LightingTroubleshooting

Imaging a small 1.2 mm bore: coaxial lighting that keeps focus sharp

Why direct light cannot enter a small bore, how coaxial lighting through the lens reaches the clips, and the light budget and exposure that sharpen the image.

LumaSpec Vision11 min read Share on LinkedIn

On this page 10 sections
  1. 01Read the symptom before touching anything
  2. 02Cause 1: direct light cannot enter the hole
  3. 03Cause 2: the coaxial light budget is smaller than it looks
  4. 04Cause 3: a stopped-down iris blurs the whole image
  5. 05Cause 4: focus plane, and the depth you can see at all
  6. 06Set exposure and strobe to the motion limit
  7. 07The silhouette route: light it from behind
  8. 08Which fix fits your case
  9. 09Prevent it at design time
  10. 10Questions

The clips are in focus, yet the saved image is dim and smeared. That combination has one root cause: a 1.2 mm bore starves the sensor of light, and the gain, exposure and iris settings you use to recover each add blur of their own. This troubleshooting guide walks the causes in the order they fail in real stations, gives the arithmetic to check each one — view cone, beamsplitter and clipping losses, diffraction, exposure against line speed — and ends with the fix that fits each case: coaxial lighting through a telecentric port, a silhouette backlight, or a change of optics.

Read the symptom before touching anything

With focus confirmed on a test target, residual blur comes from three places: sensor noise amplified by gain, motion blur from a long exposure, and diffraction from a stopped-down iris. All three trace back to one shortage — too few photons from the clips — so fix the light budget and the blur usually leaves with it.

Start with the histogram. A peak bunched near black means the light never reached the clips. A well-exposed but smeared frame points at motion or diffraction instead. Work through the causes below in order; each carries a test and a fix.

Cause 1: direct light cannot enter the hole

A ring or bar light at the mouth of a 1.2 mm bore illuminates the top face of the terminal, not the spring clips inside. Light only helps if it enters the mouth, scatters off a clip and returns through the same opening, and both trips are confined to one narrow cone.

The half-angle of that cone follows from the bore diameter D and the feature depth d:

θ = arctan(D / (2d))

Take D = 1.2 mm and clips 3 mm inside the mouth (measure your own depth and rescale everything that follows). Then θ = arctan(0.6 / 3) = 11.3°, equivalent to an aperture of about f/2.5. A ring light working at 30-60° incidence delivers effectively nothing down that cone.

Test: capture one frame with the direct light on and one with it off at the same exposure. If the histogram barely moves, the photons never entered.

Fix: through-lens coaxial illumination, where light and image share the axis. The contrast mechanism is explained in in-line illumination in machine vision, and the geometry logic in lighting geometry for image contrast. For other recessed metal features, the engraved-character coaxial setup shows the same pattern at work.

Cause 2: the coaxial light budget is smaller than it looks

Coaxial lighting solves the geometry, then loses most of the light. Three losses stack. The beamsplitter passes about half the light on the way in and half of the return: a factor of 0.25 end to end. The illuminator's exit beam is wide — assume 18 mm across at the port and rescale with your lens — while the mouth admits only a 1.2 mm column, a fraction of (1.2 / 18)² = 0.0044. And of the light a matte metal clip scatters, only the share leaving within the lens numerical aperture returns to the sensor: about NA², or 0.01 at NA 0.10.

Worked example, assuming 50% diffuse reflectance and object-side NA 0.10:

Step Flat part under the same coaxial Clip 3 mm inside the bore
Beamsplitter, light in x 0.5 x 0.5
Beam captured by the bore mouth x 1 x 0.0044
Surface reflectance x 0.5 x 0.5
Return within NA 0.10 x 0.01 x 0.01
Beamsplitter, light out x 0.5 x 0.5
Light reaching the sensor 0.00125 0.0000055

The clip receives about 230 times less light than the flat part — two orders of magnitude. That is why a coaxial light that looks bright on a flat sample goes dark inside a bore. The flat-part case in absolute units is worked through in how many lumens for machine vision lighting.

Fix: buy the margin back with current and time. Feed the port from a strobed spot or collector light:

Spot lightsThird generation spot light, power 4 W (LS-SPT-0133)LS-SPT-0133
Size
Under 50 mm
Voltage
24 V
$259.00View part
Spot lightsSpot collector light, 94.2 x 24 mm, 3 W (LS-SPT-0142)LS-SPT-0142
Size
50-100 mm
$349.00View part
Spot lights6 mm High luminance spot light, power 5 W (LS-SPT-0175)LS-SPT-0175
Size
Under 50 mm
$229.00View part

Drive the pulse at full strobe current — pulsed ratings typically allow several times the continuous current, worth 5-10x over always-on drive. Keep the iris fully open. Take the longest exposure the line speed allows, which the next section sizes. If the budget still falls short, lower the magnification: a larger object pixel tolerates more motion per pixel, so the exposure limit lengthens.

