Is 75% of new the right replacement threshold?

It is a common default and it is a proxy for something measurable. Whether it is right for a given line depends entirely on how much dose margin that line was commissioned with. A process that fails at 62% of new is losing lamp life to a 75% rule; a process that fails at 80% is being let down by it. The fourth row on this page computes where the threshold should sit for your process.

Why do my readings jump around between checks?

Measurement conditions almost always dominate. Distance is the biggest lever. A point-source lamp changes with the square of distance, so a small height difference swamps months of decay. Band selection, puck speed on a conveyor, warm-up time, and radiometer calibration drift all add to it. Fix every one of those and readings become comparable.

Does the lamp lose output evenly across the spectrum?

No, and this is a real trap. Mercury lamps age unevenly across UVA, UVB and UVC, and envelope solarisation cuts short wavelengths hardest. A broadband or UVA reading can look acceptable while the shorter-wavelength output a surface-cure photoinitiator depends on has fallen much further. Measure in the band your chemistry actually absorbs in.

Can I keep running a lamp below the required irradiance?

Only by giving it more time. Dose is irradiance multiplied by exposure, so slowing the line can restore the dose a decayed lamp no longer delivers at speed. Whether that trade is worth making is a throughput question rather than a cure question, and it has a limit, because some chemistries need peak irradiance rather than accumulated dose, oxygen inhibition being the usual reason.

What if I never recorded a new-lamp baseline?

Then the percent-of-new figure on this page is not meaningful, and the process-requirement rows are the ones to use. They need only today's reading and the cure trial. Record a baseline the next time a lamp goes in, at a fixed position, and the decay tracking becomes available from then on.

The reading came out above 100% of new. What happened?

Something other than the lamp changed. The usual causes are a baseline taken at a different position or belt height, a different spectral band, a different radiometer, or instrument drift. Lamps do not gain output with age, so a figure above the baseline is a measurement problem to resolve rather than good news.

UV Curing calculator

UV Intensity Decay Calculator: % of New vs What the Process Needs

Compare today's lamp reading with its new-lamp baseline, then check the plant's percentage replacement rule against what the process actually needs. Enter the current radiometer reading, the baseline recorded when the lamp was new, the percent-of-new threshold you replace at, and the irradiance below which this process stops curing. The calculator returns the decay figure, the headroom against the rule, and the margin against the process. And, where those two disagree, the percent-of-new at which the process genuinely fails. That last figure is the threshold the rule was trying to approximate.

What this calculator does

Track lamp output against its new-lamp baseline, and check the percentage replacement rule against the irradiance the process actually requires.
Use it for deciding whether a lamp flagged by the percentage rule really needs replacing, setting a replacement threshold from a measured process requirement, tracking decay against lamp hours to build a real curve for a system, investigating cure failures on a lamp the rule says is healthy, estimating remaining useful life between scheduled radiometer checks.
Track lamp output against its new-lamp baseline, and check the percentage replacement rule against the irradiance the process actually requires.

Formula used

% of new (today) = today's reading ÷ new-lamp baseline × 100
Headroom above threshold (points) = % of new − replacement threshold
Margin against the process (%) = (today's reading − required) ÷ required × 100
% of new where the process fails = required ÷ new-lamp baseline × 100
Reading the rule replaces at = baseline × threshold ÷ 100

Inputs explained

Today's irradiance reading: Current radiometer reading, taken at the same position, band and belt height as the baseline. A reading taken anywhere else is not comparable.
New-lamp baseline irradiance: The reading recorded when the lamp was new, in the same conditions. If no such reading exists, the decay figure on this page cannot be trusted.
Lamp replacement threshold: The percentage-of-new rule the plant swaps on. It is a proxy for the process requirement, not a measurement of it.
Irradiance the process requires: The reading below which this process stops curing at its current line speed. From a cure trial, not from the lamp datasheet.

How to use the result

Best suited to deciding whether a lamp flagged by the percentage rule really needs replacing, setting a replacement threshold from a measured process requirement, tracking decay against lamp hours to build a real curve for a system, investigating cure failures on a lamp the rule says is healthy, estimating remaining useful life between scheduled radiometer checks.
A single-point reading is not the whole cure surface. A lamp can hold its centre reading while its ends fall away, and this page cannot see that; the uniformity page can. Says nothing about spectral shift. An ageing mercury lamp can lose disproportionately in one band, so a UVA reading can look healthy while the UVC output a photoinitiator needs has fallen further. Irradiance is not dose. A lamp below the required irradiance may still cure at a slower line speed, which is a real option this page does not price. Treats the process requirement as a hard threshold, when in practice cure degrades gradually and the failure point depends on the acceptance test used. Cannot distinguish lamp decay from a dirty quartz sleeve, a degraded reflector or a drifting radiometer, all of which read as decay here.

Common questions

Is 75% of new the right replacement threshold? It is a common default and it is a proxy for something measurable. Whether it is right for a given line depends entirely on how much dose margin that line was commissioned with. A process that fails at 62% of new is losing lamp life to a 75% rule; a process that fails at 80% is being let down by it. The fourth row on this page computes where the threshold should sit for your process.
Why do my readings jump around between checks? Measurement conditions almost always dominate. Distance is the biggest lever. A point-source lamp changes with the square of distance, so a small height difference swamps months of decay. Band selection, puck speed on a conveyor, warm-up time, and radiometer calibration drift all add to it. Fix every one of those and readings become comparable.
Does the lamp lose output evenly across the spectrum? No, and this is a real trap. Mercury lamps age unevenly across UVA, UVB and UVC, and envelope solarisation cuts short wavelengths hardest. A broadband or UVA reading can look acceptable while the shorter-wavelength output a surface-cure photoinitiator depends on has fallen much further. Measure in the band your chemistry actually absorbs in.
Can I keep running a lamp below the required irradiance? Only by giving it more time. Dose is irradiance multiplied by exposure, so slowing the line can restore the dose a decayed lamp no longer delivers at speed. Whether that trade is worth making is a throughput question rather than a cure question, and it has a limit, because some chemistries need peak irradiance rather than accumulated dose, oxygen inhibition being the usual reason.
What if I never recorded a new-lamp baseline? Then the percent-of-new figure on this page is not meaningful, and the process-requirement rows are the ones to use. They need only today's reading and the cure trial. Record a baseline the next time a lamp goes in, at a fixed position, and the decay tracking becomes available from then on.

Last reviewed 2026-08-25.