Adhesive presence under a matched housing, a sealant bead on a dark enclosure, a tamper seal printed in security ink, a film of coolant on a machined face: these are the jobs visible light cannot do and UV-A can. The catch is that UV-A spans 315-400 nm, and where you sit inside that range moves the signal your camera sees by factors, not percentages. This guide gives you the band to specify per task, the physics that sets it, a worked lens-and-exposure example, and the checks to run before ordering.
The decision at a glance
If you want the answer before the physics: specify 365 nm for fluorescence contrast — adhesives, inks, sealants and contamination; 375-385 nm when the tracer is forgiving and the line is fast; 395-405 nm when the job is reflectivity and mark contrast rather than glow. Those three lines cover most UV-A stations. The table adds what each band costs, and the sections below show where the numbers come from.
| Band | Specify it for | Gain | Cost |
|---|---|---|---|
| 365 nm | Adhesives, security inks, sealant, contamination | Strongest fluorescence per radiated watt; widest filter margin | Weakest LED output; longest exposure |
| 375-385 nm | General fluorescence on fast lines | Three to five times the radiated output of the short-wave band | Slightly weaker excitation |
| 395 nm | Strong emitters; dark surfaces | Efficient, bright, still UV-A | Weak fluorescence on many materials |
| 405 nm | Marks, codes and texture on dark or glossy parts | Highest sensor response and LED output; quiet background | Almost no fluorescence; visible violet tail |
Throughout the rest of this guide, "the short-wave band" means the 365 nm line and "the violet band" means the two longer lines, so each figure appears once and the argument stays readable.
For the wider geometry question — ring versus bar versus dark field — see the lighting selection by inspection task guide. This guide stays on the wavelength axis.
The two curves that set the band
Two curves move in opposite directions as wavelength climbs across the UV-A range: the LED's radiated output rises while the fluorescence it can drive falls. Everything else follows from that crossing.
Photon energy: the short-wave end carries 3.40 eV per photon, the violet end 3.06 eV, with the middle line at 3.14 eV. A tracer must absorb the excitation photon before it can emit, so each step toward the visible edge weakens excitation — gently for most organic tracers, sharply past 400 nm.
Fluorescence is emitted at a longer wavelength than the excitation. Take a tracer emitting at 450 nm: excited at the short-wave end the Stokes shift is 85 nm; excited at the violet end it is 45 nm. A bigger shift gives a long-pass filter more running room below the emission, and the background stays darker.
The camera is less of a constraint than most people expect. Typical machine-vision CMOS sensors convert roughly 10-30% of arriving short-wave UV photons into electrons, against 50-60% near 550 nm, because sensor windows and microlenses absorb short wavelengths. With a long-pass filter fitted, reflected UV never forms the image at all: the picture is built from the visible emission, where sensor response peaks. The QE gap at the short-wave end is largely irrelevant once the filter is right.
The LED is the real constraint. Radiant efficiency — optical watts per electrical watt — falls steeply toward short wavelength: a typical short-wave UV head turns a few percent of its input into UV-A, while violet-band heads reach tens of percent. At equal drive, expect roughly five to ten times less radiated power from the short-wave head. Band selection is the trade between that deficit and the fluorescence gain.
365 nm: maximum fluorescence contrast
This band excites the widest set of UV-A-reactive materials and returns the strongest emission per radiated watt — commonly about twice the signal of the next line up on adhesive tracers, and more on weak emitters. That margin decides borderline calls: a glue skip on a dark housing, a tamper seal in security ink, a coolant film on a machined face, residue on a cleaned surface. The 85 nm Stokes shift also gives the lens filter its widest separation between excitation and emission.
The price is photons. The same size head radiates five to ten times less than a violet-band equivalent, so exposure runs longer or the LED overdrives. Alignment happens on the camera, because the output is almost invisible to the eye.
For filter stacking, ambient control and camera settings in detail, see the fluorescence excitation and filters guide. If you are still deciding between imaging reflected UV and imaging fluorescence, the reflected UV versus UV fluorescence guide separates the two.
