UV coating cure with measured dose — real mJ/cm² tied to the batch, not faith in the lamp.

The problem

A UV lamp doesn't fail — it ages. And the dose drops with no visible change in the coating.

The dose a coating actually receives depends on the lamp, the reflector and the line speed — and all three drift. A mercury lamp loses irradiance over its life, a dirty reflector eats a surprising share of the output, and a line sped up to hit the schedule cuts the exposure time. On paper, the radiometer exists to catch all of that. On the floor:

1. The radiometer run happens “when someone remembers” — no fixed cadence per lamp, and the reading ends up on a sheet, in a drawer, or nowhere.
2. Undercure is invisible — at 70% of the dose, gloss and touch are nearly identical. It surfaces in an MEK rub test, an adhesion failure or a customer complaint weeks later.
3. Speed changes are never recalculated — the line runs faster to make volume and the dose drops below the formulator's window with nobody doing the math.
4. Lamp replacement is guesswork — by hour meter or by outright failure, because there is no irradiance history per lamp to plan against.

The line lead knows it; the quality manager knows it. The classic system works almost every time. The batch it misses is the one that leaves “cured” to the eye and fails the solvent rub at the customer — and by then, several more batches have shipped under the same aging lamp.

How it fits the IRIS system

iLEAN doesn't change your radiometer — it ties every run to the lamp and the batch, and watches the trend.

Cure control does not fail for lack of a radiometer; it fails because the reading is never tied to the lamp or the batch. iLEAN acts as the putty between the radiometer, the cure tunnel and the batch record, changing none of the three.

Connect captures the radiometer. Edge ties dose and irradiance to lamp and batch. The agent watches the trend and warns before the window is breached. The person signs — never the other way round.

The iLEAN pieces applied to UV cure control:

• Connect — captures the radiometer you already use (a puck that rides the line or a probe) via USB, via its download software or, if it only has a display, from a photo of the screen that Connect reads. It also takes line speed and lamp power setting wherever the tunnel exposes them.
• Edge — identifies the line, the lamp and the running batch (QR/label or voice), so every radiometer run lands tied to all three, with date and product. It works without a network.
• Cure agent — checks dose and irradiance against the formulator's window for that product and watches the trend per lamp. A downward drift triggers the lamp-change or reflector-cleaning warning before the window is breached; a run below minimum holds the batches run since the last valid reading. The person signs — the batch is never released on its own.

See the full IRIS architecture →

Before and after

Radiometer readings on paper vs. dose tied to the batch with iLEAN

Aspect	Radiometer + sheet of paper	With iLEAN (dose tied to the batch)
Radiometer run	“When someone remembers”	Scheduled per lamp, captured at the source
Reading	Sheet of paper or memory	Dose + irradiance tied to lamp and batch
Line speed change	Dose recalculated by hand, or not at all	Recalculated and verified against the window
Lamp aging	Discovered through the failure	Trend per lamp with early warning
Undercured batch	Found by the customer	Held in-house before shipping
Lamp replacement	By hour meter or by breakdown	By measured irradiance data

Impact estimate

Impact estimate for your plant — to be validated with your numbers.

The block below is an estimate to be validated with the specific data of your plant. We put it forward so the committee has an order of magnitude; we refine it during the diagnostic.

• UV cure line for coatings, varnishes or industrial inks (wood, plastics, metal, industrial graphics) with a radiometer already in use.
• Pilot on one line (radiometer capture + lamp and batch identification + trend agent). First value expected within a few weeks.
• Indicative payback between 4 and 9 months, depending on the average cost of an undercured batch reaching the customer (replacement, freight, claim handling, brand damage).
• Hard levers: undercured batches at the customer trending to zero (estimate to be validated); lamp replacement driven by irradiance data, not by the calendar; a cure dossier per batch ready for audit.

And the quality manager's reasonable doubt

“What if the system misreads a value?” — false positives exist, and that is why the system never decides alone. Hallucination is a problem of free generation, not of anchored tasks. In tasks where the AI merely receives dose and irradiance from the radiometer and compares them against the formulator's window, the best models brought error below 1.5% ^[1]. And even so, what is critical is never decided alone: iLEAN warns and the operator or the quality manager sign. The three safety rings exist precisely for this.

^[1] OpenAI paper “Why Language Models Hallucinate”, 2025 — on the reliability of AI in anchored tasks.