Build: The UVA Exposure Unit

Purpose

To build the one instrument in this course whose failure mode is silent. Everything else announces itself: a chemical stings, a hot iron burns, a mains fault trips something. A UVA array running with the lid ajar is invisible, painless and doing exactly what it was built to do.

So this build is organised around a single structural idea: the array cannot be energised in a state where anybody can see it. Not “should not” — cannot, because a switch in the low-voltage supply is opened by the lid before the lid is open far enough to look in. Everything else here — the heatsink, the baffled vents, the matt black interior, the registration stops — is engineering in service of that, or in service of the second requirement, which is that the same exposure means the same ultraviolet dose next month.

What you get at the end is a printing area of known evenness, a dose series for each process you print, and a temperature and hours log that tells you when to re-establish both.

Learning objectives

By the end of this session you should be able to:

• write a specification for an exposure unit from the largest print you will make, the slowest process you will print and the evenness you will accept, and derive the array from it rather than the reverse;
• lay out an array so that the pitch, the height and the overhang together meet an evenness target you can state before you build;
• design the whole electrical side at extra-low voltage from a certified enclosed supply, fused on the low-voltage side, with no mains conductor inside the unit;
• calculate a junction temperature from a datasheet’s thermal resistance and say what the calculation does and does not cover;
• build vents that pass air and stop light, and explain why the same baffle does both;
• fit an interlock that removes power from the array when the lid opens, and say honestly how far it falls short of a certified fail-safe interlock and what you do about that;
• commission the unit in the right order — leak, interlock, temperature, evenness, dose — and say why no print is made until all five have passed;
• map uniformity with coated strips rather than with a meter, and explain what that buys;
• establish a dose series for a process and say when it expires.

Prerequisites

Ultraviolet for the alternative processes, in full and without exception. The bands, the exposure limits, the wavelength argument, the control hierarchy and the ozone question are all settled there and none is re-argued here. This is the one prerequisite in the course that is a safety control rather than a convenience.

The contact printing frame, because this unit is a light source for that frame and inherits its geometry, its registration and — critically — its glazing decision: an acrylic-glazed frame under a 365 nm array is a frame with a filter in it.

Low-voltage electronics for the darkroom, in full. Constant-current driving, MOSFET low-side switching, the current budget, heatsinking and the five datasheet numbers are established there and used here without being taught again.

Safety classification

Level B, and the letter is earned by the source rather than by anything chemical. The rubric’s Level B assumes splash goggles, stronger ventilation, eyewash within reach and experience with concentrated reagents or with UV and low-voltage electronics. This page is the second of those.

The specific criteria that applied. An artificial optical radiation source capable of exceeding a published exposure limit at the working position, whose hazard is not perceptible to the person exposed and against which there is no aversion response. HSE list ultraviolet curing of inks — which is physically this instrument with a different job — among the sources presenting a reasonably foreseeable risk of harming the eyes and skin of workers. That is the classification, and the level does not come down because the unit is small.

Two steps carry controls of their own, and both are avoidable. Neither is a raised step under the mixed-level rule, because the session’s baseline is already B; they are named here because a reader who skims the level and not the controls will meet both of them anyway.

Soldering, on route B, carries Part XIV’s controls unchanged: iron in its stand, extraction at the joint, lead-free solder, nothing touched for thirty seconds. Route A avoids it entirely — Waveform’s own strips carry DC plugs on both ends.

First energisation with the lid open, during alignment, is the highest-exposure moment in the whole build and it is where people get hurt. It is done with UV-blocking eyewear on, skin covered, at the lowest current the array will run at, for the shortest time that lets you see what you need to see, and never with anyone else in the room who has not been told. Better still, avoid it: the alignment checks in stage 4 can all be done with the array dark and a torch.

What is not a hazard here. No chemistry is handled during the build itself: the Chemicals section a lab page would carry is replaced by Parts, and the only substances present are the adhesives and paint of a workshop session. There is no ozone — ICNIRP attribute hazardous ozone production to very intense UVC sources, particularly below 230 nm, and a UVA emitter at 365 or 395 nm with a 10 nm half width emits nothing near it, so the ventilation in this box is a thermal control and is not there for a gas. There is no mains work, because every conductor inside the unit is at 12 or 24 V from a certified enclosed supply with a moulded lead, on HSE’s reasoning that the supply voltage should be the lowest that will do the job. And there is no burn hazard from the emitters themselves at the currents specified — but there is one from the heatsink, which is doing its job when it is too hot to hold.

Hazards

Hazard	Where it arises	Control
UVA to the eye, imperceptible	Any moment the array is energised and not enclosed	Enclosure, interlock in the low-voltage supply, external indicator, no viewing window. Eyewear and covered skin as backup for servicing only
UVA to skin	Hands in the box during a stopped exposure; leakage at the lid line	The interlock, and the leak check at T1
Chronic low-level ocular exposure	Working beside the unit for long sessions	ICNIRP’s unweighted eye limit and their explicit caution that threshold data are lacking. The control is zero detectable leakage rather than a permitted dose
Heat: heatsink, plate, enclosure	The array at full current for a full-length exposure	Thermal design in stage 3, the temperature log at T3, a thermal cut-out, and a hands-off rule until the log says otherwise
Fire	An enclosed array run for tens of minutes with a blocked vent	Baffled vents that cannot be blocked flat, a thermal cut-out on the heatsink, and the unit is never left running unattended
Electrical, low voltage	Strip, driver, interlock wiring	Certified enclosed supply, moulded lead, fuse on the low-voltage side, nothing at mains potential opened or extended
Cut and drill	Enclosure, vents, plate	Eye protection, work clamped, cut away from the body
Mercury, if you use tubes instead	A broken UVA fluorescent tube	Not this build’s route; if you take it, see Waste streams

⚠The one instruction to read before the parts list

No viewing window. ICNIRP permit observation ports made of suitably tested ultraviolet-absorbing material — certain grades of acrylic, PVC or window glass — and this course declines the permission, because tested is the operative word and a home builder cannot test. An untested window is a window whose transmission you are guessing at, in the one part of the spectrum where guessing is invisible.

You do not need one. The alternative processes are printed by inspection with the lid open and the array off, which is what the split-back frame exists for, and which is exactly what both Ware and Bostick & Sullivan describe: open one leaf spring, fold back half the paper, look, close it, carry on. The interlock makes the array dark while you do it. That sequence is safer than a window and it is also the traditional method.

