on the camera. That single input is not a downgrade; it replaces the whole scheme. A digital camera owns its scan timing, and one rule falls out of that: one frame-start edge, plus a strobe output if the light must follow. Below is what HSYNC and VSYNC actually do, why the single edge is enough, and how to wire the DAQ counter, the laser and the light.

What HSYNC and VSYNC do on an analog camera

Composite analog video is one serial voltage. Pixel brightness travels as a waveform, one line at a time, and the frame grabber's digitizer must sample each line at the same pixel positions in every frame. HSYNC is the line mark: a pulse of about 4.7 µs at the start of every line, at 15,625 lines per second on CCIR, one line every 64 µs. VSYNC is the field mark: 50 fields per second, interlaced 2:1, so 25 full frames per second, with 576 of the 625 lines carrying picture. The American RS-170 standard runs the same structure at 525 lines, about 63.5 µs per line and 30 frames per second.

External synchronization slaves the camera's scan to pulses you generate. A multifunction DAQ board produces both trains from its counter outputs: H-drive locks the camera's line clock, V-drive locks the field boundary, and camera and grabber stay pixel-aligned. Two conditions make this chain brittle. A standard analog camera only locks at the standard rates — a camera whose timing deviates from CCIR or RS-170 needs a grabber configured for its specific video timing. And every element must agree on both clocks, or lines smear and fields tear. That is the machinery a digital camera deletes.

Why one pulse replaces both

A digital camera times itself. An internal oscillator and PLL clock the sensor's lines, and the line and frame boundaries live inside the camera and inside the data it sends. On a Camera Link camera they travel as dedicated FrameValid and LineValid signals beside the pixels. A GigE Vision or USB3 Vision camera wraps whole images into packets, and the driver rebuilds the frame on the host. Nothing samples a voltage per line, so there is no pixel position to protect and no HSYNC to feed.

What remains is one decision: when does a frame start? That is the hardware trigger input. One 5 V TTL edge starts one exposure and one full readout — your old VSYNC, promoted from field pacing to frame pacing. Line timing inside the frame cannot be forced from outside, and no area-scan task needs it to be. If you are weighing the interface choice itself, analog versus digital cameras for machine vision covers it system by system.

Two ways to wire it

Scheme one fits an event that must lead the frame — a laser step, a flash, a part arriving under a sensor. Your DAQ counter fires the event first, then edges the camera trigger, and the camera integrates immediately. Budget the camera's trigger latency, typically a few tens of microseconds, and keep the event ahead by latency plus margin.

Scheme two fits a free-running camera. The camera asserts a strobe or general-purpose output at the start of integration, and that edge gates the laser, the flash, or a strobe-capable light controller. The light then burns only inside the exposure window — the same trick as using the exposure time as the shutter. One counter output can drive camera and light in parallel, and one TTL pulse can be fanned out to several devices when you need buffered copies. Give every line a common ground: a missing common reference is the first cause when a trigger does nothing. When the pulse source is a PLC rather than a DAQ, wiring an external trigger from a PLC to an industrial camera covers the control side.

Worked example: a laser step at every frame

Take an interferometry-style job: step a laser's wavelength once per acquisition, capture, subtract the previous frame. Pace the camera from a DAQ counter at 30 Hz — external pulse trains set the frame rate exactly this way. The timing follows directly:

Set a 1.0 ms exposure and allow about 0.1 ms from edge to integration. Counter A steps the laser; counter B, started 1.1 ms later, edges the camera trigger. Every frame then lands on a settled wavelength, and the subtracted pair shows motion, not drift.

The 1.0 ms window is 3 % of the period. A light strobed at full current through the window collects 33 times the photons of a continuous lamp at the same average power — that ratio is the whole argument for strobing. Watch mains flicker: a 1.0 ms window can sit inside a trough of 100 Hz ripple and read dark, the mechanism behind black bars that track the lighting.