375-385 nm: the working middle
Around this band the LED typically radiates three to five times the power of a short-wave head of the same size, while most UV-A tracers still respond, a little weaker. When line speed caps exposure, that trade usually wins: if the short-wave variant needs 50 ms and the line allows 15 ms, this head closes the gap with the ratio arithmetic shown in the worked example below. Keep the long-pass filter fitted — the excitation is still ultraviolet and still must be stripped from the image.
When exposure must fall to milliseconds, add strobing: the overdrive strobing technique guide shows how pulsed drive recovers the missing photons.
395 and 405 nm: reflectivity and a quiet background
The violet band keeps a foot in both camps: efficient to drive, and still enough UV-A to excite strong emitters such as heavily loaded adhesive tracers. The longest of the two gives up most fluorescence and buys the brightest heads, the highest sensor response and a background that stays quiet even beside fluorescent inks. Choose the first when you want weak fluorescence plus reflectivity; choose the second when you want reflectivity only.
On dark plastic, matte rubber or glossy laminates that is exactly what mark, code and texture checks need. A violet band-pass lights a laser-etched code as bright-on-dark contrast while the surrounding material stays black, and fluorescent packaging nearby stays unexcited. Reflection geometry still rules the result — the bright field versus dark field guide pairs surface type with illumination angle.
A violet spot head suits the small fields these tasks usually run. The LS-SPT-0068 is a 405 nm spot light, 73.7 x 9.4 mm, powered from 24 V with a PNP trigger input.
- Size
- 50-100 mm
- Voltage
- 24 V
- Wavelength
- 405 nm
Worked example: adhesive bead on a dark housing
Task: confirm a 1 mm fluorescent adhesive bead around a dark housing window. Field of view 60 mm; camera 2448 × 2048 with 3.45 µm pixels; stand-off around 200 mm. Band: 365 nm, because the bead is the only feature allowed to glow and the background must stay black.
- Sensor width: 2448 px × 3.45 µm = 8.45 mm.
- Object scale: 60 mm ÷ 2448 px = 24.5 µm per pixel, so the 1 mm bead spans 1 ÷ 0.0245 = 41 pixels.
- Magnification: m = 8.45 ÷ 60 = 0.141.
- Working distance with a 25 mm lens: WD = 25 × 1.141 ÷ 0.141 = 202 mm.
- Depth of view at f/4, circle of confusion two pixels (6.9 µm): DOF ≈ 2 × 4 × 6.9 × 1.141 ÷ 0.141² = 3.2 mm, or ±1.6 mm about best focus — ample for a flush bead.
If your numbers differ, the lens focal length calculator runs the same arithmetic, and the field-of-view picker below filters lenses to the result.
Now the band trade in one calculation. Assume this tracer emits about twice as strongly per radiated watt at the short-wave end as in the violet band, and a short-wave head radiates about one fifth of a violet head at the same drive — typical ratios for UV-A-reactive adhesives. Net emission at 365 nm = 2 × 1/5 = 0.4 times the violet result. If a violet prototype passes at 20 ms, budget 20 ÷ 0.4 = 50 ms at the short-wave end, or overdrive the LED instead of stretching the exposure. Measure both ratios on a sample part before freezing the design; they move the answer, not the method.
Filtering, exposure and motion
Fit a long-pass filter with its cut-on near 430 nm: it passes the blue-green emission and blocks the excitation. From the short-wave band the filter sits 65 nm above the excitation line; from the violet end the margin shrinks to 25 nm, which is why fluorescence set-ups there are fragile. Run fixed manual exposure and judge the result on the histogram, never by eye.
Motion caps exposure: t_max ≈ allowed blur ÷ part speed. Twenty pixels of tolerated blur at 24.5 µm per pixel is 490 µm; at a 100 mm/s line that is 490 ÷ 100,000 = 4.9 ms of exposure. If the chosen band cannot deliver enough fluorescence in 4.9 ms, move up to the middle of the range and recompute.
Stations that also read print sequence a white channel with the UV channel on one camera. The multi wavelength light sequencing guide and the camera emitter pattern synchronization how-to cover the timing.