Required PPE

UV-blocking eyewear for any moment the array could be energised outside the enclosure — servicing, alignment, fault-finding. ICNIRP place personal protection last in the hierarchy, and it is last here too: it is what you wear when the engineering controls are deliberately defeated, not what makes the unit acceptable.

One reader-specific instruction, and it is not general advice. ICNIRP state that people who have had a cataract operation and received an intraocular lens not designed to absorb UVA, and people with no implant at all, should be fitted with UVA protective eyewear when working with UVA sources. An intact crystalline lens absorbs most UVA before it reaches the retina; a replacement without that property does not. If this is you, the eyewear is not backup — it is a primary control and it is worn whenever the unit is running, enclosed or not.

Covered skin — long sleeves — for servicing with the array live. Eye protection of the ordinary impact kind for drilling and cutting; that is a different pair of glasses doing a different job, and the two should not be confused.

Ventilation

Ventilation here is a thermal control. Nothing in this unit produces a vapour, a dust or a gas: the ozone question belongs to wavelengths below about 230 nm, which nothing in this box reaches. Where ICNIRP call for ventilation to exhaust ozone they are describing UVC sources.

What the airflow does is carry heat out of a closed box that is dissipating tens of watts for tens of minutes. It is designed in stage 3 and verified at T3, and its specification is a temperature, not an air change rate.

The one genuine airborne product in the box is not from the electronics at all. Ware notes that the iron processes outgas carbon dioxide during exposure, which is why he puts a felt blanket or porous plastic behind the print — enough of it in a sealed sandwich degrades resolution. That is a frame problem rather than a box problem, and the frame page handles it.

If you paint the interior with anything solvent-based, that step goes outdoors or at an open window and the box is left to cure completely before any coated paper comes near it.

Parts

Two routes through the array, and the rest is common.

Route A: LED strip — simpler, lower power, no soldering

Part	Quantity	What it does	Substitutes and notes
UVA LED strip, 12 V, 365 nm or 395 nm, with a published radiant output per unit length	1.5 to 2 m	The source	The published example is Waveform’s realUV at 0.7 W per foot radiant at 365 nm and 0.9 W per foot at 395 nm, FWHM 10 nm, on 1 m and 5 m reels. Named as the worked example because it publishes those figures, not as a recommendation. Buy one reel: a second reel from another bin will not match, and the uniformity map will find it
Aluminium backing plate, 3 mm, the size of the array field	1	Heatsink and flat mounting surface	Strip on bare wood runs hot and dies early. The plate is the heatsink for this route
Certified enclosed 12 V supply, moulded lead, rated ≥ 1.5 × the measured draw	1	Power	Waveform’s own supply powers up to 5 m of their strip, which is a useful sanity check on rating. Check the plug and fuse markings as the safelight build sets out

Route B: discrete emitters — higher power, needs soldering and real heatsinking

Part	Quantity	What it does	Substitutes and notes
High-power UVA emitters on star boards, 365 or 395 nm	9 to 16	The source	Nichia’s NVSU233B is the worked datasheet: at 1,000 mA, 1450 mW at 3.85 V for the 365 nm rank and 1650 mW at 3.65 V at 395 nm, junction maximum 130 °C, thermal resistance junction to solder point 3.9 °C/W typical, 5.7 °C/W maximum
Constant-current driver, low voltage in, current set for the emitter	1 per string	Sets and holds the drive current	An LED is a diode; a constant-voltage supply and a resistor is a current that drifts with temperature. Part XIV settles this
Aluminium extrusion or finned heatsink sized in stage 3	1	Removes the heat the emitters do not radiate	Thermal interface material under every star
Certified enclosed 24 V supply, moulded lead	1	Power	Higher voltage lets you put emitters in series, which keeps the current and therefore the wiring loss down

Common to both

Part	Quantity	What it does	Substitutes and notes
Enclosure: 12 mm ply or MDF, internal depth from stage 1	1	The light-tight box	It must be light-tight in the ultraviolet, which is a stricter test than the eye applies
Matt black interior finish	~250 mL	Suppresses reflected ultraviolet	ICNIRP call for surfaces coated with non-reflective material to reduce reflected UVR. Visibly shiny surfaces are often good UV reflectors
Interlock microswitch, rated for the array current, plus a second one	2	Removes power from the array when the lid opens	Two, in series, operated by different parts of the lid. See stage 4 for what this does and does not amount to
Indicator: a bright visible LED on the outside, on the array side of the interlock	1	Says the array is live	On the array side, so it goes out when the interlock opens. An indicator wired to the supply lies
Vent baffles: two sets of offset slots with a black-lined plenum	2	Pass air, stop light	The same geometry as the room’s vent trap, lined matt black
12 V fan, quiet, with a guard	1	Moves air through the baffles	Drawing air out is better than blowing it in: dust goes where the air comes from, and dust in this box lands on your negative
Fuse holder and fuse, low-voltage side	1	Fault protection	Sized to the array, not to the supply
Thermal cut-out, normally closed, on the heatsink	1	Opens the array supply if the heatsink exceeds its rating	Cheap insurance. See stage 3 for the honest position on choosing its rating
Hours counter, or a notebook page	1	Tracks lamp ageing	A 12 V mechanical hours meter costs little; a ruled page costs nothing and is used more often
Timed channel: Part XVII’s low-voltage output, or a bought timer	1	The exposure	Switches the low-voltage array, not a mains lamp
Registration stops for the frame	3	Fixed frame position	Two edges and a corner, as the contact printer’s

🛠Four things this page will not buy for you

A part number for a UK supplier. No UVA emitter or strip was verified as available in the United Kingdom at a given price while this part was written. Nichia, Luminus and Waveform are cited because they publish the figures a buyer needs; the course names the figures, not the shop.

Your own irradiance figure. Waveform publish an irradiance-against-distance report for their 365 nm strip, in microwatts per square centimetre, measured from a single 1 m section — and its numbers live on radial charts that the mirrored text of this course’s corpus does not carry, so this page quotes no irradiance from it. Their own caveats are worth keeping anyway: multiple rows increase irradiance “in theory” linearly, and the values are not guaranteed.

A detection threshold for the leak check. No published detection threshold for any UV-indicating card or fluorescent material was found. So “no leak detected” on this page means “below the threshold of an unknown-sensitivity indicator, in a dark room, to a dark-adapted eye”, and T1 says so rather than implying a measurement.

A thermal cut-out rating derived from a standard. No source in this course’s corpus establishes at what power an enclosed LED array warrants a thermal cut-out. The page fits one anyway and sets its rating from the temperature log, which is a design decision with its reasoning rather than conformity to anything.

