It’s been a while since I wrote a somewhat acrid blog about the 300W UV floodlight unit I bought from AliExpress. The tone was acrid, because 300W in reality turns out to be about 75W. The conclusion was somewhat counter-intuitive, as I also mentioned that I found the unit so abysmal, I planned to buy some more of them. The reason is simple: while the unit doesn’t live up to its specifications, it still gives a lot of bang for your buck, and most importantly, it’s super easy to implement. Well, you be the judge of that!
A little background: I had been using a bank of UV tubes for years for alt. process prints and also some PCB manufacturing. UV tubes back in the day when I built that array were the most attractive option in terms of UV output per Euro/Dollar etc. They also were argued to be a reasonable solution by Sandy King on his seminal unblinkingeye article on the topic of alt. process light sources. If you have the time, do give that article a read; it’s nicely done.
One of my main arguments to opt for tubes back then was the matter of efficiency. Before LEDs arrived (at least at these brightness levels), fluorescents were pretty hard to beat, especially in the moderately low power densities we’re dealing with here. But today of course we have LED, and that also solves the main problem the bank of tubes solution suffers from: the diffuse nature of the light. LEDs are for all intents and purposes point sources, and even if you create a big array of them, the light projected onto, say, a contact printing frame, remains (quasi-)collimated. So long story short, LEDs are now as far as I’m concerned the preferred solution for UV contact printing.
On a side note, I pity the people who are still stuck with their massive HID (High-Intensity Discharge) lamps for carbon printing, burning kilowatts of electricity for the same UV output that LED achieves with about an order of magnitude less input power. “Yes, but a HID lamp is nearly a perfect point source!” Well, for one, this is only critical if you’re working with halftone screen negatives, which only a few people appear to be doing. For continuous tone imaging, my experience is that even the diffuse nature of a bank of tubes is adequate and if you’re not satisfied with that, quasi-collimated light from a close-distance LED array gives excellent resolution and contrast. Thirdly, if truly collimated light is necessary, do what Calvin Grier does and gang together a couple of big COB UV LED modules and place them at a few meters from your contact printing frame. Hey presto, your artificial UV-sun is there!
I might one day upgrade to the big COB array solution, but for now, I liked to keep things simple. Cost is also a factor; last time I saw pricing information on TheWetPrint’s light sources, they were around €5k for a 1600W solution – I think I could construct something of similar (or slightly higher) power at around 10% that cost level, but that’s still around 5 times as much as a small array of those 300W LED flood lights. Anyway, simplicity is the name of the game, and here is my stupid simple UV burning solution:
The construction really is simple: the floodlights have a bracket with some pre-drilled holes, so they can be bolted directly to a baseplate, for which I used a piece of chipboard I had lying around somewhere. I also mounted some simple brackets to a shelf so I can slide the chipboard unit underneath it (as shown above) – and it can be removed as well in case I want to store it. Of course, it doesn’t use much space this way, so it’s not likely to get in the way.
I can hear you think: but how about those horrible margins between the individual LED panels? Doesn’t that create massive problems with unevenness? Well, there’s always a tradeoff between distance between the light source and the print and hence light intensity (or exposure time) and evenness. The further away from the light source you get, the more the light evens out and the larger the evenly illuminated surface becomes, but the lower the light intensity and therefore the longer the exposures. With a light source like this, evidently if you creep up very close to it, the dark area between the panels will be problematic. But at the same time, the LEDs have a pretty wide angle; typically around 100 degrees or so. This means that the evenness problems drop away quickly as you move away from the light source – but so does the light intensity.
Now, the size of the individual panels are a bit larger than 4×5″, so for a 4×5″ print, I can just position the printing frame close to any of the four panels. As long as I align the print area exactly beneath the covered area of the light source, I’m good. I actually did all of the initial exposures with these flood lights that way, with a distance of about 11cm (~4.5″) between the contact frame and the light source. For 8×10″, this wouldn’t work, so I did a little test with Van Dyke Brown. I really like Van Dyke’s for alt. process light source testing because it’s a pretty fast process in terms of making a print (it doesn’t take much preparation) and it’s also pretty high contrast, so evenness problems will show up easily. Here’s how an 8×10″ Van Dyke print (without a negative) exposed to a dull grey came out, with the frame positioned around 20cm (~8″) from the 4-panel light source:
The photo of the print is a little deceptive as it seems to exhibit a lighter area in the center, but in the real print, this is not present, so it’s some kind of smartphone camera or lighting artefact. In any case, the gaps between the four floodlight panels don’t show up in any way on the print, which is great news. At a distance of 20cm from the light source, exposure times are still fairly short. I experimentally determined the difference in exposure time between the 11cm distance I used for initial 4×5″ tests and the 20cm distance I’m using now because it’s more convenient. The printing frame simply sits on the table below the light source, so I don’t need any means to raise it up to a certain height. Once again, I did some Van Dykes to determine the light intensity difference between both distances, culminating into this comparison print:
The difference worked out as slightly more than a stop for the longer distance. Note that this ‘breaks’ the inverse square law which would imply nearly a two-stop difference when going from 11cm to 20cm (nearly doubling the distance). However, the nature of the light source, which is halfway between diffusion (radiating outward in all directions) and collimation (parallel light waves) results in the inverse square law not holding. That’s why this kind of testing is necessary when changing distances in exposure systems like this one.