Which band for which job
| Task | Band | Why |
|---|---|---|
| Adhesive or sealant presence | 365 nm | Strongest excitation, widest filter margin |
| Tamper seals and security inks | 365 nm | Security inks are formulated for UV-A excitation |
| Oil or coolant contamination | 365 nm | Petroleum films fluoresce under UV-A |
| Fast line, forgiving tracer | 385 nm | Three to five times the radiated power buys back the exposure |
| Code, mark or texture on dark parts | Violet band with a band-pass | Bright-on-dark reflectivity, minimal glow |
| Print on brightened packaging | No UV-A | Paper brighteners flood the image with glow |
When two bands both pass the bench test, take the longer one: it drives easier, runs cooler and tolerates faster lines. Drop to the short-wave band only when the contrast at the middle wavelengths falls short of the pass/fail margin.
Checks before you order
- Excite a sample part with one head from each end of the range and photograph both through a 430 nm long-pass filter. Choose by emission contrast, not by which light looks brighter. The reflected UV contrast test guide walks this test.
- Read your sensor's QE curve at the emission wavelength, not at the excitation wavelength.
- Confirm the mechanics: light, filter, shroud and the 202 mm-class stand-off your FOV arithmetic demands.
- Match the head shape to the field: bar lights for wide fields, spot lights for small fields and coaxial paths, and the special lights range for structured, sequencing and custom heads.
- Budget the exposure from the ratio calculation, then confirm the head holds it at your trigger rate.
Bar lights carry most UV-A duty on production lines: the range runs from under 50 mm to over 1,200 mm lit length, across the whole band selection in this guide.
Angled or low-angle light, any length
1,403 models · from $89.00When the usual answer is wrong
Optical brighteners in paper, many plastics and some coatings fluoresce so strongly that the whole surface glows under UV-A and contrast collapses. Treat that as a stop sign, not a gain problem: move to violet-band reflectivity, or suppress surface glare with cross polarization under visible light.
Many leak-tracing dyes are specified for blue excitation around 450-470 nm. A blue spot light such as the LS-SPT-0024, which pairs a 455 nm emitter with a 32° lens, then outperforms any UV-A source.
Bright ambient light passes a long-pass filter and lifts the black level. Shroud the cell first; where that is impossible, tighten to a band-pass centred on the emission.
Below 315 nm the physics changes: standard silicon cameras barely respond and the safety regime hardens sharply. Germicidal and biofilm work in the UV-C band is disinfection, not a machine-vision lighting choice.
Next step
Bring four numbers to the decision: field of view, stand-off, cycle time and the emission colour of your tracer. The lighting configurator narrows heads by wavelength, size and drive, the lights finder lists the range by the same filters, or send us the inspection details and we will size band, head and exposure together. More lighting material sits on the lighting topic page.
Questions
Does the camera see the UV itself or only the fluorescence?
Only the fluorescence should form your feature. Standard silicon CMOS sensors respond weakly around 365 nm and strongly in the blue-green where most tracers emit, so the emitted light makes the image while a lens long-pass filter strips the reflected excitation. The result is a dark background with the glowing feature as the only bright object.
Which filter do I fit for a 365 nm fluorescence check?
A long-pass filter with its cut-on around 430 nm: it passes blue-green fluorescence and blocks the excitation. If your tracer emits nearer the violet, move to a band-pass matched to the emission peak. Fit the filter to the lens, fix it mechanically, and repeat the contrast check after any lens change.
Is a UV-A inspection light safe to work around?
UV-A is the least energetic band used in vision, but close-range LED heads still expose eyes and skin, and the output is nearly invisible so you get no blink warning. Wear UV-A blocking goggles when a head runs outside a shrouded cell, shield the station, and never look into the light. UV-B and UV-C demand far stricter controls.
Why does the violet band look brighter to my eye than the short-wave band?
The violet band sits at the edge of the visible, so the eye sees a strong purple glow; the short-wave band is almost invisible. The dim-looking head can still deliver more usable excitation per radiated watt. Judge both lights from the camera histogram under the same exposure, never by apparent brightness.
Can one station do a UV seal check and a normal label check?
Use a multi-channel light with a UV channel and a white channel, trigger them in sequence, and capture two images per part: the UV frame exposes the fluorescent seal, the white frame reads print and label position. One camera and one controller handle both as long as the cycle time allows two exposures.