Why does the spectral band matter so much?

Because a photoinitiator absorbs over a narrow range and a radiometer measures over a defined band, and if the two do not overlap the dose figure is unrelated to the cure. A material initiated at 254 nm is cured by UVC; a UVA radiometer reading 1500 mJ/cm² next to it says nothing about whether that happened. This is the most common way a process that measures correctly still fails, and it is why every dose figure needs its band stated alongside it.

Is 100 mW/cm² for 12 seconds the same as 1200 mW/cm² for 1 second?

The dose is identical and the cure is not. Through-cure depends on total dose, so the bulk sees roughly the same thing. Surface cure in air depends on peak irradiance, because oxygen diffusing into the top few microns quenches radicals and only a high generation rate outruns it. The low-intensity long exposure therefore tends to leave a tacky surface over a properly cured bulk: a result that confuses people precisely because the dose was right.

What dose margin should I run?

Enough to cover the lamp's decline between measurements, which usually means 20% or more. Mercury lamps lose output continuously and a reading is a snapshot; if you set the line at target on the day of commissioning, the process is undercured before the next radiometer check. The alternative to margin is measuring often enough to catch the drift, which most lines do not do.

Why is the cure-zone length only used for belt speed?

Because the dose calculation needs exposure time, which you can get either from the zone and the speed or by timing directly. The zone length is here so the page can convert the required exposure into the setting an operator actually changes, which is the belt speed. If you already know your exposure time, the zone input only affects that one output.

My dose is on target and the part is still tacky. What is wrong?

Almost certainly oxygen inhibition rather than dose. Air quenches free radicals in the top layer of an acrylate, leaving a thin uncured skin over a properly cured bulk. Higher peak irradiance, nitrogen inerting, or a formulation change fixes it; more dose at the same intensity usually does not. The oxygen inhibition page models that directly.

Does a higher dose ever hurt?

Yes, which is why the process window has two sides. Overcure embrittles some coatings, yellows others, and on heat-sensitive substrates the infrared that comes with a mercury lamp distorts the part long before the UV does any harm. That is a thermal limit rather than a UV one, and the heat load page sizes it.

Finishing calculator

UV Dose Calculator (mJ/cm²) with Cure Margin

Convert what a radiometer measured into the dose a part actually receives, and check it against what the material needs. Enter the irradiance at the part, the exposure time, the cure-dose target from the datasheet, and the length of the cure zone. The calculator returns the delivered dose, how it compares with the target, the exposure that would hit the target exactly, and the fastest belt speed that still cures. Dose is the product of irradiance and time, so the same number can be reached many ways, and those ways are not interchangeable.

What this calculator does

Convert a measured irradiance and exposure time into delivered UV dose, compare it against the material's cure-dose target, and give the fastest belt speed that still cures.
Use it for commissioning a line against a material datasheet, checking whether an aged lamp still delivers the specified dose, finding the fastest belt speed a cure will tolerate, diagnosing an adhesion complaint by reconstructing the dose the part actually got, comparing a proposed lamp change against the current process window.
Convert a measured irradiance and exposure time into delivered UV dose, compare it against the material's cure-dose target, and give the fastest belt speed that still cures.

Formula used

UV dose (mJ/cm²) = irradiance (mW/cm²) × exposure time (s)
Dose as % of target = delivered dose ÷ cure-dose target × 100
Exposure needed = cure-dose target ÷ measured irradiance
Fastest belt speed = cure-zone length ÷ exposure needed × 60
Dose shortfall = max(target − delivered, 0)

Inputs explained

Measured irradiance at part: Peak irradiance a radiometer reads at the part surface, in the same spectral band as the material's cure-dose specification. Not the lamp's rated output.
Exposure time under UV: Seconds a point on the part spends inside the cure zone. On a conveyor this is cure-zone length divided by belt speed.
Material cure-dose target: The dose the material datasheet specifies, in a named band. If the datasheet does not name a band, that is a question for the supplier before it is a number for this page.
Effective lamp cure-zone length: Length of the illuminated zone along the direction of travel, measured where irradiance is actually useful rather than lamp housing to lamp housing.

How to use the result

Best suited to commissioning a line against a material datasheet, checking whether an aged lamp still delivers the specified dose, finding the fastest belt speed a cure will tolerate, diagnosing an adhesion complaint by reconstructing the dose the part actually got, comparing a proposed lamp change against the current process window.
Says nothing about depth of cure. This is a surface dose; through a pigmented or thick film, what reaches the bottom is far less. Does not model oxygen inhibition, which can leave a tacky surface at a dose the bulk cured at. Ignores spectral distribution beyond naming the band. Two lamps with the same UVA output can cure differently if their peaks sit differently against the photoinitiator's absorbance. Not a cure verification. A dose figure is a process setting; gel content or a rub test is evidence. Assumes a single pass under a uniform zone; shaped parts and multi-lamp tunnels need the shadowing and multi-lamp pages.

Common questions

Why does the spectral band matter so much? Because a photoinitiator absorbs over a narrow range and a radiometer measures over a defined band, and if the two do not overlap the dose figure is unrelated to the cure. A material initiated at 254 nm is cured by UVC; a UVA radiometer reading 1500 mJ/cm² next to it says nothing about whether that happened. This is the most common way a process that measures correctly still fails, and it is why every dose figure needs its band stated alongside it.
Is 100 mW/cm² for 12 seconds the same as 1200 mW/cm² for 1 second? The dose is identical and the cure is not. Through-cure depends on total dose, so the bulk sees roughly the same thing. Surface cure in air depends on peak irradiance, because oxygen diffusing into the top few microns quenches radicals and only a high generation rate outruns it. The low-intensity long exposure therefore tends to leave a tacky surface over a properly cured bulk: a result that confuses people precisely because the dose was right.
What dose margin should I run? Enough to cover the lamp's decline between measurements, which usually means 20% or more. Mercury lamps lose output continuously and a reading is a snapshot; if you set the line at target on the day of commissioning, the process is undercured before the next radiometer check. The alternative to margin is measuring often enough to catch the drift, which most lines do not do.
Why is the cure-zone length only used for belt speed? Because the dose calculation needs exposure time, which you can get either from the zone and the speed or by timing directly. The zone length is here so the page can convert the required exposure into the setting an operator actually changes, which is the belt speed. If you already know your exposure time, the zone input only affects that one output.
My dose is on target and the part is still tacky. What is wrong? Almost certainly oxygen inhibition rather than dose. Air quenches free radicals in the top layer of an acrylate, leaving a thin uncured skin over a properly cured bulk. Higher peak irradiance, nitrogen inerting, or a formulation change fixes it; more dose at the same intensity usually does not. The oxygen inhibition page models that directly.

Last reviewed 2026-08-25.