<HOME
<Test equipment
<Electronics
<Workshop stuff

My home made UV light box for exposing photosensitive PCBs

While I mostly build my one-off electronic projects on Veroboard or similar, proper PCBs are sometimes necessary, and not all that hard to make at home—especially single-sided ones. After drawing the layout, I find the most tedious chore is drilling holes for through-hole components—and even that can be avoided by using surface-mount devices.
I use ready-made photosensitive PCB material, which I expose through two laser-printed transparencies (you know, those used with an overhead projector—or are you too young ever to have used one?) carefully aligned and taped together (a single print will have too many holes in it). Over them I place a sheet of glass (I have somewhere an old semiconductor photomask stripped clean of its emulsion, but ordinary glass taken from a photo frame does work just as well, and is more readily available) to press the transparencies against the PCB. I expose them, develop with suitably diluted drain opener (no need to buy pure NaOH pellets, which are hugely more expensive), etch with sodium persulfate (or hydrochloric acid with hydrogen peroxide, or diluted nitric acid, or whatever I have available), and strip with acetone. When using trough-hole components, I drill the board with my Porxxon mini drill which is mounted on a drill stand. To make alignment easier, I've added a USB microscope to the drill stand, With an adjustable crosshair overlay I can see exactly where the drill bit will hit the PCB. Sometimes I will even tin the copper areas.
Last time I needed a custom PCB, I finally went ahead and built myself a proper UV exposure tool, as described here. The PCB fabrication process is described further down.

I used to do the exposure with an ordinary fluorescent tube desk lamp, which produces enough UV (or almost-UV) to expose a small board in some 20–30 minutes at close distance (larger boards need the light to be further away to illuminate the whole board, or moved regularly along the length of the board, so the time is correspondingly longer). That was convenient enough, as the lamp was right there whenever needed, and I rarely was in any hurry—but you can't get the environmentally disasterous mercury-laden tubes anymore (regardless how "green" CFLs were touted as, while transitioning away from incandescents)!
But, surprisingly, I found you can still get compact fluorescent "blacklight" bulbs for your home discotheque. (Well, you could way back in 2024 anyway. And I'm not sure, perhaps these also have been outlawed, but the store was allowed to sell its remaining stock? So if you can't get them either, you'll be looking for an LED-based solution.) So I bought one of those, and built a dedicated, "real" UV exposure box to house it in. With my extremely infrequent use, the bulb should last forever, plus I bought a spare one as well. (And while laser-printable transparencies are still available, I also stocked up on those! Expect them to finally disappear from stores any day now!)

Instructions on the Net say you need a cardboard box and a suitable lamp holder and power cord. Hang the UV bulb in the box, place the PCB underneath, and expose away!
Awful.
I mean, it will probably work just fine, but when doing anything, why not overdo it? Why not go absolutely overboard with the design? I happened to have some thin aluminum sheet metal in-shelf to make a reflector of sorts, so as I I bought the UV bulb, on a whim I picked up a piece of plywood as well (actually an extra shelf for a cabinet of some kind... I've later used three of the exact same shelves to build a dust containment tub for my Proxxon mini drill setup). In one evening, I built a much nicer exposure box. Good and sturdy, no wobbly cardboard, it sits properly above the PCB and will certainly give more uniform and consistent exposures!

Here is a photo of the box upside down, looking into the reflector. The two plywood pieces on the left and right have grooves milled into them, which hold the aluminum reflector in place (and in shape). The left-hand one also has a hole for the lamp holder, roughly at the reflector's focus. Some extra support is provided by six bits of square(ish) wooden dowel, three on either side of the plywood, glued on with epoxy (you can see the inside three surrounding the base of the bulb in the photo). I didn't bother attaching the lamp holder any more permanently, so it will just slide out if the bulb is first removed.
The design process was very scientific: I plotted a couple of parabolae, and eyeballed one of suitable dimensions. :) The manufacturing process was highly technical as well: I printed out the parabola, cut the paper along the curve, and traced its edge with a pen onto the two pieces of plywood. (The paper stencil is in the photo on the right.) Using a 2 mm wide milling bit with my Proxxon IBS/E in its MB-200 stand, I hand-fed the plywood along the curve and managed to cut reasonable grooves some 6 mm deep into the 10 mm plywood. I bent the aluminum into roughly the same shape, and had little trouble fitting it into the grooves. The two other sides of the box (upper and lower in the photo) came from the same scrap of plywood, and attached with wood screws to the two others. (Sorry I didn't think of taking photos during assembly, but you can probably see what I've done from the finished product.)