Tools

Tool	For	Notes
Multimeter	Array current at the fuse; interlock continuity	The interlock test at T2 is a continuity test before it is anything else
Thermocouple or probe thermometer with a long lead	The temperature log, read from outside a closed box	The lead is the point: the log is taken with the lid shut, which is the condition that matters
Drill, step drill, jigsaw or router	Enclosure, vents, cable entries
Soldering iron and stand	Route B only	Route A needs a crimp tool instead
Thermal interface material	Route B, under every star
UV-blocking eyewear	Any live-and-open work	Listed as a tool as well as PPE because you will not do the work without it
A fluorescent indicator: a UV-reactive card, a banknote, or a white shirt	The leak check	ICNIRP note that colourless signature inks and banknote features fluoresce under UVA, which is the property this exploits
Timer or stopwatch	Everything in Testing and calibration

Estimated cost

Cost band £££, the highest in this part, and the money is in three places: the emitters, the enclosure and — if you take route B — the heatsink.

The planner prices none of it. It carries no line for UVA LED strip, UVA emitters, constant-current drivers, low-voltage supplies, heatsinks, fans, microswitches, thermal cut-outs or enclosure material. Its gap register already names the electronics and the enclosure hardware; this page adds UVA sources, thermal parts and the interlock switches to that list. Every cost statement here is a band and not a figure.

What the planner does price is the consumable this unit exists to expose: pre-coated cyanotype paper at £18.95 for 20 8 × 10 sheets (£0.95 a sheet) to £36.95 for 50 (£0.74). That is the number that should shape the build, because commissioning spends about a dozen sheets and every dose series after it spends more.

The honest comparison the planner permits: a bought contact printing frame is £74.80 to £77.99, and a bought UVA unit is not priced at all. If you find one, apply every test in Testing and calibration to it before you print, because buying moves the manufacturing and moves nothing else.

Estimated consumables cost

One run of this page means building the unit and commissioning it, which is where the consumption is. The unit, the emitters, the supply, the heatsink, the fan, the switches and the eyewear are capital.

Consumed	This session	Sourced price	Cost this session
Pre-coated cyanotype paper, 8 × 10 in — leak check, uniformity strips and dose series	12 sheets (2 for the leak and interlock checks, 9 cut into strips for the map, and the rest cut for the dose series)	£18.95 for 20 (£0.95 a sheet) to £36.95 for 50 (£0.74)	£8.88 to £11.40
Citric acid or plain water for developing the test strips	about 500 mL of 1 per cent	£10.00 per 250 g of the monohydrate, checked 7 September 2026	£0.20
Wash water	about 10 L	Not priced by the course	—
Matt black interior paint	about 200 mL	None. The planner carries no line for it	—
Thermal interface material, route B	9 to 16 pads	None. The planner carries no line for it	—
Solder and flux, route B	under 2 m	None. The planner carries no line for solder	—
Adhesive and tape	as needed	None	—

Subtotal: £9.08 to £11.60, and it is a floor rather than a total. Four of the seven rows carry no dated price and none of them is free. The figure also assumes the commissioning works first time, which it will not: budget twenty sheets, and note that cutting a sheet into strips is the single biggest economy in this part — nine uniformity readings come out of one 8 × 10 sheet cut into strips, not out of nine sheets.

If you coat your own paper rather than buying it pre-coated, the sensitiser chemistry and its costs belong to Part XXI and the planner could not price several of its components either.

Waste streams

The build itself produces workshop waste: offcuts of ply and aluminium, packaging, and a small amount of solder and flux residue on route B. None of it is photographic.

Commissioning produces the cyanotype wash water from the test strips, which is Part XXI’s stream and is treated under the course’s disposal policy: collected rather than tipped away on an assumption, with local regulation governing what happens to it.

At end of life, the unit is electrical waste. WM3’s list of waste puts household and household-type electrical equipment in chapter 20: 20 01 35* for equipment containing hazardous components and 20 01 36 for equipment containing none, with chapter 20 taking precedence over chapter 16 for domestic-type items. An LED array with no battery and no mercury falls in the second of those; if you built the fluorescent-tube version instead, the tubes are 20 01 21*, fluorescent tubes and other mercury-containing waste, which is an absolute hazardous entry and goes to a household hazardous waste route rather than the bin. Local regulation governs disposal and this page gives no jurisdiction-specific instruction.

Alternative route

Sunlight, with the same tests applied. The complete alternative, and for cyanotype it is not a poor one: Ware’s figures put average sunlight at 30 to 40 W/m² of UVA and full density at 2 to 4 minutes. What you must supply instead of the unit is a written record — date, time of day, cloud, orientation — and the print-out judgement through a split-back frame, continuing until the deepest shadows reverse. What you cannot supply is repeatability, and you should not pretend otherwise. The uniformity test still applies: the sun is even, but a partly shaded frame is not, and the strip test at T4 will find it.

A bought UVA unit. Legitimate, and it exempts you from nothing. Run T1, T2, T3, T4 and T5 on it exactly as written. A bought unit is more likely to have a real interlock and more likely to have a viewing window you should verify or cover.

A nail lamp, for postcard work only. The honest small route. A nail lamp is a UVA array with no enclosure, designed to irradiate a hand at a few centimetres. Used as sold it is the geometry every control on this page exists to prevent, so the course permits it only one way: inside a box you build, with the same lid, the same interlock, the same matt black interior and the same five tests. That is this build with a cheaper array, and it works up to about 100 × 150 mm.

No workshop, no tools, no space to keep a box. Take the sunlight route and build only the frame. The frame is the part that cannot be substituted; the light source can be the sky.

Cannot handle the coated paper or the wet processing. This unit is a light source and it does not solve that. The alternative-process parts carry their own accessibility routes; what this page can say is that the exposure step itself is a timer and a lid, which is among the more accessible operations in the course.

Build

Stage 0 — Write the specification before you buy anything (20 minutes)

Four numbers, in the notebook, before a parts order.

1. The largest print. Not the largest you dream of — the largest the alternative-process parts actually ask for. 8 × 10 in with a margin means a printing area of about 250 × 300 mm.
2. The target exposure time for the slowest process you will print. Somewhere between five and fifteen minutes is the sensible band: much shorter and small timing errors matter, much longer and a session becomes an afternoon and the box has time to heat the paper.
3. The uniformity target, stated as what it will do to a print. This page uses a quarter of a printing-out step across the printing area, established by strip test rather than by meter.
4. The constraint that decides the enclosure: nobody standing beside the unit receives ultraviolet. Not “less than a limit” — zero detectable, for the reason the lesson gives: a home unit has no calibrated instrument with a known spectral response, so a computed permissible leak would be a number you cannot check.

