![\"Camera](\"https:\/\/pickandplacemachine.com\/wp-content\/uploads\/2026\/08\/Camera-and-Lighting-Degradation-Warning-Signs-in-SMT-Vision-Systems.jpg\")
<\/figure>\n\n\n\nAOI System Upkeep Begins With Image Security, Not Dish Tuning<\/h2>\n\n\n\n
An AOI picture is not simply a photo. It is the final outcome of a regulated measurement chain:<\/p>\n\n\n\n
LED source \u2192 optical diffuser \u2192 reflection geometry \u2192 lens \u2192 photo sensor \u2192 exposure setups \u2192 photo processing \u2192 problem classification.<\/p>\n\n\n\n
Adjustment any web link and the gauged look of the board adjustments.<\/p>\n\n\n\n
A solder fillet may become darker since the joint is faulty. Or because the green lighting network has actually deteriorated. An element side might end up being soft since positioning has actually shifted. Or since the lens is contaminated. A polarity mark might disappear because the element provider transformed its ink. Or because automated exposure is masking a decreasing light source by raising gain.<\/p>\n\n\n\n
The software program sees pixels. It does not understand your maintenance background.<\/p>\n\n\n\n
I utilize a blunt analysis regulation: when a number of unconnected component families start failing in the very same picture area, suspect the imaging chain prior to blaming the production procedure. Process faults typically adhere to materials, packages, printers, placement heads, reflow zones, or particular dishes. Optical faults commonly adhere to video camera sights, illumination sectors, board coordinates, time of day, thermal state, or device vibration.<\/p>\n\n\n\n
This matters on production lines where a positioning system such as the Hanwha SM482 And also pick-and-place maker<\/a> feeds a post-placement or post-reflow evaluation station. Positioning data and AOI data need to be associated, not dealt with as contending versions of truth.<\/p>\n\n\n\n In one 2023 industrial vision research study, scientists checked around 2,000 video camera components under stable lighting and reported precision above 99.92%. The figure is impressive, however the phrase “stable light condition” carries more functional weight than the headline precision. Controlled lighting belonged to the outcome.<\/p>\n\n\n\n 1. Direct exposure time or gain maintains increasing<\/strong><\/p>\n\n\n\n Automatic exposure can hide a fading source of light, a contaminated lens, or declining sensing unit level of sensitivity. The picture may look acceptable on the operator display while exposure rises from 1.8 ms to 2.4 ms, after that 3.1 ms.<\/p>\n\n\n\n That is not adaptation. It is proof.<\/p>\n\n\n\n Fad exposure, gain, aperture setup, and light-controller result by dish. A continual increase of 10% to 15% under unchanged manufacturing conditions deserves evaluation. This is a sensible trigger, not a global OEM limit.<\/p>\n\n\n\n 2. Edge sharpness decreases throughout every element course<\/strong><\/p>\n\n\n\n Blur influencing QFP leads, 0201 resistors, silkscreen characters, and fiducials at the same time is unlikely to be four separate process failures.<\/p>\n\n\n\n Examine focus, lens contamination, video camera resonance, mechanical play, and sensor temperature level. Contrast the same high-contrast edge against a validated golden photo utilizing slope stamina, modulation transfer, or an easier edge-width measurement.<\/p>\n\n\n\n 3. Issues continue to be connected to one camera view<\/strong><\/p>\n\n\n\n Multi-camera and 3D AOI systems can create a suspicious pattern: one side electronic camera creates duplicated lifted-lead or insufficient-solder calls while other views remain secure.<\/p>\n\n\n\n Turn the board or move a controlled examination discount coupon. When the anomaly remains with the video camera instead of the physical board place, the imaging network becomes the leading suspect.<\/p>\n\n\n\n 4. Hot pixels, dead pixels, or fixed-pattern sound multiply<\/strong><\/p>\n\n\n\n Record a dark framework with the lens covered and the same gain used during production. New bright factors, upright bands, amplifier glow, or unpredictable pixel clusters can show sensor or electronic devices destruction.<\/p>\n\n\n\n One faulty pixel may be harmless. An expanding cluster under an essential evaluation home window is not.<\/p>\n\n\n\n 5. The photo adjustments throughout equipment warm-up<\/strong><\/p>\n\n\n\n A cold AOI might pass its gold board at 07:00 and produce alarms after two hours of operation. Thermal development can affect electronic camera placement and focus, while LED output and motorist habits can shift with temperature level.<\/p>\n\n\n\n Contrast cold-start, 30-minute, two-hour, and end-of-shift pictures. Temperature-dependent drift is still drift.<\/p>\n\n\n\n 6. Calibration offsets grow after every solution<\/strong><\/p>\n\n\n\n Consistently rectifying X-Y setting, height, emphasis, or camera geometry without locating the physical reason misbehaves maintenance technique. Calibration should deal with well-known tolerances, not end up being a routine that hides loosening up mounts, birthing wear, cable television stress, or vibration.<\/p>\n\n\n\n 7. Various boards show the same spatial artefact<\/strong><\/p>\n\n\n\n A haze in the upper-right edge, a vertical brightness red stripe, or a localized comparison loss that shows up across several items factors towards the camera, lens, light, diffuser, or optical home window.<\/p>\n\n\n\n The board modifications. The artifact stays.<\/p>\n\n\n\n That is the idea.<\/p>\n\n\n\n 8. Photos look “normal” but gauged pie charts wander<\/strong><\/p>\n\n\n\n Human vision is inadequate at identifying progressive change. Operators adapt to dimmer or warmer pictures, especially when the display screen automatically rescales illumination.