Bandwidth closes the loop. One 4000 × 3000 frame at 8 bits is 12 MB, so 30 fps streams 12 MB × 30 = 360 MB/s. USB3 carries roughly 350–400 MB/s in practice, so it fits; GigE carries roughly 110 MB/s, capping this resolution at about 110 / 12 ≈ 9 fps.

Analog and digital sync, side by side

The interlaced question

Buyers coming from CCIR often ask whether digital cameras do interlace. Current area scan cameras read out progressively — the area scan cameras category filters by global versus rolling shutter, not by fields, because there are no fields. Progressive readout is also what frame-subtraction work actually wants: every line is exposed in the same time window, so a fast event cannot land in one field and miss the other, as it can on 2:1 interlaced video.

Decision: keep 2:1 interlace only when a legacy device must receive CCIR video itself. That means an analog camera, a frame grabber that lists the camera as supported, and your H-drive and V-drive pulses into both — troubleshooting an external frame trigger on a frame grabber is the reference for that path. Everyone else buys progressive and triggers it.

Picking the parts

Pulsed illumination decides the shutter. A rolling shutter exposes lines in sequence across the frame, so a 1.0 ms pulse lights only the lines that were integrating during it; a global shutter integrates every pixel simultaneously and freezes the whole frame in one pulse. For any pulse shorter than a few milliseconds — or any pulse whose timing you do not control — take a global shutter.

The MV-CU020-90GC pairs a 1/1.7" global-shutter sensor with a GigE interface and a C-mount.

When you want more pixels behind the same simultaneous exposure, the MV-CH120-11UC is a 12 MP USB3 camera in the high-speed version with a global shutter.

Where light stays on through the whole exposure, shutter type stops mattering and resolution and interface take over. The LS-CMR-0166 reads 4000 × 3000 pixels at 30 fps over USB3.

The same edge that starts the frame can drive the light. The LS-CTL-0070 is a 4-channel, 200 W digital control box with a strobe feature, so camera and illumination fire on one shared trigger.

Checks before you order

• Confirm the trigger input level: 5 V TTL driven directly, or opto-isolated at 5–24 V through its series resistor pin.
• Get the trigger latency figure if the event must lead the frame. At 30 fps, 0.1 ms is 0.3 % of the period; at 1,000 fps the same 0.1 ms is 10 % and starts to eat the budget.
• Count the devices sharing the edge — camera, controller, laser — and buffer beyond two inputs.
• Set the acquisition mode to triggered before the first test: a camera that ignores the trigger is usually still in continuous mode, and double edges cause overtrigger errors.
• Compute the link: bytes per second = width × height × bytes per pixel × frame rate. Keep GigE under roughly 110 MB/s and USB3 under roughly 350 MB/s.
• Match trigger polarity and idle level across camera and light controller.

When you really do need line-by-line sync

Two jobs still want external clocks. A line scan camera advances one line per encoder tick, so it takes a clock input alongside the trigger — a different camera class, not an HSYNC cable. A linear sensor driven straight from a DAQ card takes both clock and trigger lines; driving a linear image sensor with a DAQ clock and trigger shows that pattern. Multiple area cameras on one event simply share the frame trigger in parallel; pixel-clock phase between cameras is not exposed on GigE Vision, and for nearly every task it does not need to be.

The decision

Three cases, three wirings:

1. Laser or event leads the frame. DAQ counter A fires the event; counter B edges the camera trigger latency-plus-margin later — 1.1 ms in the worked example. Take a global shutter when the pulse is short.
2. Camera free-runs, light follows. Camera strobe or GPIO output gates the illumination; size that half from the light-controller finder, where the strobe filter does the screening.
3. A legacy device must receive CCIR video. Stay analog: camera, grabber, and your H- and V-drive pulses, with the camera supported by that grabber. Everything else moves to a digital camera and one edge.

Size the camera on the camera finder, then let the camera configurator pair it with trigger cabling and a strobe-capable controller, or request a quote with your frame rate and pulse budget. More on picking and timing cameras lives on the cameras topic page.