A line-scan camera samples the part once per line, so the resolution along the motion direction is set purely by how far the belt travels between two lines. To hold 50 µm per line at 50 mm/s the camera has to run at 1 kHz — that is 50 mm/s ÷ 0.05 mm. Each line then has 1000 µs to expose and read out, which is the number that decides whether your light is bright enough, and the cross-web axis has to be sampled at the same 50 µm if you want square pixels.
| Belt speed | 50 mm/s |
| Along-motion resolution (per line) | 50 µm |
| Required line rate | 1 kHz |
| Line period (1 / line rate) | 1000 µs |
| Exposure cap for 1 px of smear | 1000 µs |
| Exposure cap for ½ px of smear | 500 µs |
| Exposure cap for ¼ px of smear | 250 µs |
| Lines per metre of product | 20000 |
Line rate is speed divided by resolution, not multiplied by it. Written with consistent units, line rate = belt speed ÷ resolution along the motion direction. In millimetres per second and millimetres per line the result is lines per second; divide by 1000 for kHz. Dividing micrometres by millimetres (a factor of 1000) is the single most common error in this calculation — it produces an answer three orders of magnitude out and is easy to spot because no camera on the market runs that fast.
Substitution: 50 mm/s ÷ 0.05 mm = 1000 lines/s = 1 kHz
Smear along the motion direction is speed × exposure time. To keep it under a chosen fraction of the 50 µm sampling interval, the exposure has to satisfy exposure ≤ fraction × 50 µm ÷ 50 mm/s. At one full pixel of smear that ceiling is 1000 µs; at half a pixel it is 500 µs. In practice you aim well below the ceiling and make up the light with a brighter bar, because a real line also has readout and jitter on top.
| Line rate | Travel per line | Lines per metre | ½ px exposure cap |
|---|---|---|---|
| 5 kHz | 10 µm | 100000 | 100 µs |
| 10 kHz | 5 µm | 200000 | 50 µs |
| 20 kHz | 2.5 µm | 400000 | 25 µs |
| 40 kHz | 1.2 µm | 800000 | 12.5 µs |
| 80 kHz | 0.6 µm | 1600000 | 6.2 µs |
Same belt speed, different camera setting. Doubling the line rate halves the resolution interval and halves the exposure budget — the two move together, which is why a faster line always costs light.
The line rate sets the sampling along the motion direction; the lens and camera pixel pitch set the sampling across it. For equal sampling in both directions the lens has to run at magnification = pixel pitch ÷ resolution. On this page that is 50 µm of along-motion sampling, so a 5 µm pitch camera needs 0.1X, a 10 µm pitch needs 0.2X and a 20 µm pitch needs 0.4X. Get this wrong and the part is stretched or squashed by the ratio of the two, even though each axis looks sharp on its own.
| Camera | Pixel pitch | Line length | Magnification for square pixels | Resulting FOV width | Inside PMS-LLC |
|---|---|---|---|---|---|
| 2K | 14 µm | 28.67 mm | ✓ 0.28X | 102.4 mm | yes |
| 4K | 7 µm | 28.67 mm | ✗ 0.14X | 204.8 mm | no |
| 4K | 10 µm | 40.96 mm | ✗ 0.2X | 204.8 mm | no |
| 8K | 7 µm | 57.34 mm | ✗ 0.14X | 409.6 mm | no |
| 8K | 10 µm | 81.92 mm | ✗ 0.2X | 409.6 mm | no |
| 16K | 5 µm | 81.92 mm | ✗ 0.1X | 819.2 mm | no |
Line length is pixel count × pitch and must stay inside the 82 mm image circle of the PMS-LLC series. The FOV width is simply pixel count × 50 µm, independent of which camera you choose — a larger sensor buys field, not resolution. At its widest setting the family images 82 mm ÷ 0.22X ≈ 373 mm, so a wider web is a multi-camera line, not a wider lens.
