The neighbors let us know, without mincing many words, that there will be Trouble if we don’t set the up the giant yard tree for Christmas again this year. We’ve been told that kids living on the opposite corner of the development would howl in protest if they weren’t driven home – the long way – via our little corner.
Fortunately, last year’s anchoring system proved so strong and sturdy that I plan to extend the tree to its full 25′ height. This means that there won’t be any extra node string ends on the ground this year, and the entire tree will be about 6′ taller.
But it’d be boring not to add something new to the collection.
For a few years, we’ve wanted to decorate the house with snowflakes of varying size and shape. Sadly, the daydreams have so far not materialized. Certainly there are a wide variety of pre-fab wireframe snowflakes available for sale. But I wasn’t terribly interested in the overall look.
A few years back, when we first started working with RGB LEDs overmolded on a cable, I drew up some simple snowflakes in the CAD program, then sent them to the shop which does all the metal forming for the products we sell during the day. A 48″ x 96″ sheet of .06″ aluminum, fully tessellated with snowflakes, cost about $400 for materials and laser time. This, by the way, was the after-hours cash price, so who knows the actual street price of the silly things. In this business, it’s often best not too ask.
(Note that all pictures below can be enlarged to full size by clicking on them twice.)
Advantages: Discrete RGB is cool, but there’s already almost 4000 nodes on the big tree.
Disadvantages: The LED package is nontrivially large. Wiring harness is bulky and adds about 3″ of depth to each snowflake. The wiring spaghetti backstage is ugly. Installing those nodes in aluminum holes is absolute murder on the fingers. For covering the whole house in snowflakes, the installation must be immaculate.
Plus, I didn’t just want bits of light arranged in a pattern. I wanted a look which was seamless, fluid and beautiful. So for 2012, the design constraints are
- No nasty wiring harnesses
- Low profile
- Continuous color / shape
- Ability to appear nearly invisible during the day
- (Optional) color changing capability
- (Optional) relatively decent pricing, as quantity will approach 60-100 pieces in varying size.
So without further ado, here’s a summary of thoughts, experiments and misfires which have percolated through over the past few months.
Hypothesis #1 – Backlit Plastic
During a quick trip to Lowes I procured a series of translucent plastic sheets. One set, designed for use in commercial fluorescent lighting fixtures, was embossed with an interesting honeycomb pattern. This plastic can be cut by hand or by machine, so it’s possible to make any desired shape. Placing some RGB lighting tape leftover from the 2011 Superbowl project made an interesting effect. However, where the LEDs touched the plastic, their shape and spacing was clearly visible. Moving them farther and farther behind lowered the intensity but diffused the hot spots. 3-4″ seemed a good distance to work with. But making 100 fairly low-cost light boxes to ensure good diffusion just didn’t seem fun. Nor did mounting them on the side of the house. The winter winds here can take your breath away, so safety is always a concern.
Hypothesis #2 – Sideglow Fiber Optic Cable
This is a pretty neat product. While normal fiber optic cable is designed to transfer light from one end to the other with minimal losses, sideglow cable contains added magic which slightly disperses the light along its length. The resulting product can be used as a replacement for neon tubes, but in a much more flexible way. Traditionally, it’s driven by a 200 watt metal halide light bulb and a wheel containing dichroic glass filters.
Perhaps, I thought, the cable could be molded in the outline of a snowflake. This moves the effect away from ‘light blobs in a row’, which is important for this project.
Some folks stateside sell the stuff at what seems to be very inflated prices. Decent eBay sellers seemed to be from Asia, so I cruised Alibaba for a few hours in hopes of skipping the various middlemen.
The cable comes in varying thicknesses from 1-10 mm. I brought in samples of several sizes, plus an 18 watt (2 outputs, each with a 3W RGB LED) drive source.
The verdict? Nice but expensive. Price ex Asia for 6mm cable (the smallest which seemed to emit enough light to be useful) is $1 / foot plus freight. Any drive source less than 3W each of R, G and B is inadequate. 6mm cable has a bend radius of not less than about 40 mm. So fine detail is out. Driving high-brightness LEDs isn’t a big deal, but a properly packaged constant-current DMX-driven LED driver at 3 channels, even not counting our design time, would be ~ $30 each.
Thus making the house snowflakes a $5-6k project. Strike 2.
The driver would be relatively bulky as well. So that’s strike 3.
Also note that the photo above is terribly, terribly overexposed. That much brightness out the side of the cable doesn’t happen in the real world.
Hypothesis #3 – EL Panels
The next step in the quest for something low profile and evenly illuminated led to EL wire and paper. Electroluminescent wire has been around for quite a while, and made especially popular as part of dancers’ costumes, shown briefly between commercials on television talent shows. It’s fairly inexpensive, flexible and continuous in intensity. However, it doesn’t dim well, it’s only available in single static colors.
The existence of EL wire, however, leads the curious researcher to something called EL paper. It’s half a millimeter thick and can be cut to any shape with regular scissors. Drive voltage is 100+v, with current dependent on the paper’s area. It can be dimmed, but not easily. Still, this could be a neat, neat effect. Continuous light output in an arbitrary pattern.
I found a few factories who could make panels 2 meters wide x 1 meter tall. Sadly, the best pricing I could find, even when buying in largeish quantities directly from the maker, hovered around $0.023 per square centimeter. Add designing a DMX-driven high voltage outputter (cough, cough… these don’t seem to even exist!) for each snowflake and the project becomes ferociously impractical. Plus, still single color per panel.
Hypothesis #4 – Edge Lit Acrylic
Here’s where things become very interesting. Folks in the sign-cutting industry have, for a long time, made some striking art by edge-lighting pieces of plastic. Further, designs can be etched in to the plastic, either using a stencil and sandblast method, a CNC router or laser ablation. Each section of the plastic disturbed in some way catches light.
I bought from the local plastic shop four pieces of 3/8″ cast acrylic plate. 3/8″ is expensive, but it’s also the same thickness as my RGB LED tape. These were precut in squares, one each of 6″, 12″, 18″ and 24″. Retail price for the plastic was $70.
These guys have a nice little Epilog laser for cutting and etching, but its table size was too small to handle the 18″ and 24″ pieces. I was referred instead to a shop which owned a drop-dead-gorgeous 150w Kern Laser, featuring a cutting area of 50″ x 50″. Plenty of size and power for this project.
I emailed over some vector graphics files and we were in business.
We happened to have some extra high-power DMX dimmers in the shop – prototypes from a custom design built for a touring concert group. These six-channel dimmers have a super heavy duty output stage, and digitally dims at a frequency far higher than the broadcast video cameras they were using. Each of the two RGB banks can handle 10 meters of double-density (60 LEDs per meter) RGB LED tape. For just fooling around, this is massive headroom.
I wrapped two snowflakes in LED tape and turned on the dimmer. The results were spectacular. Smooth, even illumination everywhere the snowflake pattern was ablated, and complete transparency otherwise. The etched patterns are nearly invisible, even in daylight. The prototypes have a very thin profile and are exceedingly simple to drive. Each snowflake, even at full brightness, requires maybe 300 mA per color at 12v. Over the next month or so, I’ll gin up a dozen or so 36-channel (12 x RGB bank) dimmers with an Art-Net front end. Each snowflake will connect to the dimmer using a length of thin and inconspicuous CAt5 cable. These will mesh nicely with the Art-Net infrastructure already in place.
As an interesting side note, it’s actually less expensive, component-wise, to design using ethernet than to include a properly optically and galvanically isolated input stage for DMX.
There’s only one downside at this point: retail price for the laser time was $225. Even divided by four snowflakes, it’s way too much money. But the look is stunning and perfect, so we’ll figure out a less expensive way to proceed. Perhaps even the vinyl stencil + sandblasting method will prove affordable at this scale. I need to investigate whether a CNC router can produce roughly the same effect but at a fraction of the cost. It may be a good excuse to buy a smaller (36″ x 36″) router just to have. Or a biggish laser table.
Anyway, stay tuned for fall. The flock of snowflakes should be spectacularly beautiful.
If anyone has suggestions for pulling off this look without totally breaking the bank, I’d love to hear from you in the comments below.