I also did a comparison between the LED light source and the fluorescent tube array I used before. I did mess this up in two ways: I didn’t properly fix the Van Dyke Brown print, so it’s horribly fogged now. I generally evaluate these when they’re fresh, so I don’t spend much attention to properly processing these kinds of prints, at least not the fix and wash stages. Moreover, the distance between the light source and the print wasn’t exactly the same, but still fairly close. All considered, the result is indicative (although not dead-accurate) of the difference in exposure strength between both light sources:
The result indicates two things: firstly, the contrast is roughly the same. The fluorescent tubes throw a pretty broad UV spectrum that trends towards a smaller wavelength than the LEDs; the fluorescents peak around 370nm or so if memory serves, while the LEDs used here are a narrow-band 395nm type. Apparently, at least for Van Dyke Brown, the difference isn’t particularly meaningful. This may be different for other processes. Secondly, the fluorescent bank (rated at around 200W RMS power) is significantly slower than the LED unit, with the difference being close to two stops by the looks of it. The LED unit is around 300W RMS, so the LEDs do seem to have a slight edge in terms of efficiency.
Note that the difference in efficiency between LED and fluorescent for general purpose (white) lighting is often cited as LED being 40% ~90% more efficient (e.g. here and here). At least in this application, it seems to be on the lower end of that bandwidth. And if I’m brutally honest, I think the higher figures (closer to 80%) are exaggerations that purposefully compare favorable LED technology with unfavorable fluorescents (e.g. aged lamps, inefficient ballasts etc.) Still, LEDs have the edge in general and also in this particular case.
As said before, the construction of this unit is very simple. As I’m currently working on my color carbon project, my focus isn’t really on building complex equipment. I wanted something I could do quick & dirty and just get on with the job at hand, and this setup fits the bill. Here it is, in all its basic glory:
At the top is the chipboard baseplate and the individual flood lights are held in place with screws and some tie wraps to keep them strapped together, which adds to the rigidity of the setup – otherwise the individual units have the tendency to tilt and turn a little when they’re bumped into etc. Each unit came pre-wired with a switch and a plug, so I just cut the cords on each of them and then tied them together using two Wago 221-415 connectors so that they use a single plug and switch. Since the total setup only draws around 300W, the switch and plug are used well within their maximum ratings, as is the cable. No problems there. Sure, it’s not going to win any prizes for aesthetics or electrical safety, but there are no real-world problems here.
Well, stuff does get hot, I have to admit. I might add a fan at a later stage, but so far, I’m mostly running 6-7 minute exposures (that’s what my color carbon requires; 6m30s to be precise) and four of those immediately after another don’t give any trouble. That’s probably also the longest I’ll use this light source at a time, since I prefer to keep my carbon exposures on the short side, so within 30 minutes. I’m way too impatient for longer exposures, and generally I don’t consider them necessary at all. While the units do heat up considerably, there are no actual problems in this area at this point, so for now I’ll accept it as it is.
Above you can see the floodlight exposure unit in its current place, which is under an existing shelf and right above the fluorescent tube array. The latter is mounted underneath a collapsible table that can be folded downward against the wall to make space. It also serves as a table top, which is why I installed it at the exact same height as the worktop to the right, which spans the width of the room. It’s convenient to have large, more or less continuous work surfaces. There’s also a little black breadbox, which is an experimental 10W UV COB LED exposure unit I made for PCB work. It works very well for that purpose, but it’s too small and slow for any serious alt. process printing. It is nice and compact though, and it’s more or less a point source, so it prints very sharply.
Future steps will include replacing the manual switch on the new LED unit with something a little more convenient. My other exposure units both feature a 7-segment display and a rotary encoder allowing the exposure time to be set in seconds. While that works OK, I’d like to add another feature: an integration sensor. This will essentially be a UV sensor hooked up to a microcontroller that records the light intensity periodically and sums up / integrates the total exposure. The advantage of this is that I can simply determine a suitable exposure time using one particular UV exposure unit and one particular distance, and then make identical exposures at other distances or even using different equipment. Well, in theory that should work; practice will be a little more challenging because there’s going to be some challenging regarding UV spectra and the spectral response of the sensor used, but I’ll probably go for a simple approach with some compromises in this area to keep it feasible.
So much for now for the UV exposure stuff! As I mentioned, this was something I encountered on the way in my carbon endeavors, so it’s really a sideshow to the actual printmaking project. For that purpose, it seems to work well enough and it didn’t cost more than an hour or so to assemble, so it didn’t sidetrack me too much. Onwards, to glory! (Or massive failure, of course.)