This is how the box stands on the desk. The appropriately masked PCB goes underneath it. The bulb is some 13 cm above the surface of the desk, and there are slots at the bottom of the box and at the sides of the reflector to allow air circulation. (Actually, the bottom slots were dictated by the size of the leftover plywood, after I had cut the first two pieces to hold the reflector.) Whereas the 15 W bulb heats the aluminum reflector to somewhat over 40°C, the PCB being exposed remains pretty much at room temperature.
The standard E27 lamp holder protrudes from the left-hand side of the box, and has an on/off switch in its cord. You can see the three support pieces placed around it, just like inside the box, in the photo. (The "extra" piece of square dowel along the lower edge of the left-hand side has no purpose here—it was part of the shelf I sourced the plywood from, and I just didn't bother to remove it. )

The bulb's actual tubes are about 80 mm long, and at the distance that they are, the 100 mm wide PCB ought to be pretty uniformly illuminated along its entire width. With the reflector, the illumination seems visually very uniform in the perpendicular direction as well; I'd be surprised if it couldn't expose the whole 160 mm long PCB in one go. (Not that I fabricate such large PCBs very often. The first PCB I used this device on was just 20×30 mm—though I did make four of them at once, so 80×30 mm then. But on the few instances that I have exposed a whole 160×100 mm using a desk lamp, I've shifted the lamp along the board's 160 mm length several times during quite a long exposure.) Here's a photo of the box on its side (with the UV bulb vertical), with a blank paper fluorescing in front of it. The rectangle drawn on the paper is the size of a 100×160 mm PCB. (And here is a professional-grade schematic of where the ultraviolet CFL bulb and its holder are behind the paper, and what their relative sizes are.) Not bad, huh? According to Gimp's color picker tool, the bright central parts seem to be only some 20% brighter (on a linear scale) than the darkest corners of the rectangle area. The required exposure dose for the PCB is certainly less critical than that!

The process

I'm using up my >10 years old PCBs first (have I really gone that long using just Veroboard???). They still work sorta-kinda ok, only their peel-off backing leaves some black goop around the edges of the board—that adhesive seems to degrade faster than the photoresist! However, fresh boards will likely do with a much shorter exposure time. I'll update that here once I get around to it, but here's my recipe for these long ago expired boards:

Expose 10 minutes with the 15 W bulb. In an exposure time test, 4 minutes already looked very good, but 8 minutes resulted in just perceptibly less residues. In another test, there was no further difference between 8 minutes and 16 minutes. I tried etching the 4 min and 8 min test PCBs, and the 4 min was no good, but the 8 min was pretty ok, considering the age of the PCB material! So I settled on 10 min exposure time, just to play it safe. If I were trying to make ultimately narrow traces, where exposure time might be more critical, I'd certainly use a fresh PCB, not one with a decade old expiry date! :)
Develop 1+ minute with agitation in domestic Pirkka Putkenavaaja drain opener solution, diluted with water to 1/10 of its original concentration. I think Pirkka is at the milder end of the NaOH concentration range, judging by its density; other brands of drain opener will likely need to be diluted more. About 1% NaOH by weight should be a reasonable concentration to aim for, if the concentration happens to be specified on the bottle.

Pro tip: Underexposure can rarely be adequately compensated with overdevelopment, so it's better to err on the side of overexposure, and to develop only as long as visually seems necessary! The same goes for overetching: Unless there's serious photoresist residues remaining on the PCB due to underexposure, underdevelopment or over-aged boards, the etching time should be quite independent of the above. But any residues that do hinder etching are unlikely to be helped much by overetching the board. Instead, the traces will just get narrower and narrower.

Etch 10+ minutes with agitation in a solution of 20 g sodium persulfate for every 100 ml of water. Use as hot water as you can get from your faucet (usually that's about 60°C). The etch rate will go down as the solution cools(*), so especially with small etchant volumes (for small PCBs), you may need to carefully heat up the solution in a microwave halfway through. (Take the PCB out first, and don't go anywhere near boiling!)
Once the board begins to look good, overetch it for a while to make sure you leave no short circuits anywhere. Then rinse it off and inspect it under a magnifier, and etch some more if necessary.
Unlike above, where I tested four different exposure times, this is a "production batch" of four PCBs at once. (There's no sense in making only a single one of such a tiny board—the saving is minuscle, and either I'll mess up the first prototype and need another PCB, or I'll find use for another one later, or someone else will become interested and request a PCB!) The edges of the traces are all ragged and "fuzzy" with these over-aged PCBs, so for anything more demanding than a SOIC (at 1.27 mm pitch) I'd absolutely use a fresh board instead!