Cause 3: a stopped-down iris blurs the whole image

Chasing depth of field inside the hole, it is tempting to stop down to f/11 or beyond. Diffraction then blurs everything. The Airy disk diameter on the object side is:

Airy diameter = 1.22 x λ / NA

At λ = 0.55 µm and object-side NA 0.10 (roughly image-side f/4, typical for 1-2x telecentrics): 1.22 x 0.55 / 0.10 = 6.7 µm, about 2.7 of the 2.5 µm object pixels. Stop down to NA 0.05 and it doubles to 13.4 µm — five and a half pixels of blur — while the light halves again. The Airy disk and diffraction limit guide works this pixel matching through in detail.

Test: image a static clip wide open and stopped down at equal exposure. If the wide-open frame is crisper, diffraction was the blur.

Fix: iris fully open, focus at the clip plane, accept the resulting depth of field. When part height varies, a telecentric lens keeps defocus blur symmetric so sizes stay repeatable — see telecentric symmetric blur. For a standard entocentric lens, size the stop first with the depth of field calculator.

Perspective with a standard lens compared with a telecentric lens Left: the same part at two distances appears at two sizes through a standard lens. Right: a telecentric lens accepts only rays parallel to its axis, so the part appears the same size at both distances.Standard lensfarnearsensorThe same part looks bigger when nearer:its measured size depends on distanceTelecentric lensOnly rays parallel to the axis enter:the same size at any distance
Perspective with a standard lens compared with a telecentric lens. Left: the same part at two distances appears at two sizes through a standard lens. Right: a telecentric lens accepts only rays parallel to its axis, so the part appears the same size at both distances.

Cause 4: focus plane, and the depth you can see at all

Two geometry checks come before any purchase.

Focus on the clips, not on the mouth. Sweep focus in 0.2 mm steps on a static part; the sharpest frame sets the working distance.

Then confirm the clips are visible at all. A telecentric lens sees along rays within its numerical aperture, so at depth d the view cone has narrowed to a spot of 2 x d x tan(θ_NA). Wall features vanish once d exceeds:

d_max = (D / 2) / tan θ_NA

With NA 0.10 (θ_NA = 5.7°) and D = 1.2 mm: d_max = 0.6 / 0.1 = 6.0 mm. Contrast collapses earlier, because wall returns arrive at grazing angles, so treat 3-4 mm as the useful limit for clips in a 1.2 mm bore. Deeper than that, no lighting fix helps: change the optics — a 360° inspection lens images the wall from inside the bore — or inspect before assembly.

Set exposure and strobe to the motion limit

With the light fixed, take every microsecond the motion allows:

t_max = p_obj / v

The object-side pixel is p_obj = 3.45 µm / 1.36 = 2.5 µm on a 5 MP 2/3" sensor. The exposure that keeps motion blur within one pixel:

Line speed Exposure limit
10 mm/s 250 µs
50 mm/s 50 µs
100 mm/s 25 µs

At 50 mm/s, 50 µs of full-current strobe usually beats 2 ms of continuous light at a safe current level. Two details decide whether it works. Fire the pulse a few microseconds early, so the LED reaches full output during the exposure — rise-time behaviour is covered in LED strobe rise time. And use a light with an integrated strobe driver so the camera's strobe output triggers it directly; what that changes is covered in integrated strobe drivers in machine vision lights.

If the part sits clamped during capture, exposure can run to tens of milliseconds and the whole budget problem shrinks. Confirm that with a static test before buying more light.

The silhouette route: light it from behind

If the rear insulator of the terminal is translucent, skip coaxial entirely. Put a spot light behind the part: the glowing plastic becomes the background and the clips silhouette dark against it. No beamsplitter, no bore clipping on the imaging side, and contrast comes from extinction, which forgives surface finish.

Test: back-light a sample and check that both clips darken cleanly against the glow at every rotation angle the part can present.

Limit: a silhouette shows presence, position and bending. It cannot show surface detail such as scratches or plating flaws.

Blue suits translucent plastic: it scatters strongly and sits near the silicon sensor's response peak. The colour choice is worked through in which LED colour gives the best contrast. For a broad glow behind a small part:

Spot lightsSpot light, 49.9 x 66.6 mm lit area, 455 nm Blue, 32° lens (LS-SPT-0024)LS-SPT-0024
Size
Under 50 mm
Voltage
24 V
$279.00 · quotedView and quoteBuild a system with it

A 455 nm spot with a 32° beam lens spreads an even pool behind the fixture. Place it 50-100 mm back and judge uniformity on the live image.

Which fix fits your case

Case A — surface detail on clips within about 3-4 mm of the mouth. Object-side telecentric with a coaxial port, strobed spot, iris open, exposure at the motion limit:

Telecentric lensesObject-side telecentric lens, 1.36x, 73 mm WD, 2/3", coaxial light port (LS-TCL-0811)LS-TCL-0811
Sensor format
2/3"
Mount
C-mount
Image circle
8-11 mm
Type
Object-side telecentric
$789.00View partBuild a system with it

The LS-TCL-0811 gives 1.36x at a 73 mm working distance on a 2/3" C-mount camera. On a 5 MP 2/3" sensor with 3.45 µm pixels, that is an object pixel of 3.45 / 1.36 = 2.5 µm and a field of 8.8 / 1.36 = 6.5 mm — enough to follow clip bends, cracks and plating damage inside the mouth. Match the feeder on the light finder and the optic on the lens finder, or start at the light configurator.

