Reflected UV vs UV fluorescence imaging: which one you need

Two imaging techniques share the same ultraviolet lamp and yet need different cameras, lenses and filters. Reflected UV imaging photographs the UV that bounces off the part; fluorescence imaging photographs the visible light the part re-emits after absorbing UV. Pick the wrong one and the hardware order is wrong too: a UV-transmitting lens is useless if the filter stack passes the wrong band, and a standard camera rarely returns usable UV-A contrast. This guide separates the two, gives the band structure you should quote, shows which defects each technique finds, and works the lens calculation both share, so you can commit to a wavelength, a filter position and a focal length before you order.

What is the difference between reflected UV and UV fluorescence imaging?

Reflected UV imaging returns the wavelength you project — 365 nm from the LED lands on the sensor as 365 nm, unshifted — while fluorescence uses the UV only as a pump: the surface absorbs it and re-emits a longer wavelength, typically blue near 450 nm.

The reflected path never changes wavelength, so the sensor works at the LED's own band and every optical part must pass UV. The fluorescence path converts: the absorbed photon raises an electron, the electron relaxes, and the material emits at a longer wavelength — the Stokes shift — with the difference lost as heat. Light energy out therefore always stays below light energy in, which is why fluorescence images run dim and exposure planning matters.

That physical split decides the hardware. Reflected UV keeps the whole chain inside the ultraviolet, where silicon is insensitive and glass is lossy. Fluorescence moves the signal into the visible and near-infrared, where standard cameras and lenses are at their best. If the defect changes how the surface reflects or absorbs UV — a scratch, an oil film — you are in the first camp. If it fluoresces, or you can treat it with a fluorescent penetrant or dye, you are in the second.

A monochromatic source sharpens reflected UV work further. With all light at one narrow wavelength, the optics need no chromatic correction across a spectrum, and the shorter the wavelength, the finer the detail the system can separate. Narrowband UV LEDs turned UV imaging from a lamp specialist's field into an affordable production tool.

Which UV band should you inspect in?

Use UV-A, 315-400 nm, unless you have a specific reason not to: LEDs in this band are bright, easy to filter, and ordinary lens glass still transmits them. The CIE bands run UV-C 100-280 nm, UV-B 280-315 nm and UV-A 315-400 nm, with visible light starting at 400 nm.

Below UV-A the glass changes. Standard optical glasses absorb below about 290 nm, so the UV-B region already demands UV-transmitting glass, and UV-C work needs fused silica, fused quartz or calcium fluoride optics. Below roughly 180-190 nm the air itself absorbs the radiation — the vacuum-UV region — and imaging only works in a vacuum or a nitrogen purge.

For production inspection this collapses to a default: UV-A for scatter, absorption and penetrant work, with 395-405 nm violet LEDs as a brighter option that brings more visible scatter to filter out. UV-B and deep UV stay specialist: mercury-line imaging near 254 nm for organic contamination, and photolithography. The wavelength ranges of machine vision light put these bands next to the visible and infrared alternatives.

Glass choice shows up directly in lens catalogues. The LS-SWL-0280 is specified 190-700 nm on C-mount for 4/3" sensors, with a 16-22 mm image circle: it reaches through the UV-B region and still passes visible emission, which no standard glass lens does.

Why does a standard camera and lens fail in reflected UV?

Because silicon's response falls away below 400 nm and the glass in ordinary lenses and sensor windows absorbs what remains: a visible-grade lens cuts off hard below about 290 nm, and a colour mosaic removes UV contrast on top of that.

Each part of a visible-light chain blocks UV in turn. Silicon responds weakly at UV-A wavelengths and many sensor cover windows attenuate further. Catalogue lens glasses stop transmitting as the wavelength drops. A colour camera's Bayer mosaic absorbs UV without adding useful colour, because a single narrow band carries no colour information at all. Work monochrome, with a camera whose spectral response is documented at your wavelength.

Then match the lens to the band rather than to the sensor format alone: the transmission range is the spec that decides. An inspection at the LED's wavelength needs a lens that passes that wavelength cleanly, and the UV, SWIR and thermal lens category filters by type so you can shortlist candidates before comparing details. The SWIR and UV lens basics guide covers what else changes optically once you leave the visible band.

Contrast needs the filter even when brightness looks fine. Illuminate with UV while the lens admits visible light and the image records mostly visible reflection: substrate and background wash together. A UV bandpass on the lens, passing the illumination band and blocking everything longer, is what pulls a defect out of a plastic substrate — the difference between a UV photo and a UV measurement.

What does a UV fluorescence setup need instead?

A camera that sees the emission, a lens that passes it and a filter that blocks the excitation: most silicon cameras already respond across 400-1000 nm, so fluorescence needs far less specialised hardware than reflected UV.

The excitation comes from the same UV-A LEDs, listed under the UV colour option on the special lights category. The camera is ordinary: monochrome preferred, colour when emission colour itself carries information. The lens is ordinary too — any design corrected for visible light images a blue or green emission well.

The filter is where fluorescence is won. Surface-reflected excitation is orders of magnitude brighter than the weak emission, so a long-pass or bandpass on the lens that rejects the excitation band is not optional. Exposure needs respect as well, because the emission carries only a fraction of the pumped-in energy: plan for strobed overdrive, a wide aperture or a longer integration. The strobe overdrive and duty cycle guide covers the current and timing side, and the fluorescence lighting guide works through excitation-and-filter pairing in detail.