(*) A chemists' rule of thumb says the reaction rate will approximately halve for every 10°C reduction in temperature. So if etching takes 10 minutes at 60°C, expect to etch almost three hours at room temperature!

The reaction between the persulfate ion and copper is:
Cu + S₂O₄^2- → Cu²⁺ + 2 SO₄^2-
and from their molar masses you can calculate that a gram of sodium persulfate will etch 0.27 g of copper, or 0.030 cm³, which is about 8.5 cm² surface area on a single-sided PCB with the typical 35 μm copper. 20 g of sodium persulfate in 100 ml of water, per the above recipe, should therefore be just sufficient to etch an entire 160×100 mm PCB completely, especially if you make copper fills in any large unpopulated areas, rather than etching them completely free of copper. However, 100 ml is a dismally small volume of liquid to spread over a PCB that size. The liquid would only be some 6 mm deep—and even shallower, if you etch in a pan any bigger than the PCB! So go ahead and mix up a sufficient amount of solution, some 300–400 ml at least for a PCB that size, and then you can also rest easy knowing the etchant won't run out of oomph halfway through the copper. After all, the etch rate will also decrease with decreasing etchant concentration, not just with decreasing temperature!

After etching, wipe the board free of the photoresist, first with acetone, then with isopropyl alcohol. Then, if you're using any through-hole components, drill the board.
Finally, if you like, you can tin the board by applying Bera-Fix solder tinning compound with a cotton bud, and melting it with a heat gun. (Or you can spread flux over the board, and trace all copper areas with a bead of molten solder on the tip of your soldering iron.) Then scrub away the flux residues and loose solder beads with acetone again, and finally with isopropanol. Inspect the result carefully—there's still solder beads and residues visible in the photo, demanding more scrubbing! The solder plating also looks needlessly thick, I guess I was a bit too liberal with the Bera-Fix this time.
The PCB is for a two-channel balanced to unbalanced audio converter, which comprises a dual op-amp and a bunch of resistors and capacitors. I made it in SMD only because the end application didn't have much room inside. (Ok, I made it in SMD because I was too lazy to drill holes. Happy now?) It went into a mixer whose AUX RETURN inputs were, for some unfathomable reason, unbalanced!

Radiation, electrical and chemical hazard warning!

Please be aware that UV bulbs, though they may be sold even for recreational use at home, may present a significant hazard to eyesight! Do not stare into the bulb under any circumstances, and do not assume that the bulb will only emit the supposedly less-harmful "UV-A" radiation! There are numerous documented instances of cheap Chinese UV bulbs blasting out significant amounts of the more harmful UV-B (or even UV-C?), with party guests suffering eye or even skin problems the following day! Put the bulb in a box where you can't at least see it directly, and do avoid staring into direct reflections as well!

Also, don't make the household AC-powered lighting fixture yourself, if you're at all uncertain of how to do it properly! I'm sure you can scavenge a bulb holder, cord and switch already wired up and ready to go from some cheap ready-made lighting appliance at your local hardware or home furnishing store such as Ikea.

Finally, wear appropriate hand and eye protection when handling the chemicals for developing, etching and stripping your PCBs! Drain opener (sodium hydroxide, a.k.a. lye) is especially hazardous for your eyes—a single drop can actually blind you! (It's astonishing that the stuff is sold in supermarkets, yet half our population isn't blind already!) And your skin will absolutely not appreciate the highly oxidizing and/or acidic copper etchants, nor the acetone, for that matter (despite it being commonly used as nail polish remover)! These substances will also have cumulative effects, just in case the first contact doesn't kill you sufficiently. And dispose of used solutions appropriately—dissolved copper is a toxic heavy metal, you know, right? Right??? You see, I'm actually a chemist by training, but I foresee a greater than 50% probability that you're not! If nothing else, at least wear goggles, for fsck's sake!!!

So do this at your own risk, and don't blame me for anything bad that happens!

