T – 4 weeks
I stayed in Brisbane for a few weeks either side of Easter. I had no projects in mind for this period, apart from maybe whipping up something quick to light my campsite at a doof I was going to. Before I’d even left — long before I’d had the chance to show off my Kaiju suit to my local hackerspace to show it off — a friend informed me of a charming man who had recently joined, who had a plan to build an art car in the coming weeks, just in time for the very festival I would be attending. He put me in touch, and by the time I’d made it to HSBNE, I’d already signed on.
T – 3 weeks
Back in Brisbane, I bring my newest project along to HSBNE and a few people are suitably impressed (I’ll see myself out, thanks). My charming new friend tells me he’s in the midst of organising the purchase of the vehicle which will become the basis for our project. I figure that not all of the strip needs to be addressable, and we can save some construction costs by casting solid colours on some sections using the much cheaper common-anode strip, a handful of DIP WS2811 chips and some MOSFETs (see below). Knowing the rough size of the vehicle, I order what seems like an appropriate amount of LED strip, including spares to deal with the usual unforeseen consequences.
I spend the next week designing a circuit which can use the current-sinking ability of the WS2811 chip to PWM a BJT, which in turn drives the gate of a MOSFET, to switch the power for the common-anode strip. It seems to work in a circuit simulator. That obviously means it’s going to work flawlessly the first time in the real world, right? I start programming the addressable part and getting the power system in order.
T – 2 weeks
The team gets to work on the construction. I help when I can, and start building a bunch of current-sink-to-FET translators when I can’t. Near as I can tell, they seem to work.
T – 1 week
The LED strip arrives, and I get to work hooking up a WS2811 chip to one of my triple-FET veroboards.
Pop. There goes a WS2811. Hmm, maybe they can’t actually take 12V on VCC like the “datasheet” says.
Pop. Oh, right. Need a resistor there.
…no smoke! No lights either. Wait, the red one lights up briefly. Hold on, I missed a connection.
Pop. Seriously? Good thing these come in packs of 100.
While I do end up with …something, it doesn’t work out as I’d planned — whatever I try, either the lights come on only some of the time, one of the channels doesn’t work, or my increasingly-frustrated brain messes up somewhere and I let the smoke out of yet another WS2811. The most annoying thing about this exercise is that my breadboarded prototype seems to work just fine until I add a resistor that the circuit shouldn’t have worked without. Amusingly, it is here that I learn the lesson that the faster I rush, the less I accomplish over time. Cooler heads really do prevail.
Eventually I give up on the WS2811 bridge, and change the controller from an Arduino Uno to a spare Mega ADK, and use the plentitude of PWM pins on the Mega to drive the FET gates directly. While the threshold voltage for the FETs is right on 5V (meaning RDS never quite reaches RDS(on)), it does manage to colour-fade correctly. Near enough, meet good enough.
Meanwhile, the connectors I found for the 4-wire strip work fine so long as they are soldered to the strip. Yes, they’re labelled as solderless connectors, but simple contact doesn’t always conduct as well as it should, and some flux-cored solder removes any surface corrosion and seals the contact area nicely.
T – 3 days
I get all the LEDs installed. The controller drives them just fine, although now I’m running three data lines to each controller board instead of one. C’est la vie. Power works. Controller works. My long-ass USB cable lets me sit far enough from the Queenslander to see the changes after I upload its program. I tweak it so that the timings of the windows are evenly-spaced. Again so that the windows don’t all change to the same colour in a sequence. Speed it up. Slow it down. Accentuate the sparkle trail. Remove the sparkle trail.
I’m buggered from getting it all installed, but gleeful that I can now see the payoff. I cut the lightbox windows from some scrap plastic (leftover PETG from my Laser Angel Wings), then get busy with an orbital sander to frost them. For kicks, I use the space’s laser cutter to etch the front surfaces.
The Big Lebowski Event
We did it! We brought The Wonky Queenslander to life, just in time for Rabbits Eat Lettuce.
The day of the drive to the festival (Thursday), it rains all afternoon. The rain wouldn’t let up until Saturday afternoon. The site starts wet and ends up a huge mudpit. That doesn’t stop anyone from having an awesome time, however!
Lessons learned
- Four weeks is just long enough to throw a larger project like this together in a hurry.
- Transistors are not simple switches, there is some serious semiconductor wizardry going on inside them, and you need to understand that wizardry to use them in practice.
- “Solderless” connectors can work better when soldered.
- Common-anode LED strip is remarkably fault-tolerant if you attach R+G+B+A to both ends (it could have a single break in each and still work fine)