Why is the delivered fraction so low?

Because the losses multiply. Reflector efficiency of 0.75, ageing at 85% and a window at 80% is 0.75 × 0.85 × 0.80 = 51%, so half the catalogue number never reaches the part. Each term looks individually reasonable, which is exactly why the compounding surprises people and why designing from the rated figure alone produces an undercured line.

Which loss should I attack first?

The window, because it is free. Cleaning quartz takes minutes and can return 10–20% immediately. Reflector cleaning and re-focus is next, cheap but bounded by the optics' condition. Lamp replacement is last because it costs money and downtime. The point of separating the three is that a shortfall tells you nothing about which one to fix unless they are shown apart.

Should I use this or measure?

Measure, whenever you can. This page exists for the cases where you cannot. A sealed tunnel, a bondline behind a substrate, a machine still on a drawing. A radiometer reading at the part is direct evidence and this is a chain of four estimates multiplied together. Where you have both, calibrate the reflector factor so the prediction matches the measurement, and then the page becomes useful for extrapolating.

Is window transmission really spectrally flat?

No, and that is a stated limitation rather than a detail. Solarised quartz attenuates short wavelengths far harder than long ones, so a window passing 90% of UVA might pass much less UVC. A process initiated in the short bands therefore loses more than this page suggests, and the flat-transmission assumption is optimistic exactly where the stakes are highest.

Does this work for LED arrays?

Poorly. LED output is far more stable with age, so the ageing term barely applies, and LED optics are typically close-coupled rather than reflector-focused, so the reflector factor does not mean the same thing. The LED array power density page models those systems on their own terms. Use this one for mercury arc systems, which is where the loss chain genuinely behaves this way.

Where does working distance fit?

Inside the reflector efficiency factor, which is a simplification. Distance has its own strong effect and its own law, and for a mercury arc lamp that law is not inverse-square in the near field, because the lamp is a line source rather than a point. The lamp distance page models that directly and is the right tool if distance is the variable you are actually changing.

UV Curing calculator

UV Irradiance at Part Calculator (Loss Chain)

Predict how much of a lamp's rated output actually reaches the part. Enter the rated peak irradiance, the reflector and focus efficiency, the ageing derate, and the transmission of any quartz window between lamp and part. The calculator works down the loss chain and returns the irradiance at the part, what each stage costs, and the fraction of the catalogue number that survives. Use it where you cannot get a radiometer to the part; where you can, measure instead, because a measurement beats a prediction every time.

What this calculator does

Predict the irradiance reaching a part from the lamp's rating, working through the reflector, lamp age and quartz window losses that stand between them.
Use it for estimating irradiance where a radiometer cannot physically reach the part, sizing a lamp for a machine that does not exist yet, explaining why a measured reading is half the catalogue number, deciding whether a shortfall is a cleaning job or a lamp purchase, sanity-checking a supplier's claimed irradiance at the part.
Predict the irradiance reaching a part from the lamp's rating, working through the reflector, lamp age and quartz window losses that stand between them.

Formula used

After reflector losses = rated irradiance × reflector efficiency
After ageing = after reflector × ageing derate ÷ 100
Irradiance at part = after ageing × window transmission ÷ 100
Lost to ageing = after reflector − after ageing
Lost to the window = after ageing − irradiance at part
Fraction delivered = reflector × ageing ÷ 100 × window ÷ 100 × 100

Inputs explained

Rated peak lamp irradiance: Manufacturer's rated peak output at the specified focal distance, for a new lamp. A catalogue number, not a measurement.
Focal / reflector efficiency factor: Fraction of rated output the optics actually place on the part, covering reflector condition, focus and working distance. Typically 0.6–0.85 for a maintained system.
Lamp aging derate: Output remaining as a percentage of new, from the lamp's hour count and its decline curve. 100% is a fresh lamp.
Quartz window transmission: Percentage passing the quartz window, sleeve or filter between lamp and part. Clean quartz is 90%+; clouded, solarised or overspray-coated quartz can be far worse and is free to fix.

How to use the result

Best suited to estimating irradiance where a radiometer cannot physically reach the part, sizing a lamp for a machine that does not exist yet, explaining why a measured reading is half the catalogue number, deciding whether a shortfall is a cleaning job or a lamp purchase, sanity-checking a supplier's claimed irradiance at the part.
A prediction, not a measurement. Every term is uncertain and they multiply, so the output carries all of that uncertainty. Window transmission is not spectrally flat in reality. Solarised quartz attenuates short wavelengths much harder than long, so a UVC process loses more than this suggests. Ignores distance explicitly; working distance is buried inside the reflector factor rather than modelled, which the distance page handles directly. Says nothing about uniformity. This is a peak at a point, not an average across a part. Does not cover LED arrays well; their output is far more stable with age and their optics work differently, so the power-density page suits them better.

Common questions

Why is the delivered fraction so low? Because the losses multiply. Reflector efficiency of 0.75, ageing at 85% and a window at 80% is 0.75 × 0.85 × 0.80 = 51%, so half the catalogue number never reaches the part. Each term looks individually reasonable, which is exactly why the compounding surprises people and why designing from the rated figure alone produces an undercured line.
Which loss should I attack first? The window, because it is free. Cleaning quartz takes minutes and can return 10–20% immediately. Reflector cleaning and re-focus is next, cheap but bounded by the optics' condition. Lamp replacement is last because it costs money and downtime. The point of separating the three is that a shortfall tells you nothing about which one to fix unless they are shown apart.
Should I use this or measure? Measure, whenever you can. This page exists for the cases where you cannot. A sealed tunnel, a bondline behind a substrate, a machine still on a drawing. A radiometer reading at the part is direct evidence and this is a chain of four estimates multiplied together. Where you have both, calibrate the reflector factor so the prediction matches the measurement, and then the page becomes useful for extrapolating.
Is window transmission really spectrally flat? No, and that is a stated limitation rather than a detail. Solarised quartz attenuates short wavelengths far harder than long ones, so a window passing 90% of UVA might pass much less UVC. A process initiated in the short bands therefore loses more than this page suggests, and the flat-transmission assumption is optimistic exactly where the stakes are highest.
Does this work for LED arrays? Poorly. LED output is far more stable with age, so the ageing term barely applies, and LED optics are typically close-coupled rather than reflector-focused, so the reflector factor does not mean the same thing. The LED array power density page models those systems on their own terms. Use this one for mercury arc systems, which is where the loss chain genuinely behaves this way.

Last reviewed 2026-08-25.