Then the sanity calculation, which decides whether your array is plausible. Ware’s platinum-palladium figure is a full-scale image in about two minutes at 50 W/m² at 365 nm, and his simple cyanotype takes five to ten minutes under “an average 365 nm source”. Take 25 W/m² as a modest target at the paper — half of sunlight — over 0.075 m² of printing area:

P = E × A = 25 × 0.075 = 1.9 W

Radiant power that must land on the printing area

P is the radiant power in watts, E the target irradiance in watts per square metre and A the printing area in square metres. Not all of the array’s output lands there: some goes to the walls, some past the edges, some back into the box. If you assume 60 per cent reaches the printing area — an assumption, not a measurement, and one your dose series will correct — the array must emit about 3.2 W of ultraviolet.

Against Waveform’s published 0.7 W per foot at 365 nm, that is about 4.6 feet, say 1.5 m of strip. Against Nichia’s 1.45 W per emitter at 1 A, it is three emitters at full current — which is why route B is usually built with nine to sixteen emitters run well below maximum instead: same total, much cooler, and far more even.

📓What that calculation is and is not

It is an order-of-magnitude check that stops you buying half a metre of strip for an 8 × 10 unit or sixteen 3 W emitters for a postcard one. Its 60 per cent coupling factor is an assumption and its 25 W/m² target is a choice.

It is not a prediction of your exposure time, and nothing on this page treats it as one. The dose series at T5 is the measurement, and if it comes out at three times the calculation you have learnt something about your box’s coupling rather than found an error.

Stage 1 — Array layout, and the arithmetic that ties pitch to falloff (30 minutes)

Three dimensions, and they are not independent: the pitch p between rows or emitters, the height h from the array to the paper, and the overhang by which the array extends past the printing area on every side.

Pitch against height governs the ripple. Directly under an emitter the paper sees one close source plus several distant ones; midway between two it sees two at middle distance. The difference shrinks as h rises and as p falls. As with the contact printer’s light box, the working rule is h at least equal to p, preferably twice — but a UV unit cannot buy its way out of a bad ratio with a diffuser, so the rule binds harder here.

Overhang governs the corners. A corner of the printing area is further from most of the array than the centre and sees it more obliquely, so it receives less. The remedy is to make the array bigger than the printing area — at least h of overhang on every side is the starting point. Overhang costs emitters and it costs nothing else, which makes it the cheapest evenness you can buy.

And diffusion is not available to you. ICNIRP record that polymethyl methacrylate and polycarbonate normally do not transmit below about 370 nm, which puts an ordinary acrylic diffuser directly in the way of a 365 nm array. A UV unit’s evenness comes from geometry. That single fact is why this array is larger, flatter and closer to the frame than the contact printer’s, and why the overhang matters so much more.

The unit in section: array, heatsink, baffled vents, lid interlock and the frame's fixed position

1. 1UVA array at pitch p, on the lid, facing down — route A is strip on a 3 mm plate; route B is 9 to 16 emitters on stars
2. 2Heatsink, fins outside the box — the heat leaves the enclosure rather than circulating in it
3. 3Height h, array to paper, with h at least p and preferably 2p — no diffuser is available: most acrylic cuts off around 370 nm
4. 4Overhang, at least h on every side — the cheapest evenness in the build
5. 5Matt black on every interior surface — ICNIRP: shiny surfaces are often good UV reflectors
6. 6Baffled inlet and fan-drawn outlet, both black-lined — no straight line from inside to outside; the fan draws out, so dust does not come in with the air
7. 7Two interlock switches in series, in the low-voltage array line — the lid lifts about 2 mm before the first opens
8. 8Indicator on the array side of the interlock — wired to the supply instead, it would lie
9. 9Three battens locate the frame — same exposure means the same dose only if the frame is in the same place

Drawn readable rather than to scale. The one proportion worth getting right on paper first is the overhang against h; everything else is checked by measurement in Testing and calibration.

Stage 2 — The electrical side, entirely at extra-low voltage (40 minutes)

Every conductor inside this box is at 12 or 24 V, and the reason is HSE’s: limit the supply voltage to the lowest that will do the job, because 230 V AC can kill and the risk is greatest in wet surroundings. A room where prints are washed is wet surroundings.

The supply. A certified, enclosed, marked constant-voltage unit with a moulded lead, rated at least 1.5 times the measured draw. Nothing about it is opened, extended or re-terminated. If the lead is too short, buy a longer one or move the box.

The fuse goes on the low-voltage side, sized to the array rather than to the supply. A supply rated 1.5 times your load will happily deliver 1.5 times your load into a fault.

The switching element is a low-side transistor, exactly as Part XIV established. The timed channel drives its gate; the interlock and the thermal cut-out sit in the supply line in series, so either can remove power regardless of what the timer is doing.

And there is no mains conductor inside the unit — including the one you might be tempted to add. Switched mains output is Part XVII’s subject, and its survey of 5 September 2026 answered the question this page used to leave open: no acceptable certified module is sold at hobby cost, so the course builds none and the timer’s only output is low voltage. If you want the unit switched by a wall timer, use a plug-in timer between the socket and the supply’s own plug — no fixed wiring, nothing opened. If you want a fixed switched circuit, that is work for an electrician and not for this page.

!Why the low-voltage rule earns its keep specifically here

On the safelight and the contact printer, extra-low voltage was a general safety policy. On this unit it is what makes the central control possible.

An interlock has to interrupt the power to the array, reliably, thousands of times, operated by a lid, built by you, in a box you also made. At 12 V and a couple of amps that is an ordinary microswitch doing an ordinary job, and a fault in it is a dead array. At 230 V it would be a safety-critical mains switching problem requiring a rated, certified device correctly installed, and a fault in it would be something else entirely.

The safety architecture of this instrument depends on its supply voltage. That is worth noticing, because it is the clearest example in the course of a safety rule that buys capability rather than costing it.

Stage 3 — Thermal design, and the calculation you can actually do (40 minutes)

An LED converts most of its input to heat, and an enclosed array run for ten minutes is a small oven. Three consequences: the emitters age faster and dim, the box warms the paper, and at the extreme something scorches.

