Let there be workshop lights

When I lived in Boston, my workbench was illuminated by fluorescent tube lights mounted on a home-made frame. The frame was built specifically for the room it was in, so when I moved to New Haven I had to rebuild it. This took a year because grad school, and also because I decided that I would build something repeatable (rather than an afternoon-at-Home-Depot one-off) and not designed (too) specifically for one room.

Oh, also, I decided to learn Fusion 360 along the way, which was useful, but the learning curve didn’t help me get this done faster.

The parts finally arrived from Ponoko and Fastenal, and now I have a shiny new light. Not a figure of speech: on a bright sunny day, this light is brighter than indirect sunlight.

Choice of lights

This decision goes a few years back, and basically it comes to three factors: I wanted lights that are comparably bright to daylight, have daylight-like color spectrum, and don’t cast sharp shadows. That allows me to comfortably work at night.

There are a few different solutions to this set of constraints; none are cheap. The solution that I settled on involves residential high-output T5 fluorescent lights.

“Residential” means “ballasts aren’t noisy” and “bulbs have an option for daylight-like spectrum”, as opposed to industrial lighting, which is typically harsher and often accompanied by audible 60 Hz hum.

“High-output” means “same-size bulbs, but with more light”, which helps keep the whole setup a reasonable size; high-output bulbs generate roughly 50-80% more light than their plain counterparts. However, high-output bulbs are more expensive, and they need high-output sockets and circuitry, because they draw more power.

“T5” refers to the size of the fluorescent tube (and therefore also the housing); T5 tubes are 5/8” in diameter.

I like these lights because they last forever (I’ve had to replace one bulb in five years, because I dropped it), they are bright (the four lamps I am using output about as much light as 14 common household (60 W incandescent) bulbs), they are efficient (around 80-90 lm/W, same as household LED bulbs), and they can be daisy-chained to simplify wiring.

The cost of this lighting is around $50 for a 24” light (of which I used three) and $75 for a 48” light (of which I used one), so my setup includes $225 in lights alone. As I said: expensive. As I’ll say later, this is also an overkill and I probably could have achieved acceptable results at half the price. I sourced my lights from Pegasus Lighting.

Light frame

These lights are already pretty good at mitigating shadow problems (sharp shadows, or hands and tools casting shadows on workpiece), because they have a built-in diffuser. I wanted to make this even better by making sure that my workbench is illuminated from multiple directions, so I basically decided to make a rectangularish frame roughly the same size as my workbench, mount the lights on the bottom side of the frame, and mount the whole affair above the workbench.

I knew that I already had three 24” and one 48” lights, so I designed the frame around that. (Insert days’ worth of flailing in Fusion 360 here.)

I wanted to try laser-cutting this. Of all the makery fabrication options, 3D-printing and laser-cutting are the two that can be done online at reasonable prices.

3D-printing the whole frame was going to be too expensive. Besides, I don’t know enough about structural integrity of 3D printed materials to suspend a few pounds of lights above my head on a 3D-printed frame.

Meanwhile, Ponoko does laser-cutting at decent prices, and with excellent service.

Laser-cutting has its own problems, the biggest being that joinery is a pain because everything is flat. Most laser-cut objects I have seen are either completely flat or completely ugly, and since this light frame could be completely flat, I concluded I had a shot at not ending up with something too ugly.

Given these constraints, this turned into an interesting puzzle, which I solved like this:

I picked Ponoko’s black melamine MDF for its sturdiness and its esthetic appeal, and managed to cram the entire set of parts into a single piece of melamine to keep the material cost down:

Final cost of Ponoko fabrication was $56.30.

I decided to use binding posts and screws to attach the pieces. The principal advantage of using binding posts is that their outside surface is unthreaded, which means that I could get a tight fit between the fasteners and the MDF. I sourced the fasteners from Fastenal.

Unfortunately, tolerances of Ponoko MDF and the fasteners worked out such that I couldn’t just buy one set of screws and know they will fit. If I bought shorter screws, and then ended up getting a very thick piece of MDF, the screws would be too short for the overall thickness of MDF; if I bought longer screws, and then ended up getting a very thin piece of MDF, the screws would be too long for the binding posts.

And that’s not even considering the fact that the Home Depot corner brackets I was going to use to attach the frame to the wall did not have a specified thickness. I could have gone with a more engineering-oriented outfit than Home Depot, but then I’d be looking at more shipping costs and blah blah blah.

Since the total cost of screws of either length was about $2 for 30, I just bought two lengths of screws. I’ve never regretted too many screws in my life.


Ponoko and Fastenal orders arrived a day apart, and — as is usually the case with these things — I quickly went from being elated at everything that worked out exactly as planned to being dispirited at everything that didn’t.

One problem — which I designed for — was that the holes for the fasteners were slightly too small. This was easily fixed with a cordless drill.

This kind of a thing is hard to avoid because laser kerf is an unknown quantity. Since it depends on laser speed, and laser speed changes to accommodate different line curvatures, the only way to really deal with laser kerf is to make a test cut at the exact curvature, using the same material on the same laser, and measure the result.

Rather than ordering a test piece from Ponoko, I just designed the fastener holes on the smaller side, figuring slightly-too-small holes are much easier to fix than slightly-too-large holes.

Which they were, and after fixing the holes, the fit between them and the binding posts was tight enough to keep the frame together without any screws screwed into the binding posts.

But the real problem was that I did not account for the size of the connectors between adjacent lights. After daisy-chaining the lights on my floor, there wasn’t enough room in the corners of the frame, and the connectors were running into each other.

At this point I gave up on daisy-chaining and I connected power directly to each light. This uses one connector per light (instead of two) and makes everything fit (just barely), at the expense of turning power wiring into more of a mess-of-wires affair than I would have liked.

It also involved a bunch of wire nuts and splicing and other nonsense that took up way too much time and energy.


There was light at the end of the tunnel, though. Having widened all the holes, wrangled all the wiring, and attached everything to the wall, I now bask in gloriously diffused 11000 lm approximately 5 feet above my workbench.

My test for workbench lighting is: if I get my head as close to the workbench as I can while still able to focus my eyes, can I see what I am working on without straining my eyes? These lights easily pass the test.

In fact, these lights are arguably too bright. Having them in your peripheral vision is annoying (a concern if someone else might occupy the same space), and looking directly at them will have you squinting.

All said and done, my cost was around $350, and includes lights ($225), frame ($75 with shipping), a wall-mount switch, and a bunch of random hardware. Most of the random hardware was needed to deal with the wiring problems created by inability to daisy-chain.

You could probably obtain adequate lighting with just two bulbs, which brings the cost of lights down to $100 or $125 (depending on which two you keep). Alternately, you could use four regular bulbs (not high-output), which would cut the price of lights to around $125 and reduce total light output by 25-50%, but keep the omnidirectional nature of the light mounting.

I am not going to post the CAD files for the frame, because lights not fitting on the frame is a giant hassle, but if you are building your own lighting, you may find these useful:

And these turned out to be an excellent fastener for laser-cut pieces: