All these LEDs on the outside have to be powered and managed by something on the inside. Kevin and I have settled on a ‘Hub’ design where each hub consists of a PSU (Power Supply Unit) supplying 5 volts, and an ESP32 controller running WLED firmware. These hubs will take in 120 volts + Ethernet, and will emit 5 volt power + control signals for the LED strips. They will be located close to the LED strips to minimize distance — an important factor at low voltages because of voltage loss in long wire runs.
I was not anticipating how hard the design would be. Kevin and I have spent several hours on phone calls working through the details, and I’ve spent a couple dozen hours modeling different possibilities in my CAD program to figure things out. Lots of iterations, lots of little changes along the way, but I believe we’re converging on something that will be close to the final design. There are a lot of constraints to satisfy, and as anyone with systems design experience knows, lots of somewhat conflicting constraints equals a hard problem. Price, power, reliability, ability to have sheet metal actually be created in the design, ease of assembly before playa, and especially ease of assembly ON the playa — lots of variables to juggle to come up with something that is by some definition, “optimal”.
The cost thing snuck out of nowhere… ok, it swaggered menacingly straight out of the power budget. We’ve got a LOT of LEDs… 16,000 of them… and they suck down a LOT of power. Although giant PSUs are available, they cost an arm and a leg, and its cheaper to assemble a lot of smaller ones. 22 or 24 of them… call it 25 because we need a spare… anything times 25 is a lot.
The wires turn out to be expensive. Voltage loss in low voltage circuits is very material, especially at 5V. My first design had hubs placed pretty much in the middle of the structure, which unfortunately made for maximum runs of cable — over $1K in wire! Yikes. I considered cheap plastic boxes for the hubs, but the PSUs require good airflow for cooling, and I also need to protect them from water (yes, it MOST DEFINITELY rains on the playa, cf BM23, as well as my local festival territory of Seattle), so they’d need to be attached to the T-Slot implying an aluminum bracket. I could have gone low end and just arranged boxes on the ground, but… half of the feeds were on top of the prism, 12′ away. 12′ times 3 runs * 8 strips = a boatload of money in wire. For control, we’ve got more leeway, so it wasn’t a hard constraint, but it does does add to the cost if we want to keep things simple to assemble both before and on the playa — each hub should be identical and fungible, requiring an ESP32 controller at almost $30 each in addition to the PSU. Original PSU choice was almost $60 (*25 = $1,500!) but ended up with a $30 one.
Here’s the first design: up in the air, midway up the prism, but now I had 6′ to just get up and down… no help:
So, I needed hubs that would be attached to the structure, up in the air away from the water, with plenty of airflow, and close to the LEDs they’re powering. Enter designs 2, 3, 4, and 5. Sheet metal reality dictates physical constraints (oops, the bending brake interfered with the bend), cost (holy crap, $60 for a bracket?), and physical reality (huh, the strain relief cable glands farther away from the inputs than the pigtails are long). I ended up with the design featured up top on the post and in the diagrams below. Easy to put down low, up high, not super expensive ($22 for the bracket), with good physical protection of the components, good physical integrity, ease of field assembly… I am hoping this is close to the final design.
As with most things in system design, this is like poetry: subtract out everything that isn’t essential. Its trivially easy to design complex systems that do complicated things, its much harder to design simple systems that accomplish the same complex goals. In systems design, we talk about “DF*” — “Design For…”. In this one, we’ve got “Design for Cost”, “Design for Manufacturing”, “Design for Power Efficiency” and possibly the most important one, “Design for Playa”.
I have to admit, partway through this weekend of painful optimization, I considered bagging the overwhelming LED complexity (this is, after all, supposed to be a music centered project initially…) but imagining just how cool this will look out on the playa, LEDs blazing and dancing… yeah, let’s keep up the pain. I’m very glad I’ve got a fellow co-conspirator in Kevin willing to jump into the complexity pain bucket to emerge with solutions that work in the real world… otherwise I really would just say ‘screw it, put up a spotlight and call it a day’. As I’ve discovered over and over again in my career, complexity is best tackled as a team because its all so daunting by yourself.
Next step: prototype the first production hub, then order a ton of expensive components for work parties in early June.