The calculation the datasheet supports. Nichia give a junction temperature maximum of 130 °C and a thermal resistance from junction to solder point of 3.9 °C/W typical and 5.7 °C/W maximum. Take one emitter at 1,000 mA and 3.85 V:

ΔT = P × R(θJS) = 3.85 × 3.9 = 15 °C

Junction rise above the solder point

P is the electrical power into the emitter in watts and R(θJS) the junction-to-solder-point thermal resistance in degrees per watt. Using the maximum 5.7 °C/W instead gives 22 °C. So to keep the junction under 130 °C, the solder point must stay below about 108 to 115 °C, and Nichia’s operating temperature range tops out at 85 °C anyway, which is the tighter constraint.

🧮What that calculation does not cover, and why the log exists

It covers one link in the chain: junction to solder point, inside a part somebody else designed and characterised.

It does not cover solder point to star board, star board to thermal pad, pad to heatsink, or heatsink to the air in a box with one small fan — and those three or four links are the ones you built, with no datasheet and no characterisation. Their total resistance is easily several times the one you can look up.

So the calculation tells you your ceiling and the temperature log tells you where you are. That is why T3 is not optional and why it is taken with the lid shut: an array measured with the box open is an array in a different thermal environment from the one it works in. The course publishes no thermal claim it has not measured, and it has measured nothing on your box.

The physical answers, in order of effectiveness. Put the emitters on aluminium with thermal interface material under every one. Put the fins outside the enclosure, so the heat leaves rather than circulating. Draw air out with the fan rather than blowing it in, so the box runs slightly below atmospheric and dust is drawn through the baffles rather than settling on your negative. And spread the load: sixteen emitters at 350 mA make the same light as five at 1 A and are far easier to cool, as well as far more even.

Vents that pass air and stop light. The same geometry as the room’s vent trap and for the same reason: two sets of offset slots with a black-lined plenum between them, so that no straight line runs from inside to outside. Light travels in straight lines; air does not have to. Line the plenum matt black, because ICNIRP note that visibly shiny surfaces are often good ultraviolet reflectors and a shiny baffle is a mirror that pipes the light round the corner for you.

The thermal cut-out. A normally-closed bimetallic switch bolted to the heatsink, in series with the array supply. No source in this course’s corpus establishes at what power such a cut-out becomes necessary, so the course fits one on the reasoning that it is cheap, passive and cannot be forgotten, and sets its opening temperature from the temperature log: about 20 °C above the highest steady temperature T3 records, and below anything in the box that could soften or scorch.

Stage 4 — Enclosure and interlock (50 minutes)

The enclosure is light-tight in the ultraviolet, which is a stricter test than your eye can apply. A gap that shows no visible light may still pass UVA, which is why T1 uses a fluorescent indicator rather than an eye. Line the lid seam with a felt or foam light-trap strip, run the cable through a grommet with a bend in it, and paint every interior surface matt black.

The interlock, and the honest account of what it is. Two microswitches, in series, in the low-voltage supply line to the array, operated by two different parts of the lid, positioned so that the lid lifts about 2 mm before the first one opens — long before the gap could pass a beam to an eye.

ICNIRP are specific that where direct access to a source is required, interlocks should be fail-safe, manufactured, installed, tested and used to agreed relevant technical standards. Their own worked example is a door switch on a germicidal irradiation unit that removes power from the lamps when the door opens, and that is precisely the arrangement you are copying.

This course will not claim your two microswitches are a certified fail-safe interlock, because they are not. What they are is a redundant pair of ordinary switches in a series arrangement that fails safe in the common failure modes — a stuck switch, a broken lever, a lost adjustment — because either one opening kills the array. What the arrangement does not have is a certificate, a rated life, or an independent test. The course’s answer to that gap is not a stronger claim; it is three cheap compensations:

1. Redundancy, which is the second switch.
2. A test performed twice, every time, at T2 and at the start of every session — not a commissioning test but a habit.
3. An indicator wired on the array side of the interlock, so the state of the array is visible from across the room and does not depend on believing the switch.

The indicator is not decoration. This instrument’s characteristic failure is a live array nobody knows is live. Wire the indicator so that it is powered by the same line the array is: if it is lit, the array is energised; if it is out, the interlock has done its job. An indicator wired directly to the supply tells you the box is plugged in, which is not the question.

Stage 5 — Control, timing and the hours counter (25 minutes)

A manual switch for setting up, so you can energise the array deliberately with the lid closed to check the indicator and the leak. A timed channel for exposures, from Part XVII’s low-voltage output or a bought timer, switching the same 12 or 24 V. As on the contact printer, exposure is by time at a fixed current and there is no dimmer: an LED’s output is not proportional to its current, current shifts the junction temperature and therefore the output, and a knob position is not a record.

The hours counter is the part people skip and then regret. LED output falls slowly with accumulated hours and fluorescent output falls faster; this course could find no published decay curve for UVA sources of either kind, so it offers no replacement interval and instead offers the only thing that survives not knowing: a running total of hours, and a dose series re-established at an interval you choose and write down.

Ware’s observation about the professional units is the same problem solved with money — a NuArc carries a light integrator that measures irradiance continuously and accumulates dose, reading out in units of exposure rather than time, precisely because the source’s emission varies. Your hours counter is the poor relation of that integrator, and the dose series is what you do instead of owning one.

Stage 6 — Registration (10 minutes)

Three battens on the floor of the box: two along adjacent edges and one at the corner between them. The frame goes against all three, every time, in the same orientation. Mark the sheet sizes you use on the floor of the box in a paint that does not fluoresce.

This costs nothing and it is what turns “twelve minutes” into a dose rather than a duration. Change the frame’s height in the box and every number in the dose series is void, because the irradiance changed.

Stage 7 — The label and the log (10 minutes)

On the outside of the box, where you will read it:

UVA exposure unit. Peak wavelength ____ nm. Array current ____ mA at ____ V. Printing area ____ × ____ mm. Array-to-paper height ____ mm. Warm-up ____ min. Highest heatsink temperature recorded ____ °C on ____ (date). Uniformity: worst strip ____ against centre, ____ (date). Interlock tested ____ (date).

And a page in the notebook headed hours, with a line per session. Both stay blank until Testing and calibration fills them, and until then this is a box with an ultraviolet source in it and no evidence about either.

Testing and calibration

In this order, and no coated paper meets a negative until all five have passed. The order is not arbitrary: each test assumes the previous one.

