We’ve seen a couple of UV lamp builds for exposing photosensitive PCBs and erasing EPROMs, but [John] over at pcboard.ca decided if it’s worth doing, it’s worth overdoing. They designed a UV exposure board using twelve 1 Watt UV LEDs, an impressive amount ultraviolet light that you probably shouldn’t look at for too long.
We’ve seen UV exposure boxes before, usually made with a bunch of 5mm UV LEDs soldered to a piece of protoboard. These projects do their job, but the exposure time is on the order of minutes. The PCboard.ca UV lamp can expose a PCB in just 20 seconds.
The build began with four pieces of aluminum bar, 1 inch wide and 1/8″ thick. The 12 star LEDs were glued down to this bar with thermal adhesive and serve their purpose as a rather large heat sink.
[John] performed a little test to determine how long it would take this monstrous UV source to expose a PCB. By copying a PCB mask four times and placing it over an unexposed board, [John] made a PCB with exposure times of 60, 45, 30, and 15 seconds. After developing and etching, all but the 15-second exposure was fully etched, an amazing result that will probably lead to some very, very rapid prototyping.
All the more impressive is the fact that only four 1-watt LED drivers were used for this build. That’s right, this UV lamp is actually operating at about a quarter of its maximum rating, or about 285mA per LED. We’d hate to see this thing operate at full power, protective eyewear or not.
33 thoughts on “Do Not Look Into 12 Watt UV Lamp With Remaining Eye”
If the led is rated at 1 watt, 285mA is not “a quarter” of it’s maximum continuous current. This kind of led usually have a forward voltage between 3 and 4V .. that gives, for 1W, something like 300mA of continuous current.
Indeed. The article links to both the LED modules used and the led driver boards.
The led drivers will keep a constant current of 285ma through several leds hooked in series.
The leds used suggest a forward voltage of 3.2v and 350ma current.
The led driver manual does say that the minimum input voltage should be 16V when power 4 leds like in this configuration, but the article says they are using a 12v source.
They may be able to get more brightness from these led modules by going to a 16v source, only running 3 modules per driver board on a 12v source and/or using 350ma led driver boards to match their led modules.
But the claim of only running at 1/4 output now seems dubious.
You are right, mostly. A constant current driver will drive the entire string of leds with each led getting roughly the same current. If you wanted to run them at 1/4 the power you would have to have them in parallel, but even that wouldn’t work since one led is going to have a slightly lower voltage drop than the rest so that LED will draw more current. That means you have an uneven light source.
Raising the voltage should get a touch more light. The drivers usually want a few volts over the combined voltage drop of the LED string.
There are 3 leds in series, not 4.
“with each led getting roughly the same current.”
They’re getting exactly the same current due to laws of physics.
in order to get even illumination a hexagonal pattern is much better. Everytime someone stacks LEDs like this a kitten dies or something.
It is true, though they are easier to stack this way (heatsinking purposes).
But maybe not by much. If it doesn’t have to be a leveled plane, one could layer the aluminium bars over and under each other. Creating a hexagonal.
The LEDs are connected up in Four parallel strings, with each string of three connected up in series – so the driver at 12v is correct.
P=E*I – Ohm’s law. People who claim to be technologists who don’t know this are hatfarts.
.285 amps times 3.2 volts = .912 watts.
Would your staff writer like to explain to me how .912 watts is 1/4 of 1.00 watts? Inquiring minds want to know.
The author of the article clearly either didn’t read the original information, or doesn’t understand the concept of constant current LED drivers. He saw a “1W LED driver” powering 3 LEDs at just under their rated current.
(0.285/3)/0.35 ~ 0.27 ~ 1/4 <– faulty data, correct math.
Sorry but they don’t get off the hook that easily. Look at the last paragraph:
Ignorance on full display here. The forward voltage drop on these LEDs is 3.2 volts, not 1 volt.
Even the boner is a boner. 4 watts is NOT 1/4 of 12 watts. Take some remedial math, guys!
Hackaday staff are not noted for the technical accuracy of their posts.
That’s not to detract from their entertainment value as a hacker aggregation site.
Perhaps a typo? I read that as about 285mW per LED.
