It's almost a guarantee that if you haven't yet incorporated the ubiquitous WS2812x 'NeoPixel' or APA10x 'DotStar'-style "digital" LEDs into a project, you will no doubt have seen the myriad works of many a maker, ranging from blinking, bedazzling digital jewelry to drive-in scale digital LED video walls.
No matter the form factor, getting these bare LED emitters to look their best can sometimes take quite a bit of work! A range of factors, from ensuring that you have provided an adequate power supply to diffusion of the light emitted from these devices — and everything in between! — means that there is usually a direct trade-off between the amount of thought that goes into the incorporation of these LEDs, and how fantastic the final product looks!
Debra Ansell (AKA @GeekMomProjects) has been documenting her extensive experience with these "smart" LEDs for some time now, over on her Twitter feed, and her latest innovation has oodles of potential written all over it!
Through a clever combination of lateral thinking, and considerations of design for manufacture, she's arrived at a really quite neatly simplified way to construct arbitrary sized digital LED matrices that are cost-effective, and look great!
With a framework formed of laser-cut plastic, we can see from the images below, that where conventional thought would have most of us place our pixels at the central point of a cell within such a matrix, Ansell has turned that thought on its side, and has instead chosen to run strips of LEDs down each row, but instead facing upwards, along the columns of the matrix.
With yet more laser-cut plastic, this time some translucent circles notched to align with the LEDs, the intense, almost sometimes "too much" output of the pixels is evenly diffused, to a far more pleasing, pastel-toned palate, aptly smoothing out those those nasty hotspots that can otherwise leave your eyes trying to look away!
It's quite a clever design, for a number of reasons beyond the aforementioned cost and ease of manufacture. Why do I say this? Well...
(Disclaimer: Project infact does contain a small number of wires.)
Arranged in a serpentine fashion (that's zig-zag if you're using a number of the LED matrix libraries), the only wiring needed in this build is the single link between the output of one row, and the input of the next, along with the +V / 0V power connections to each row.
While there isn't much of note in soldering some wires between points, what is to be noted is the way in which Ansell has implemented and routed the power connections within in the framework itself, opting to use adhesive backed copper tape in place of conventional wiring. That's a neat trick!
Not only does it save the time and prep of soldering discrete wiring, it's low-profile, and needs no additional mechanical fixing — the self adhesive backing means it just sticks where it's needed
While maybe not scalable to the largestof displays, where current draw can reach many tens (even hundreds) of amperes, the logic of using what is effectively the same material used within the strips themselves makes a great deal of sense — even if it wouldn't occur to many of us at first to do so!
Some people might not be aware that these LED strips are also commonly available in varying package sizes, and more importantly, pixel pitch.
As this entire design can be designed as a parametric model, requiring only the pixel pitch, and width of a strip in order to calculate the dimensions of the framework, it means that the matrix can be easily scaled to whatever strip you can lay your hands on at the time!
In fact, it's projects such as this one that would nicely lend themselves to being written up as a fairly simple OpenSCAD script, requiring a user only to input the two values discussed for their strip, plus the desired pixel count in both X and Y dimensions. Hey presto, collect your DFX files and head to the laser cutter!
Another feature of the tape that comes in quite handy in this design is that it is pretty much solid coper tape — meaning that there is no chance of any light bleeding between the rows.
(I'll apologize now for my terrible drawing annotations. It's not the fault of the stylus, I'm just not what they call "artistic." At all.Sorry. I'm working on it...)
With some attempt from the author to show the typical sort of angles of emission that you could expect from these 5050-style packaged, un-lensed parts (give or take 10-20 degrees...) — I hope I have managed to illustrate why this design works well without needing light barriers between adjacent columns — most of the light is emitted straight out of the front face of the package.
Admittedly, these parts are not known for collimation of the light emitted from the three separate RGB die within the package.
With correct diffusion applied, as we've seen here, I think we can all agree that despite the mechanical simplicity, the results are stunning! This is a brilliant use of material properties, and truly showcases what can be done with a lot of LEDs and a little lateral thinking!