A cross in the last column is not a lens problem: the camera is over-resolved for this interval, so its pitch ÷ 50 µm falls below the 0.22X bottom of the band. The fix is a coarser-pitch camera with fewer pixels — no lens can create 0.1X on an 82 mm image circle without a field wider than the line can hold.
These are the published POMEAS line-scan lenses for a 16K line (82 mm image circle, M72 mount, F4.0–F16, under 0.002% distortion). Pick the model whose magnification band contains the figure in the table above.
| Model | Magnification | Working distance | Resolving power | Distortion | Mount |
|---|---|---|---|---|---|
| PMS-LLC026-14640M-M72 | 0.22X – 0.3X | 752 / 644 / 571 mm | 5 μm | <0.002% | M72 |
| PMS-LLC033-14640M-M72 | 0.3X – 0.37X | 482 - 571 mm | 5 μm | <0.002% | M72 |
| PMS-LLC05-14640M-M72 | 0.47X – 0.54X | 417 / 383 / 357 mm | 5 μm | <0.002% | M72 |
| PMS-LLC075-14640M-M72 | 0.7X – 0.85X | 304 / 286 / 267 mm | 5 μm | <0.002% | M72 |
| PMS-LLC1-14640M-M72 | 0.92X – 1.08X | 见规格书 | 5 μm | <0.002% | M72 |
| PMS-LLC133-14640M-M72 | 1.33X | 216 mm | 5 μm | <0.002% | M72 |
| PMS-LLC2-14640M-M72 | 1.75X – 2.25X | 179 - 690 mm | 5 μm | <0.002% | M72 |
| PMS-LLC3-14640M-M72 | 3X – 3.5X | 157 / 150 / 144 mm | 10 μm | <0.002% | M72 |
Values are as published on each product page. Where a model offers three settings the working distances are listed in the same order.
Three things break before the geometry does. First, light: at 1000 µs per line the bar has to deliver roughly 1× the energy per line that a 1 ms area-camera exposure would, so line-scan illumination is sized in W per metre, not in lumens. Second, the trigger: the line rate has to follow an encoder, not a timer, or belt-speed variation turns into resolution variation. Third, vibration and lens resolving power — a lens quoted at 5 µm cannot deliver a 50 µm sample with margin if the mount rings at the line frequency.
1 kHz. Divide the speed by the resolution interval: 50 mm/s ÷ 50 µm (= 0.05 mm) = 1000 lines/s = 1 kHz. The line period is the reciprocal, 1000 µs, and one metre of product is sampled 20000 times.
The hard ceiling is one line period, 1000 µs, which already allows a full pixel of smear. For half a pixel it is 500 µs and for a quarter pixel 250 µs. Treat the half-pixel figure as the working target for measurement and the full-pixel figure as the limit for presence detection.
It does not follow the line rate directly — it follows the resolution you end up with. Because this page fixes 50 µm along the motion direction, the cross-web axis needs the same 50 µm, and that means magnification = pixel pitch ÷ 50 µm. Halve the resolution target and the magnification doubles, so a line-rate change usually pulls a lens change with it.
Until the line rate leaves the camera specification. Holding 50 µm, a 100 kHz camera would reach 5000 mm/s. Beyond that you either accept a coarser resolution interval or split the web across two cameras so each runs at half the line rate.
Use an encoder whenever the part has to be measured or stitched without distortion. A fixed line rate assumes the belt speed never varies; a 2% speed error becomes a 2% length error in the image, which is larger than most of the tolerances this resolution was chosen for. Fixed rate is acceptable for presence and label checks only.
Yes. 1 kHz is a comfortable operating point for a line-scan camera, and at that speed most camera families still offer a second tap or a binning mode if you need more light.
The method behind these numbers, and the pages that continue the calculation: motion blur: exposure time against conveyor speed, magnification table: sensor width × field of view, inline cycle time and station count, real working distances by model, PMS-LLC05 16K line-scan lens. Working on a line that does not match any row here? Send the speed and the resolution you need: talk to an application engineer.
Back to the full line-scan timing table: speed × resolution → line rate