Math aside, this is pretty cool. Who wouldn’t like a board in 30 seconds?
Using cheap 5mm leds I’m able to expose in 2min flat. I guess that’s just too long for you impatient types?
The nice thing about an array of smaller LEDs is that it gives a more even exposure. You could diffuse the light but then you’re losing the best advantage of using LEDs – the narrow viewing angle minimizes light getting to the board at angles which might under-cut the mask.
Nice LED tanning bed. :D
Aside from my above snarky comment, I think the idea of using the photoresist process is clearly better than the toner transfer methods for these reasons.
1. The pattern is printed on clear acetate or mylar and can be touched up or have opaque foil laminated to its surface. And the sheet is then available for making multiple boards.
2. Registration for double sided boards is MUCH easier than with toner transfer because the index marks can be placed directly over reference holes in the the PCB.
3. There is no concern about getting temperatures right. If a laminator is too hot the toner can spread and smear. This results in grief when using fine pitch parts.
4. By the time the cost of PulsarFX paper is taken into account the cost of sensitized boards is very competitive with normal board material. The mylar sheets are less expensive than the special toner paper and MG only charges pennies more for photosensitized board.
I agree with the above commenter that a better light consistency is achieved by putting six lamps in a circular pattern and then having a single lamp at the center.
The heat sink business seems pretty unneeded – the lamps come mounted on a heat sink and they are only on for a few seconds per exposure. I’d just add a timer circuit to keep me from forgetting to turn the lamps off.
A 1W LED of that type will get stupidly hot in 30 seconds, when it’s that hot the light output will go down, as will the expected service life of the LED.
The aluminium star is not a heat sink, it’s a heat spreader that also makes it easier to solder wires to the LED.
This from adafruit website
There is no mention of the need for an additional heat sink.
Well, they’re lying. Or they don’t know it needs extra cooling.
I wouldn’t call Adafruit a reliable source of knowledge about thermal design.
People with a knowledge of “thermal design” generally can do better than describe the temperature as “stupidly hot.” Generally they speak in terms of degrees.
Well, since a “1W” LED is actually a “350mA” LED and can vary from 1.8V for red to about 4.3V for some UV, it is impossible to name a temperature or even a power value.
Also the ambient temperature, mounting orientation, if it’s in a confined space and what it’s mounted to makes large differences in die temperature.
if you read the adafruit project:
Step 6. Mount LED onto ‘heat sink’
Next up, we use JB Weld to attach the LED onto the lightbulb mount, this keeps the LED from moving around and also gives it a nice big heat sink! The LED gets extremely warm otherwise and could be damaged.
also from the product description:
You can also try driving it directly with a 3.3V power supply, that will probably work OK as well (although its not ideal).
Doesn’t sound like someone I would trust with thermal design
You guys can complain about the math all you want.
I’ll… ahem… lighten this up a little.
My girlfriend just YELLED at me for showing her the picture.
She read that you should look into it for too long because it could damage your eyes…..
Good lord! hahaha!!! Time to find a new girlfriend I think.
*SHOULDN’T look into it for too long.
She seriously thought she was going to go blind from a .jpg of UV light.
While she’s at work, I’m going to set it as her desktop wallpaper. :P
Hehe that would be awesome though, cure your photoresist with an old laptop display and a UV light picture :)
Hey, at least she cares. There *is* more to girlfriends than just technical aptitude…
Quick, send a dozen of these to Deneva! (If your a Star Trek fan your laughing your ass off right now.)
I am a reefer (saltwater) and thought I should mention these kits since they would save you some time. http://reefledlights.com/shop/uv-violet-light-accent-bar-lab-kit/
I used the white one for a photobooth that needed more lighting.
@Jod, you owe me a new keyboard.
Seriously, she actually said that?
Re. PCBs, what about using a B/W
or small grid type LCD ie 3310
then projecting a single UV LED
through it onto the board?
Tried that with a picture frame
but it needs to be B/W due to the
I still think it’s easier just to use fluorescent tubes from those insect zappers. Two tubes, a ballast and a 555 starter circuit cost about the same as two or three 1W LEDs.
Please be kind and respectful to help make the comments section excellent. (Comment Policy)