Case B — presence and orientation only. Less magnification buys light and depth:

Telecentric lenses0.5x, WD 110 mm, 1", C-mount telecentric lens with coaxial light (LS-TCL-2194)LS-TCL-2194
Sensor format
1"
Mount
C-mount
Image circle
11-16 mm
Type
Object-side telecentric
$1,539.00View partBuild a system with it

The LS-TCL-2194 carries its coaxial light built in: 0.5x at a 110 mm working distance for a 1" C-mount camera. At 0.5x the object pixel grows to 6.9 µm and the field to about 18 mm, so the 50 mm/s exposure limit relaxes from 50 µs to 138 µs, and the longer standoff clears fixture hardware.

Case C — translucent rear insulator. The silhouette route above, with no coaxial hardware at all.

Case D — clips deeper than about 6 mm. Change the optics, not the light, as Cause 4 shows.

Case E — no telecentric lens in the station. A coaxial light with its own beamsplitter mounts in front of a standard C-mount lens:

Coaxial lights60 x 60 mm Coaxial light, 4.4 W red / 7.2 W white (LS-CXL-0027)LS-CXL-0027
Size
50-100 mm
Voltage
24 V
$269.00View partBuild a system with it

The LS-CXL-0027 is a 60 x 60 mm coaxial light, 4.4 W in red or 7.2 W in white. Red sits near the sensor's response peak and keeps the Airy disk small. It serves presence and orientation checks at close range; size the focal length for your field with the calculator below.

Lens focal length, field of view and working distance
Units
Solve for
Focal length
– mm
Field of view
– mm
Working distance
– mm
Magnification
– x
Angle of view
– °
Pixel on the part
– mm

Thin-lens model: m = sensor width / field of view, f = WD × m / (1 + m). Accurate to a few percent above about 150 mm; check the lens's minimum object distance.

Find lenses with this focal length Open the full calculator

One line to decide by: shallow clips needing detail — Case A; presence only — Case B or C; deep features — Case D; existing station — Case E. Spend on the lighting fix before the algorithm: contrast is set by glass and photons, and no filter recovers photons that never arrived.

Prevent it at design time

Fix three numbers before the mechanics are cut: the measured clip depth d, the line speed v, and the smallest feature that must resolve, which sets the object pixel. Check d against d_max = (D / 2) / tan θ_NA with a margin of about 1.5x — for a 1.2 mm bore at NA 0.10 that means clips within roughly 4 mm. Check t_max = p_obj / v against the strobe's pulse rating, not its continuous rating. Order the optic and its feeder together: the coaxial lights, telecentric lenses and spot lights pages show what pairs, and a focus target at clip depth on the fixture makes changeover a two-minute job. If a case stays open, send the bore drawing and a sample image through the contact page, or browse the surrounding guides on the lighting topic page.

Questions

How deep can a camera see inside a 1.2 mm bore?

The lens numerical aperture sets it, not the light. With NA 0.10 the view cone stays within 5.7° of the axis, so wall detail fades out around d = 0.6 / tan(5.7°) = 6 mm, and usable contrast ends nearer 3-4 mm because returns arrive at grazing angles. Beyond that, use a 360° bore lens or inspect before assembly.

Why is the image dark even with a telecentric lens and coaxial light?

Two fixed losses stack. The beamsplitter passes roughly half the light in and half the return, so at best a quarter of a direct light's throughput survives, and telecentric designs are slower than standard lenses. Plan for about four times less light than a flat, directly lit part, then win it back with strobe current, an open iris and the longest exposure the line allows.

Can I inspect inside the bore without a coaxial port?

Yes, in two situations. If the rear insulator is translucent, backlight it with a spot light and read the clips as silhouettes — no beamsplitter loss at all. Or mount a coaxial light with its own beamsplitter in front of a standard C-mount lens for presence checks, accepting a larger housing around a 1.2 mm part.

What exposure time avoids motion blur on a moving line?

Divide the object-side pixel by the part speed. At 1.36x with 3.45 µm sensor pixels the object pixel is 2.5 µm, so 50 mm/s allows about 50 µs, 10 mm/s allows 250 µs and 100 mm/s only 25 µs. Strobe the light inside that window and drive the pulse at the full rating, typically several times the continuous current.

Should the coaxial light be infrared?

Usually not. Silicon sensors lose sensitivity toward the near infrared, and the longer wavelength enlarges the Airy disk, softening detail that is already diffraction limited. Red around 630-660 nm is the sensible default; choose infrared only to defeat ambient light or when the part fluoresces under visible wavelengths. Test both on real parts before ordering.