<\/p>\n\n\n\n Track suggest strength, percentile values, local harmony, red-to-green-to-blue ratios, edge comparison, and saturation percentage. The pie chart frequently subjects destruction weeks before an operator complains.<\/p>\n\n\n\n Camera degradation usually minimizes photo integrity. Lighting degradation adjustments exactly how physical attributes are shared.<\/p>\n\n\n\n Those failures overlap, however they are not the same.<\/p>\n\n\n\n An SMT AOI might use ring illumination, dome illumination, low-angle light, coaxial light, structured forecast, or numerous wavelength channels. Each arrangement reveals various physical characteristics: part edges, solder shape, polarity markings, coplanarity, reflectivity, or surface structure. Machine-vision recommendations from Cognex and Keyence similarly highlight that illumination geometry establishes contrast and assessment repeatability.<\/p>\n\n\n\n Look for these patterns:<\/p>\n\n\n\n Global intensity loss:<\/strong> The entire picture becomes darker, and direct exposure compensation climbs.<\/p>\n\n\n\n Regional falloff:<\/strong> One side or quadrant sheds illumination, commonly indicating an LED sector, diffuser, adapter, or placement issue.<\/p>\n\n\n\n Color-channel imbalance:<\/strong> Red, green, or blue ratios change even though the board, part lot, and video camera setups remain unchanged.<\/p>\n\n\n\n Flicker or frame-to-frame variance:<\/strong> Chauffeur instability, loosened connections, trigger timing, power quality, or overheating might be involved.<\/p>\n\n\n\n Transformed glare actions:<\/strong> Solder joints instantly appear bigger, smaller, rougher, or flatter since the angle or diffusion of the light has altered.<\/p>\n\n\n\n Workout drift:<\/strong> Intensity or color supports only after duplicated flashes or prolonged procedure.<\/p>\n\n\n\n Localized darkness:<\/strong> A light, video camera, board assistance, conveyor rail, or mechanical cover has relocated.<\/p>\n\n\n\n Overheating is entitled to certain suspicion. LED lighting support keeps in mind that too much heat can decrease LED service life and advises strobing where proper to restrict thermal loading.<\/p>\n\n\n\n And below is my unpopular opinion: replacing an LED only after it goes dark is maintenance cinema. An examination light can continue to be visibly operational while coming to be metrologically unsuited.<\/p>\n\n\n\n A plant-wide AOI false call rate is nearly worthless by itself.<\/p>\n\n\n\n Intend the average price rises from 0.7% to 1.2%. Management sees a small adjustment. However segmentation might show that Cam 3 climbed from 0.4% to 4.8% on reflective QFN joints throughout the mid-day change while every other network continued to be stable.<\/p>\n\n\n\n That is a maintenance signal.<\/p>\n\n\n\n Sector false calls by:<\/p>\n\n\n\n The 2024 OANet research study matters below due to the fact that it argues that image-only inspection leaves beneficial machine parameters unused. For upkeep groups, that means cam settings, controller information, ecological worths, and historical image data should be assessed next to classification outcomes– not stored in separate systems and failed to remember.<\/p>\n\n\n\n The economic context is not small. SEMI reported that worldwide semiconductor tools billings reached US$ 30.38 billion in Q3 2024, up 19% year over year. As devices financial investment rises, tolerating unmanaged destruction in evaluation properties becomes harder to safeguard.<\/p>\n\n\n\n Do not clean the lens merely since the AOI discovered more solder problems. Do not rewrite the program just since manufacturing urges nothing changed.<\/p>\n\n\n\n Test both cases.<\/p>\n\n\n\n Begin with a validated gold board or calibration artifact. Run it consistently under secured direct exposure, gain, focus, lighting outcome, conveyor setting, and dish revision. If the captured image adjustments while the artefact does not, the vision system has relocated.<\/p>\n\n\n\n Then compare the AOI trend against upstream and downstream procedure proof.<\/p>\n\n\n\n An increasing insufficient-solder telephone call price might come from stencil obstruction, paste problem, print pressure, or separation speed. Regular pattern treatment, including tools such as an MF310D double-pneumatic pattern cleaning device<\/a>, assists get rid of one resource of uncertainty prior to designers condemn the electronic camera.<\/p>\n\n\n\n Post-reflow alarms need to likewise be checked versus thermal evidence. A Wickon Q10 thermal profiler for SMT procedure verification<\/a> can aid determine whether solder appearance altered since the optical system drifted or due to the fact that the reflow profile transformed.<\/p>\n\n\n\n8 Camera Destruction Warning Signs Engineers Generally Miss<\/h2>\n\n\n\n
AOI Lights Degradation Has a Various Finger print<\/h2>\n\n\n\n
![\"Camera](\"https:\/\/pickandplacemachine.com\/wp-content\/uploads\/2026\/08\/Camera-and-Lighting-Degradation-Warning-Signs-in-SMT-Vision-Systems-1.jpg\")
<\/figure>\n\n\n\nFalse Telephone Calls Disclose the Failure Pattern– When You Segment the Information Properly<\/h2>\n\n\n\n
\n
Different Vision-System Drift From Real SMT Process Drift<\/h2>\n\n\n\n
![\"Camera](\"https:\/\/pickandplacemachine.com\/wp-content\/uploads\/2026\/08\/Camera-and-Lighting-Degradation-Warning-Signs-in-SMT-Vision-Systems-4.jpg\")
<\/figure>\n\n\n\n![\"Camera](\"https:\/\/pickandplacemachine.com\/wp-content\/uploads\/2026\/08\/Camera-and-Lighting-Degradation-Warning-Signs-in-SMT-Vision-Systems-3.jpg\")
<\/figure>\n\n\n\n![\"Camera](\"https:\/\/pickandplacemachine.com\/wp-content\/uploads\/2026\/08\/Camera-and-Lighting-Degradation-Warning-Signs-in-SMT-Vision-Systems-2.jpg\")
<\/figure>\n\n\n\n