Commissioning order, and what each test lets you do next

1. 1T1 — Leak checkEstablishes that the enclosure contains the light. Until it passes you cannot safely run the array long enough to do any other test.
2. 2T2 — Interlock, tested twiceEstablishes that opening the lid removes power. Until it passes, every subsequent test is done with a hazard the controls do not cover.
3. 3T3 — Temperature log, lid shutEstablishes that a full-length exposure is thermally survivable, and sets the thermal cut-out. Until it passes you do not know how long you may run the array.
4. 4T4 — Uniformity map on coated stripsEstablishes that the field is even enough for the target. It needs T3, because the map must be made at the running temperature the unit will actually work at.
5. 5T5 — Dose seriesEstablishes the exposure for each process. It needs T4, because a dose measured at the centre of an uneven field is a number that applies to one square centimetre.
6. 6Only now: a real printWith the exposure written on the label, the hours log opened, and the whole chain of evidence dated.

Each step is a precondition for the next, which is why a unit half-commissioned is a unit that is not commissioned.

T1 — The leak check

What you are looking for is fluorescence, not light. ICNIRP note that colourless signature inks and banknote features fluoresce under UVA, and that is the property to exploit: many white papers, white shirts, laundered fabrics, UV-reactive beads and banknotes all glow visibly under UVA that your eye cannot see directly.

1. Close the lid, energise the array on the manual switch, and darken the room completely.
2. Dark-adapt for at least five minutes. This is the step people skip and it is the step that finds leaks. Kodak’s own room-darkness test uses the same five minutes for the same reason.
3. Move a white card, a banknote or a UV-reactive indicator slowly along every seam: the lid line on all four sides, both vents, the cable entry, every fastening, and the joint between the heatsink and the lid. Watch the indicator, not the box.
4. Then do it once more with the card held where your face would be during an exposure, and once at the position of any other person in the room.

The criterion is zero detectable fluorescence. And the page is explicit about what that means: no detection threshold for any UV-indicating material was sourced, so “no leak detected” means “below the threshold of an indicator of unknown sensitivity, to a dark-adapted eye” and not “below an exposure limit”. The reason the course sets the bar at zero detectable rather than at a computed number is given on the lesson page: a permissible leak would require a calibrated instrument with a known spectral response, which a home unit does not have, and a limit you cannot measure against is not a control.

If you find a leak: black tape from the inside, more felt on the seam, a deeper plenum in the vent, a bend in the cable’s path. Then re-test. A unit that leaks at one seam usually leaks at three.

T2 — The interlock test, performed twice

Twice, because a control tested once is a control tested in one state.

Test 2a, electrically, array disconnected. Multimeter on continuity across the interlock pair. Lid shut: continuity. Lift the lid slowly and note the gap at which continuity breaks — it should be about 2 mm and it must be before any line of sight into the box exists. Repeat for each switch alone by bridging the other, so you know both work rather than knowing one does.

Test 2b, functionally, array connected and running. Lid shut, array on, indicator lit. Lift the lid. The indicator must go out before the lid has opened far enough to see inside. Watch the indicator, not the interior. Close it, and confirm it comes back on.

Record the date on the label. Repeat 2b — five seconds — at the start of every session. A microswitch lever bends, an adjustment creeps, and a lid sags on its hinges over a year.

If the indicator stays lit when the lid opens, stop. Unplug the supply at the socket before investigating. The commonest causes are a switch wired normally-open instead of normally-closed, an indicator taken from the supply side rather than the array side, and a lever that has been bent so the switch never travels.

T3 — The temperature log, with the lid shut

The probe goes on the heatsink, at the hottest point you can reach — between two emitters rather than at an end — with its lead run out through the cable entry so the box stays closed.

1. From cold, energise the array and read the temperature every minute for twice your longest intended exposure, or until the reading stops rising for three consecutive minutes.
2. Record ambient temperature too. A log without an ambient is a log that will confuse you in July.
3. Note the time at which the reading is within 1 °C of its final value. That is your warm-up, and it goes on the label: an array printed with cold is an array delivering more than the dose series says.

Minutes	Heatsink, °C	Ambient, °C	Notes
0			from cold
1
2
5
10
15
20

Read the shape. A curve that flattens is a heatsink that is winning. A curve still climbing at twice your exposure time is a heatsink that is not, and the fixes in order of effectiveness are: more emitters at lower current, fins outside the box, a bigger heatsink, more airflow. A curve that flattens and then climbs again is usually a fan that has stalled or a vent that has been blocked by whatever you put down beside the unit.

Then set the thermal cut-out about 20 °C above the highest steady reading, and below anything in the box that could soften. Write the highest steady reading on the label.

T4 — The uniformity map, on coated strips

The honest instrument for this field is the chemistry, because it responds in the band the paper uses. A broadband UV meter of unknown spectral response measures something adjacent.

1. Coat one sheet, or take one pre-coated sheet, and cut it into nine identical strips. They must come from one sheet and one coating: two sheets coated at different times differ by more than the thing you are measuring.
2. Lay the strips across the printing area in a three-by-three grid — corners, edge centres and middle — in the frame, under the glass, with the frame in its battens.
3. Warm the unit up first, then expose all nine simultaneously for a time that puts the centre strip in mid-tone. Simultaneously is the whole method: it removes source drift, warm-up and timing from the comparison in one step.
4. Wash and dry all nine together, then lay them out in their grid positions and compare.

Where the nine strips go, and what the array must overhang

1. 1Array field: emitter rows at pitch p — h at least p, preferably 2p; there is no diffuser to fall back on
2. 2Printing area, 250 x 300 mm — an 8 x 10 in sheet with a margin
3. 3Overhang, at least h on every side — the cheapest evenness in the build, and the fix the map usually calls for
4. 4Nine strips, three by three, corners included — one sheet, one coating, one simultaneous exposure
5. 5The centre strip is the reference — the other eight are read as differences from it
6. 6The frame's registration battens — the map is only valid for this frame position

Plan view. Take the corners: a three-point map along one axis cannot see the fall-off that an array of finite size produces at its corners, which is the fault this test most often finds.

Read them against the centre strip. If you have a densitometer, read them; if not, sort them by eye against a neutral surround, which is coarse and honest. The target from stage 0 is a quarter of a printing-out step across the printing area.

📓Reading the map's shape

Dark centre, light corners, symmetrically: not enough overhang, or h too small. Add overhang first — it is the cheaper fix and it costs no evenness elsewhere.

A gradient from one side to the other: the array is not parallel to the frame, or one end of a strip run is at a lower voltage than the other because it is fed from one end. Feed long runs from both ends and check the lid sits square.