Antti J. Niskanen <uuki@iki.fi>

	I used to do the exposure with an ordinary fluorescent tube desk lamp, which produces enough UV (or almost-UV) to expose a small board in some 20–30 minutes at close distance (larger boards need the light to be further away to illuminate the whole board, or moved regularly along the length of the board, so the time is correspondingly longer). That was convenient enough, as the lamp was right there whenever needed, and I rarely was in any hurry—but you can't get the environmentally disasterous mercury-laden tubes anymore (regardless how "green" CFLs were touted as, while transitioning away from incandescents)! But, surprisingly, I found you can still get compact fluorescent "blacklight" bulbs for your home discotheque. (Well, you could way back in 2024 anyway. And I'm not sure, perhaps these also have been outlawed, but the store was allowed to sell its remaining stock? So if you can't get them either, you'll be looking for an LED-based solution.) So I bought one of those, and built a dedicated, "real" UV exposure box to house it in. With my extremely infrequent use, the bulb should last forever, plus I bought a spare one as well. (And while laser-printable transparencies are still available, I also stocked up on those! Expect them to finally disappear from stores any day now!)
	Instructions on the Net say you need a cardboard box and a suitable lamp holder and power cord. Hang the UV bulb in the box, place the PCB underneath, and expose away! Awful. I mean, it will probably work just fine, but when doing anything, why not overdo it? Why not go absolutely overboard with the design? I happened to have some thin aluminum sheet metal in-shelf to make a reflector of sorts, so as I I bought the UV bulb, on a whim I picked up a piece of plywood as well (actually an extra shelf for a cabinet of some kind... I've later used three of the exact same shelves to build a dust containment tub for my Proxxon mini drill setup). In one evening, I built a much nicer exposure box. Good and sturdy, no wobbly cardboard, it sits properly above the PCB and will certainly give more uniform and consistent exposures!
	Here is a photo of the box upside down, looking into the reflector. The two plywood pieces on the left and right have grooves milled into them, which hold the aluminum reflector in place (and in shape). The left-hand one also has a hole for the lamp holder, roughly at the reflector's focus. Some extra support is provided by six bits of square(ish) wooden dowel, three on either side of the plywood, glued on with epoxy (you can see the inside three surrounding the base of the bulb in the photo). I didn't bother attaching the lamp holder any more permanently, so it will just slide out if the bulb is first removed. The design process was very scientific: I plotted a couple of parabolae, and eyeballed one of suitable dimensions. :) The manufacturing process was highly technical as well: I printed out the parabola, cut the paper along the curve, and traced its edge with a pen onto the two pieces of plywood. (The paper stencil is in the photo on the right.) Using a 2 mm wide milling bit with my Proxxon IBS/E in its MB-200 stand, I hand-fed the plywood along the curve and managed to cut reasonable grooves some 6 mm deep into the 10 mm plywood. I bent the aluminum into roughly the same shape, and had little trouble fitting it into the grooves. The two other sides of the box (upper and lower in the photo) came from the same scrap of plywood, and attached with wood screws to the two others. (Sorry I didn't think of taking photos during assembly, but you can probably see what I've done from the finished product.)
	This is how the box stands on the desk. The appropriately masked PCB goes underneath it. The bulb is some 13 cm above the surface of the desk, and there are slots at the bottom of the box and at the sides of the reflector to allow air circulation. (Actually, the bottom slots were dictated by the size of the leftover plywood, after I had cut the first two pieces to hold the reflector.) Whereas the 15 W bulb heats the aluminum reflector to somewhat over 40°C, the PCB being exposed remains pretty much at room temperature. The standard E27 lamp holder protrudes from the left-hand side of the box, and has an on/off switch in its cord. You can see the three support pieces placed around it, just like inside the box, in the photo. (The "extra" piece of square dowel along the lower edge of the left-hand side has no purpose here—it was part of the shelf I sourced the plywood from, and I just didn't bother to remove it. )
	The bulb's actual tubes are about 80 mm long, and at the distance that they are, the 100 mm wide PCB ought to be pretty uniformly illuminated along its entire width. With the reflector, the illumination seems visually very uniform in the perpendicular direction as well; I'd be surprised if it couldn't expose the whole 160 mm long PCB in one go. (Not that I fabricate such large PCBs very often. The first PCB I used this device on was just 20×30 mm—though I did make four of them at once, so 80×30 mm then. But on the few instances that I have exposed a whole 160×100 mm using a desk lamp, I've shifted the lamp along the board's 160 mm length several times during quite a long exposure.) Here's a photo of the box on its side (with the UV bulb vertical), with a blank paper fluorescing in front of it. The rectangle drawn on the paper is the size of a 100×160 mm PCB. (And here is a professional-grade schematic of where the ultraviolet CFL bulb and its holder are behind the paper, and what their relative sizes are.) Not bad, huh? According to Gimp's color picker tool, the bright central parts seem to be only some 20% brighter (on a linear scale) than the darkest corners of the rectangle area. The required exposure dose for the PCB is certainly less critical than that!