Most industrial emissions land between 430 and 700 nm, inside any UV lens's transmission. Some materials and dyes emit in the near-infrared instead, and those need a lens whose range runs well past 700 nm while still passing visible light — the 400-1700 nm window of an SWIR-range design. The LS-SWL-0246 covers that band at F1.8 on C-mount for 1/2" sensors.

How does the filter change between the two techniques?

The filter positions are opposite: reflected UV puts a UV bandpass on the lens that passes the illumination band and blocks everything above roughly 400 nm, while fluorescence puts a filter that blocks the excitation and passes only the emission above it.

For reflected UV, the bandpass kills daylight and factory lighting, rejects the visible tail every LED leaks, and stops the part's own fluorescence from contaminating a reflection measurement. For fluorescence, the lens filter does the reverse job, and its crossover must sit inside the Stokes gap between excitation and emission. With 365 nm excitation and a blue emission near 450 nm, the gap is 450 − 365 = 85 nm wide, and a long-pass edge around 420-430 nm sits comfortably inside it. When the shift shrinks to a few tens of nanometres, narrow bandpasses on both the light and the lens take over. The same width-against-contrast trade the optical filters guide works through for visible bands applies here unchanged.

Budget exposure for the filter itself. A bandpass passes less than its peak rating suggests once coating, angle of arrival and bandwidth are real, so read the transmission curve and put the loss into the illumination margin rather than discovering it on the line.

Which defects does each technique actually find?

Reflected UV finds defects that change how a surface reflects or absorbs UV — scratches and digs on polished parts, oil and grease films, paint repairs — while fluorescence finds whatever can fluoresce: penetrant in cracks, adhesive beads, coatings, coolants and organic contamination.

Scratches are the classic reflected-UV case, because short wavelengths scatter strongly from fine ridges that longer wavelengths ride over. Darkfield geometry turns that scattering into signal: light arrives at a shallow angle, a perfect surface reflects it away from the lens, and every dig and scratch lights up bright on dark.

Absorption gives reflected UV its second family of targets. Organic materials take up UV strongly, so oil and grease traces show against clean metal — most visibly in the deep UV — and petroleum products read differently by type, which helps identify a residue's origin. Repainted or repaired regions that look identical in white light can separate under UV. The reflected UV contrast test guide shows how to score that contrast on your own parts before buying anything.

Fluorescence owns the treated-defect world. Fluorescent penetrant remains the standard crack-finder; adhesive and gasket beads, sealants, coatings and coolant residue all fluoresce; leak paths show as bright traces. Where the emission escapes but visible surface scatter does not, fluorescence also reads content that white light hides.

How do you size the lens for either technique?

The geometry is identical for both, so one calculation covers them: m = sensor width / FOV and f = WD × m / (1 + m); at a 200 mm working distance and 60 mm field on a 2/3" sensor it lands on a 25 mm lens.

Worked example: a 2/3" sensor is 8.8 mm wide, and the part needs a 60 mm field at a 200 mm working distance. Magnification m = 8.8 / 60 = 0.147. Focal length f = 200 mm × 0.147 / (1 + 0.147) = 25.6 mm, so the nearest standard focal length is 25 mm. The LS-SWL-0291 fills that slot: 25 mm focal length, 2/3" sensor format, 8-11 mm image circle, F2.0, C-mount, and a 240-1000 nm transmission range that carries the UV-A inspection and visible emission imaging on the same barrel.

One optical difference remains between the techniques. A single narrow wavelength needs chromatic correction only there, which is why monochromatic reflected UV images stay sharp even on simple optics. A fluorescence emission spreads over tens of nanometres, so ordinary achromatic correction across the visible does that work instead. Either way, focus at the band you image: the focal plane in the UV-A band does not sit exactly where the visible focus sits on the same lens, so refocus after changing bands.

When is the obvious UV answer wrong?

Skip UV whenever visible or near-infrared contrast already separates the defect: a coloured LED with a bandpass, a coating that shows under red, or a plain white-light check all cost less hardware than UV-A optics, on cameras you already own.

Three cases catch teams out. First, you need less ultraviolet than you think: before quoting deep-UV optics, reshoot the scratch under UV-A darkfield, because shorter wavelengths scatter harder but UV-A often already separates the flaw. Second, the part fluoresces on its own, and that background glow drowns a reflected-UV measurement; image the emission and reject the excitation instead, or move the excitation band. Third, the cell cannot be dark: fluorescence competes with plant lighting and wants shrouding or a strobed exposure, while a reflected-UV bandpass holds its contrast with little shrouding at all. When visible light is still in play, the LED colour contrast guide is the first experiment worth running.

What do you check before ordering?

Five checks decide the bill of materials: the wavelength that gives contrast on your parts, the camera's response at that wavelength, the lens transmission across it, the filter edge positions, and the light's output and drive.

• Contrast at the wavelength: handheld UV-A and violet sources at the planned geometry on sample parts, before any purchase order.
• Camera: monochrome for either technique unless emission colour carries data, with a spectral response curve covering your band.
• Lens transmission: the excitation band for reflected UV, the emission band for fluorescence, and an image circle at or above the sensor diagonal.
• Filter edges: a UV bandpass on the lens for reflected work, the fluorescence edge inside the Stokes gap, and the on-band transmission curve in hand to budget exposure.
• Light and drive: a UV-A LED, or a violet 395-405 nm one, driven by a controller that can strobe; the light controller cable adapters guide pairs connectors with current, and the lighting accessories category holds the extension cables and mounting parts.

When the wavelength is decided, the lens finder filters the UV lens range by type, format and mount, and the light configurator pairs an excitation LED with a matching controller. If the two techniques still tie on your parts, the contact page routes a sample review to the applications team.