A regular ripple matching the emitter pitch: h is too small for p. Raise the array or add emitters between the existing ones. You cannot diffuse your way out of this one, which is the whole difference between this unit and the visible-light box.

One dark strip with no pattern: usually the strip, not the unit. Re-cut and repeat before you rebuild anything.

T5 — The dose series

Finally, the numbers the unit exists to produce — one series per process, because the absorbed fraction differs between them.

Take a coated sheet, cut it into strips, and expose a series of times spaced a stop apart — 2, 4, 8, 16 minutes for cyanotype is a sensible bracket, given Ware’s five to ten minutes under an average 365 nm source. Include a step wedge on at least one strip, so the series tells you about the tonal scale and not only about the overall level. Process them together, dry them, and read them.

For a printing-out process, the endpoint is not a density but an appearance, and it is well described: continue until the high values are light green, the mid-tones firm blue and the deepest shadow tones reversed to a pale blue-grey, which Ware calls a solarised look. Bostick & Sullivan describe the same judgement as watching the darkest areas begin to reverse and become lighter, and note that the print must be over-exposed for a permanent image because the blue that appears first washes away in plain water.

Write the result on the label and in the log, with the process, the paper, the coating date, the array-to-paper height and the date. And note when it expires: it is void when the height changes, the paper changes, the emitters have accumulated enough hours to matter — a figure you do not know, which is why you count them — or a year has passed, whichever comes first.

Troubleshooting

Symptom	Likely cause	Test that distinguishes it
Indicator stays lit with the lid open	Switch wired normally-open, indicator taken from the supply side, or a bent lever	Unplug at the socket, then T2a with the meter
Exposure times far longer than the stage 0 calculation	Coupling worse than the assumed 60 per cent, acrylic glazing in the frame, or a 395 nm array on cyanotype	Swap the glazing for glass and repeat one strip. If it halves, the acrylic was the filter
Exposures creeping longer over months	Emitter ageing, a dirty enclosure interior, or a fan that has stopped	The hours log, then a fresh dose series. Repaint the interior if it has yellowed
Prints paler at the edges	Overhang or h too small	T4, and read the shape
A regular ripple in the print	h too small for p	T4; raise the array, do not add a diffuser
Heatsink hotter than the log	A blocked vent, a stalled fan, or a higher ambient	T3 again with the ambient recorded
A faint fog across the whole sheet	Coated paper stored or handled near a window or a fluorescent or LED room light	Not the unit. Coat, dry and store in the dark
First strip of a session is different from the rest	No warm-up	T3’s warm-up figure, then apply it
The print is soft although the unit is even	Gap in the frame, not the light	The contact frame’s pressure check
Fluorescence visible at a seam that passed before	The seam has opened, or the felt has compressed	T1, and re-tape from the inside

Questions

1. Your array emits 3.2 W of UVA and your printing area is 0.075 m². State the irradiance at the paper if the coupling were 100 per cent, then if it were 40 per cent, and say what measurement would tell you which is nearer the truth.
2. Explain why an indicator wired to the supply rather than to the array side of the interlock is worse than no indicator at all.
3. A reader argues that since UVA is much less biologically effective than UVB — 0.00011 against 1.000 in ICNIRP’s weighting — an enclosure is over-engineering for a 365 nm array. Answer them with the relevant limit and one arithmetic step.
4. Using Nichia’s figures, calculate the junction rise above the solder point for one emitter at 700 mA and 3.7 V, at both the typical and maximum thermal resistance. Then say which parts of the thermal path your calculation has not touched, and how you would find out about them.
5. Your uniformity strips show the four corners clearly lighter than the centre and no other pattern. Give the two possible causes and say which you would try first and why.
6. Why does this build put the uniformity map before the dose series rather than after it? Give the failure that the other order produces.
7. You are offered a second-hand UVA unit with a viewing window of unknown material. List the checks you would make before using it, and say what you would do if the window’s transmission could not be established.
8. Explain, to somebody who thinks the low-voltage rule is a general caution, why it is specifically the thing that makes this instrument’s interlock a reasonable project for a home builder.

Further experiments

Measure the coupling factor you assumed. Expose one strip at the centre of the printing area and one lying face-up outside the frame, right against the array, for the same time. The ratio of the exposures needed to reach the same appearance is your box’s coupling, measured. Compare it with the 60 per cent stage 0 assumed, and correct the calculation for next time.

Race the sun. Coat one sheet, cut it in half, and expose one half in the unit and the other in direct sunlight, both to the same visual endpoint, both timed. You will have measured your unit in units of sunlight, which is the most portable calibration there is — and against Ware’s 30 to 40 W/m² for average sunlight, it gives you an irradiance figure for your array without owning a meter.

Test reciprocity on your own unit. Expose one strip at the full array current for time t, and another with half the emitters disconnected for 2t. If reciprocity holds over that range, the two should match. This is the experiment the lesson says the corpus does not answer for the iron processes, and it is well within reach.

Settle the wavelength question for your own process. If you can borrow or afford a second short strip at the other common peak, build a second small array and run identical dose series. Ware’s calculation says the 365 nm route should win decisively for cyanotype — an absorbed fraction of 0.65 against 0.03 at 420 nm — and predicts nothing at all about Van Dyke or palladium, which have different internal filters. Measure it and you will know something the literature does not tell you.

Log the ageing properly. Re-run a single-strip dose check every fifty hours on the counter and plot the exposure against accumulated hours. In a year you will have the decay curve for your own array that no manufacturer in this course’s corpus publishes.

Recap

This unit exists to make an invisible hazard impossible rather than unlikely, and everything structural follows from that: an enclosure with no viewing window, two interlock switches in series in the low-voltage supply so the lid removes power before it opens far enough to see in, an indicator on the array side of that interlock, and a matt black interior because shiny surfaces reflect ultraviolet well. The course does not claim the home-built interlock is a certified fail-safe device; it compensates with redundancy, a test done twice, and an indicator that does not depend on believing the switch.

The array is designed from a specification rather than from a parts bin: printing area, target time, evenness target, and zero detectable leakage. Pitch, height and overhang set the evenness between them, and a UV unit cannot buy evenness with a diffuser because most acrylic stops transmitting below about 370 nm — so geometry does the work and overhang is the cheapest part of it. Every conductor runs at 12 or 24 V from a certified enclosed supply, which is not only a safety policy here but the thing that makes an ordinary microswitch an adequate interlock.

