← Back to MVisionPro
About the library

MVisionPro Engineering Library · method

Setting the exposure for a moving part, and deciding whether the light should be pulsed

In short

Work out object-space sampling first, the pixels per millimetre on the part: resolution along the direction of motion divided by the field of view along it. The blur you allow then limits the interval in which the image is formed. Under continuous light that interval is the camera's exposure time, and the camera has to be able to set it. Under a short synchronized pulse that dominates the image, the pulse duration takes that role and the exposure may be longer, provided the pulse lies wholly inside a window where the rows you need are exposing and ambient light leaves no trail. Neither branch proves the picture is good enough; that is settled on a sample of the code region at the real speed and light.

Sampling, and how much blur you may allow

Sampling comes from resolution divided by field of view, never from the sensor's pixel pitch, which describes the sensor rather than the image of the object. The Camera and lens calculator reference carries the derivation.

One pixel of allowed blur suits detection and presence work, where a feature has to be found rather than measured; half a pixel is what our calculator applies to measurement, where a smeared edge moves the sub-pixel estimate the result rests on. Both are defaults of our tool rather than a normative tolerance (our calculator's defaults), though Basler publishes a comparable one-pixel rule. Tighten it when a smeared edge gives a wrong number rather than a retry: halving the allowance halves the limit.

Which interval counts, the camera exposure or the light pulse

Under continuous illumination the sensor collects light for the whole exposure, so the exposure time is the smearing interval and the limit must be a setting the camera supports. The specified minimum exposure matters here: if the interval you need is shorter than the camera's shortest setting, the answer is in the geometry or the speed.

Gardasoft describes short controlled pulses as the way to freeze motion, and the reason is the light, not the shutter. The part keeps moving for the whole exposure, but when nearly all the light forming the image arrives during the pulse, only the travel within that pulse is written into the picture, so the pulse duration replaces the exposure time in the blur calculation. The pulse must fall wholly inside the interval in which the rows you need are exposing, which Basler calls the flash window, and ambient light must stay weak enough to leave no visible trail, as Allied Vision requires of the flash in its camera manual.

The shutter decides how hard that is: a global shutter exposes all pixels over one common interval, so the window is the exposure itself. A rolling shutter exposes rows at different times, and a Global Reset mode, offered on some models, starts the rows together yet may still read them out one by one, so the timing must come from the model documentation.

What the shorter interval costs

Shortening the interval costs light: Cognex notes that shorter durations stop motion better but may need a larger aperture or more gain. The aperture is the adjustable opening in the lens, and the f-number expresses its relative size, so opening the aperture means moving to a smaller f-number. That lets in more light, narrows the depth of field, the range of object heights that stay acceptably sharp, and can bring aberrations and vignetting, as Basler notes. Gain, the amplification applied after the sensor, raises signal and noise together. A pulse is the remaining lever, concentrating the available energy into the short interval instead of spreading it over the cycle. Its timing is stated per product, so send us the light and controller part numbers and we confirm pulse width, delay and duty cycle from the vendor document.

Worked example: a code on a conveyor

The camera is 2448 × 2048 px with 3.45 µm square pixels (datasheet). The scenario assumes an object area of 160 × 120 mm in the picture, parts travelling along the image width at 1200 mm/s, an allowance of 1 px and a 0.30 mm code module against a 5 px target. It also assumes a working distance of 500 mm, which the calculation measures from the lens principal plane rather than the camera body, so check the mechanical distance against the lens documentation.

The frame is not four to three: with square pixels a 160 mm full-frame width comes with a 133.856209 mm full-frame height (calculated), so the 120 mm of object height sits inside the frame with margin. That requirement asks for 25.069221 mm of focal length and selects a standard 25 mm lens.

Recomputed on the 25 mm lens at 500 mm, the full field is 160.4664 × 134.2464 mm and sampling is 15.255530 px/mm, so at 1200 mm/s a 1 px allowance caps the interval at 54.625 µs (calculated). The requirement figures had been 15.3 px/mm and 54.466231 µs: rounding the focal length down widened the field and cost sampling while the limit barely moved.

The 0.30 mm module spans 4.58 px in the installed field, while the 5 px target corresponds to 0.32775 mm, so this module misses that target (calculated).

Under continuous light, 54.625 µs is inside the Standard exposure range of the camera below (datasheet); under a pulse, keep the pulse at or below that and the exposure may be longer.

Calculate for your case

Calculate for your case

The link carries the supported conditions; blurPx, fNumber are not part of the page and are available through the tools API.

The prepared link opens the calculator with the camera, both field dimensions, the working distance, the feature size, the pixel target, the speed and the motion axis set. For the selected pair the page shows a field of view of 160.5 × 134.2 mm, 15.3 px/mm and 55 µs, this example's figures after its own rounding. The line "At the minimum exposure time of 50 µs, calculated motion blur is 1.00 px" answers a different question: those 50 µs come from the 0.30 mm feature at the 5 px target, not from the pair you selected. "Provides 4.58 px instead of 5" is the same shortfall the example computes, so the page closes with "No complete pair; send the application to an engineer". The link carries neither the 1 px allowance nor the f/4 aperture; for either, use the optics route of the tools API and record the values you assume.

Equipment that fits this example

The MV-CA050-12UM matches the example geometry: a global-shutter CMOS camera in 2/3" format, with a Standard exposure range of 15 µs to 10 s and an UltraShort range of 1 to 14 µs (datasheet). The 54.625 µs continuous-light limit falls inside the Standard range, a range comparison and nothing more.

The MV-CT0MU0060/R is the rolling-shutter case: exposure from 8 µs to 1 s, with exposure modes that include Rolling and Global Reset (datasheet), usable in a pulsed setup only once its row timing is confirmed. Mount, interface, cable and lens image circle remain separate checks for both.

Common mistakes

When this rule breaks

The calculation models constant speed and nothing around it. Acceleration and vibration, motion out of the object plane, rolling-shutter skew, defocus and the decoder's own tolerance sit outside it, so a part that jitters on a worn belt can fail at a limit the arithmetic approves.

In line-scan imaging the line period sets its own limit: the required line rate is speed times sampling along the direction of motion, and the usable exposure is bounded by the line period, the duty cycle and any camera maximum, whichever binds first. Time-delay integration changes effective integration again, and the number of physical sensor lines does not reveal it.

Next step

Decide which branch you are in before choosing a light, and write down the allowance and aperture you assumed. Then validate on a sample: look at the code region for separable elements, clipped highlights and noise, and measure reading across the operating range the line requires.

Sources

Related

How this material was prepared

Prepared with MVisionPro AI agents from stated sources and the calculation core; MVisionPro retains editorial responsibility. A physical test or human engineering review is claimed only when explicitly stated. Read the editorial method.

Editorial status: verified. Content updated 2026-09-19.