The Line-Scan Camera You Picked by Line Rate Alone — Why It Still Misses Defects: TDI and the Resolution Design Relationship

Introduction — Why Doubling the Line Rate Made Defects Harder to See

On inspection lines where PCB panel arrays move continuously on a conveyor, it’s common to select equipment based on nothing more than the line rate (kHz) printed on the camera’s spec sheet. But right after pushing that line rate up to speed the inspection along, you can run into the paradoxical situation where small defects — micro solder bridges, pad-loss — leak through even more often than before. The cause is usually not the line rate itself, but how much actual signal the sensor captures at that line rate, and how little contrast the lens loses in delivering that signal. The shorter the exposure time, the less light per line, and once a low-contrast defect boundary is buried in noise, no algorithm — however sophisticated — can reconstruct information that was never captured to begin with. Leave this unaddressed and defects that should have been screened out at the early inspection stage leak through to downstream processes and final assembly, coming back as rework costs that far outweigh the productivity gained from the higher inspection speed.

The solution isn’t a single speed metric like line rate — it’s tying TDI (Time Delay Integration) signal accumulation, lens-sensor MTF (modulation transfer function) matching, and pixel resolution matched to the target defect size together into one unified design value.

Body — Where Speed, Signal, and Contrast Meet

P (Point): A line-scan camera’s resolution should be defined not by pixel count, but by “pixels per defect size.”

R (Reason): A PCB pad or solder-mask surface, where matte green resin and semi-glossy copper pads are mixed together, is a target with wide reflectivity variation. To detect a 30–50 µm-wide micro-bridge or pad-loss defect on a surface like this, at least 3–5 px need to span a single defect boundary for the algorithm to distinguish that boundary from noise. If you raise the line rate while leaving pixel pitch unchanged — that is, only increase transport speed — cross-web resolution stays the same, but down-web (transport-direction) resolution degrades in inverse proportion to transport speed, producing a directional problem where defects disappear with the same camera.

E (Example): Teledyne DALSA’s Linea HS2 8K TDI camera is reported to maintain 5 µm pixel resolution even at a 1 MHz line rate, because the TDI structure accumulates signal from multiple lines in the charge domain, compensating with signal accumulation for the light lost to shorter exposure time. By contrast, Allied Vision’s allPIXA pro 6000px, a 3-line CCD-based color line-scan camera, covers the 400–1000 nm band (visible to near-infrared) at 6000 px and a 10 µm pixel pitch at a 34 kHz-class line rate — a design that weights spectral information and color-contrast capture over raw speed. Summarizing both products as simply “high line rate” misses the actual difference in defect-detection performance between them.

P (Point, restated): So what needs checking when selecting a camera isn’t the maximum line-rate figure — it’s the down-web pixel density actually achieved at the target transport speed, and the MTF the lens sustains under that condition.

The same principle applies on the lens side. Use a lens whose image circle wasn’t generously designed to match an 8K line-scan sensor’s diagonal length, and contrast drops sharply toward the sensor’s edges, making it hard to expect uniform detection performance across the full panel width. In machine-vision practice, a lens’s effective resolution is typically defined as the spatial frequency (lp/mm) at which 20% contrast is still maintained; however high the sensor’s own resolution, if lens MTF falls short of this bar, software post-processing cannot restore contrast that’s already been lost. This is the same principle that hardware optical-setup defects cannot be fully covered by software image processing alone, applying just as much to line-scan systems.

Core Skeleton Standard

The following design values are based on a continuous PCB panel-array inspection line. Actual defect shape and material vary, so final parameters must go through sample testing.

Implication: None of these three items is set independently — each constrains the others. The design sequence has to be: defect size determines required pixel density, and pixel density determines the lens MTF standard required at that density. Follow that order and you avoid the situation where line rate goes up while actual detection power goes down.

A Caution for Comparison — TDI Line-Scan Isn’t Always the Better Choice

TDI has the advantage of securing sensitivity under low-light, high-speed conditions through signal accumulation, but on a line where transport speed is uneven or vibration is present, line-to-line registration error accumulates and can instead produce directional blur artifacts. In an environment like this, simple line-scan or area-scan may be more stable, and which side is actually favorable depends on the target’s reflectivity, transport stability, and required defect size — this cannot be confirmed before sample testing.

Field Checkpoints

• Has it been confirmed that the WD (working distance) between lens and camera is actually achievable under the target FOV/depth conditions, and that field placement is possible without mechanical interference?
• Has it been calculated whether down-web pixel density at the target transport speed meets the 3–5 px baseline relative to the minimum detectable defect size?
• Has it been confirmed via spec sheet that the lens’s spatial frequency at 20% MTF contrast sufficiently exceeds the sensor’s Nyquist frequency?
• When adopting TDI, has the transport line’s speed variation and vibration level been measured in advance to rule out line-registration error?

References

• LUCID to Showcase New 8K Line Scan, NIR, STARVIS, and 3D RGB-D Cameras at VISION 2026
• New 8k TDI camera supports 1 MHz line rate
• Allied Vision Launches allPIXA pro 6000px
• TDI line scan cameras: Precise results with fast movements and low light
• Optimizing System Resolution: Matching Lens and Sensor MTF
• 8K high resolution line scan lenses