The thermal calculation you can actually do covers one link of the chain, junction to solder point, and the temperature log covers the links you built. And the commissioning runs leak, interlock, temperature, uniformity, dose — in that order, because each is a precondition for the next, with no print made until all five have passed and every result written on the box with a date on it.

Check your understanding

Question 1. Why is the interlock placed in the low-voltage supply line to the array rather than anywhere else in the circuit?

• Because a low-voltage switch is cheaper than a mains-rated one
• Because at 12 or 24 V an ordinary microswitch can reliably interrupt the array's power, which makes a home-built redundant interlock a reasonable project — the same function at 230 V would be safety-critical mains switching needing a rated certified device
• Because the timer cannot switch mains
• Because interrupting the signal to the driver would be slower

Show the answer and why

Answer: Because at 12 or 24 V an ordinary microswitch can reliably interrupt the array's power, which makes a home-built redundant interlock a reasonable project — the same function at 230 V would be safety-critical mains switching needing a rated certified device

It is the clearest case in the course of a safety rule buying capability rather than costing it. The whole architecture of this instrument depends on being able to remove power from the source with a switch operated by a lid — built by you, in a box you made, operating thousands of times. At extra-low voltage that is an ordinary component doing an ordinary job and its failure mode is a dead array. At mains potential it would be a different engineering problem with a different failure mode, and it is exactly the work the course sends to a competent person.

Question 2. The unit passes the leak check. What has actually been established?

• That leakage at the operator position is below the ICNIRP exposure limit
• That no fluorescence was detectable, to a dark-adapted eye, using an indicator of unsourced sensitivity — which is a pass against the course's stated criterion of zero detectable, not a measurement against a limit
• That the enclosure is light-tight in the visible band
• That the interlock is working

Show the answer and why

Answer: That no fluorescence was detectable, to a dark-adapted eye, using an indicator of unsourced sensitivity — which is a pass against the course's stated criterion of zero detectable, not a measurement against a limit

The page is deliberate about this. Converting a leak into a permissible standing time needs a calibrated instrument with a known spectral response, which a home unit does not have, and no detection threshold for any UV-indicating material was found in this course's corpus. So the criterion is set at zero detectable rather than at an invented number, and the pass is reported as what it is. Note also that visible light-tightness is the weaker test: a seam can show no visible light and still pass UVA, which is why the check uses fluorescence rather than the eye.

Question 3. Your uniformity strips show a regular ripple whose spacing matches the emitter pitch. What is the fix, and what is different about this fault in a UV unit compared with the visible-light contact printer?

• Add a diffuser, exactly as on the contact printer
• Raise the array or add emitters between the existing ones — a diffuser is not available, because most acrylic does not transmit below about 370 nm and would filter a 365 nm array
• Repaint the interior matt white to scatter the light
• Reduce the drive current so the emitters are less distinct

Show the answer and why

Answer: Raise the array or add emitters between the existing ones — a diffuser is not available, because most acrylic does not transmit below about 370 nm and would filter a 365 nm array

The fault is the same and the toolkit is smaller. In both instruments a ripple at the emitter pitch means the height is too small for the pitch, and in both the geometric fixes are more height or more emitters. The contact printer has a third option because an opal diffuser is cheap in visible light; the UV unit does not, since ICNIRP record that polymethyl methacrylate and polycarbonate normally do not transmit below about 370 nm. Matt white would be wrong for a different reason: this interior is matt black precisely to suppress reflected ultraviolet.

Question 4. Why must the nine uniformity strips be cut from one sheet and exposed simultaneously?

• To save paper, which is the largest consumable in the build
• Because coating and exposure variation between sheets and between runs is comparable with or larger than the field variation being measured, so one coating and one exposure removes drift, warm-up and timing from the comparison
• Because the strips must all be the same size
• Because the developer exhausts between runs

Show the answer and why

Answer: Because coating and exposure variation between sheets and between runs is comparable with or larger than the field variation being measured, so one coating and one exposure removes drift, warm-up and timing from the comparison

The economy is real and it is not the reason. What you are trying to measure is a small difference across the field, and every other source of variation — coating thickness, sensitiser age, warm-up state, timing, source drift — is potentially as large. Exposing all nine at once under one coating makes those common to every strip, so the only thing left that can differ between them is position. It is the same logic as the sensitometer's uniformity map and as the fog test's before-and-after control.

Question 5. Nichia give a junction-to-solder-point thermal resistance of 3.9 °C/W typical and a maximum junction temperature of 130 °C. What does calculating a 15 °C rise at 3.85 W actually tell you about your finished unit?

• That the heatsink is adequate, since 15 °C is well within the margin
• Only that the solder point must stay below about 115 °C — the calculation covers one link of the path and says nothing about star to pad, pad to heatsink, or heatsink to the air in your box, which are the links you built and did not characterise
• That the emitter can be run at 1400 mA safely
• That no thermal cut-out is needed at this power

Show the answer and why

Answer: Only that the solder point must stay below about 115 °C — the calculation covers one link of the path and says nothing about star to pad, pad to heatsink, or heatsink to the air in your box, which are the links you built and did not characterise

A datasheet characterises the part its maker built. Everything downstream of the solder point is your construction, undocumented, and its total resistance is easily several times the figure you looked up. That is why the calculation gives you a ceiling and the temperature log — taken with the lid shut, in the thermal environment the array actually works in — gives you your position relative to it. It is also why Nichia's 85 °C operating limit binds before the 130 °C junction limit does.

Question 6. A reader wants the unit switched by a wall timer so exposures can start without them present. What does this page permit?

• A relay module wired into the mains side of the supply inside the unit
• A plug-in timer between the socket and the certified supply's own moulded plug — no fixed wiring and nothing opened; a fixed switched circuit is work for an electrician
• Nothing: timed switching is forbidden entirely
• A mains relay triggered from the low-voltage side, since the trigger is low voltage

Show the answer and why

Answer: A plug-in timer between the socket and the certified supply's own moulded plug — no fixed wiring and nothing opened; a fixed switched circuit is work for an electrician

The rule is about what is opened, not about what triggers what. A plug-in timer is a certified appliance used as intended, between two other certified appliances, and adds no conductor to the unit. Options one and four both put a mains conductor inside a home-built enclosure. Option four also assumes a product that does not exist: Part XVII surveyed the market on 5 September 2026 and found no certified, sealed, low-voltage-triggered mains appliance switch on sale to a hobbyist, which is why the course builds no mains channel at all. Unattended running has a separate objection anyway: an enclosed array is not left running with nobody in the room.

Check my answersClear