I’m Dan and I have a problem.
This is Nacho. He’s an eighteen-month-old green iguana and his enclosure is criminally small, so I built him a new one.
The constraints, in the order they were given to me:
- It has to match my wife’s furniture aesthetic.
- It has to be awesome and badass.
- It has to be mobile, and fit over a radiator in the children’s science room.
- It has to be intensely automated, because that is how I roll.
Only one of those is mine.
Up front, so nobody has to write it in the comments: the finished enclosure is five feet wide, and a full-grown green iguana wants more than that to turn around comfortably. Reptile people are right to be annoyed with me. This is not Nacho’s forever home — it’s the enclosure that fits the room he’s in now, and it’s a very large improvement on what he had. I’d rather say that plainly than pretend otherwise.
Welding something bigger than I’ve welded
I’m not a fabricator. I’m not really even a welder. The plan was five feet wide by four feet tall by two feet deep, on stilts, in square steel tube, and I was missing most of the fixturing that would make that straightforward.
I welded the two sides flat, figuring I’d work out how to connect them into three dimensions afterwards. This took far longer than it should have, mostly because heat does what heat does — steel warps and expands and the rectangle you tacked is not the rectangle you get.
When I brought the sides together, they weren’t square to each other. Diagonal measurements that should have matched didn’t. There’s a stretch of this build that is just me measuring corner to corner and doing Pythagoras on a shop floor, trying to work out how far out of true I actually am.
The fix was to build a reference surface first: big aluminium blocks, extruded aluminium across them, wood on top, all shimmed dead level. Once I had a flat plane I could trust, I could clamp to it and stop guessing.
That’s the lesson from this whole section. I should have built the reference surface before the first weld, not after the sides didn’t fit.
Afterwards, the welds got cleaned up with an angle grinder and painted so they wouldn’t rust. They look considerably better than they were.
Wood, polycarbonate, and brackets that let it come apart
The structure is steel, but everything you see is ¾-inch red oak plywood and ¼-inch polycarbonate.
Rather than weld tabs everywhere, I designed and printed a set of brackets, then drilled and tapped holes directly into the steel tube so every panel screws in. It means the whole thing can come apart, which matters for something that has to be carried through doorways and lives on wheels.
Polycarbonate at ¼ inch is more flexible than you want for a five-foot span. The fix was aluminium channel sized for ¼-inch stock, slid onto the edges — cheap, and it stiffens the panel dramatically.
The doors are the best part of this build
Sliding doors on an enclosure this size are a real problem. They have to move easily with one hand, seal well enough to hold humidity, and not bind when the frame inevitably moves.
I printed the runners in nylon. Nylon has a very low coefficient of friction, so the doors slide on plastic-on-plastic with no rollers, no hardware, and nothing to seize up. A little dry lube and they glide.
Nylon is also a pain to print — you need a heated enclosure and it likes to warp — which is why I mention it at every opportunity. It took a few attempts to find a way of fusing the printed sections together that actually held.
They work better than I expected. Better, honestly, than the doors on furniture I’ve bought.
Water, and the still suit
The bottom of the enclosure is a water system.
Water passes down through the substrate and a semi-permeable membrane into a collection tank, which is a Lexan drawer that slides out so I can actually service it. Pumps on either side carry water back up through 3-inch PVC columns to the roof, where check valves stop the lines draining down so nothing has to re-prime. At the top of each column sits a waterfall outlet and a fogger, and whether the foggers run is decided by measured humidity.
The roof has five holes: a basking light, a ceramic heat emitter, the two water columns, and a central exhaust vent. The lights are recessed into the roof rather than sitting on top of it, because I have a cat and the cat smacks things.
The vent is where the odd idea lives. Bringing fresh air in means dumping humidity out, which is exactly wrong for a tropical animal. So I built a moisture-retention circulation system — fresh air comes in, but the water vapour is caught and kept rather than exhausted. It’s a still suit, and yes, that’s the Dune reference. Frank Herbert has been rattling around this shop for years.
The drawer leaked on the first fill. That was a mess, and I did fix it — I said the details were coming in Part II.
The electronics, which you never got to see
The brain is a Raspberry Pi, and the reason it’s a Pi rather than an ESP32 is that I wanted this thing to be set-and-forget, redundant, and watchable on a camera while I’m not home. It’s an animal, not a light strip.
- An I²C relay board for lights, heat, pumps and foggers. I didn’t know these existed until I found one, and it removed a whole rat’s nest of transistors.
- A PWM fan for controlled intake rather than on/off.
- SHT30 temperature and humidity sensors — properly industrial parts, not hobby modules — fanned out through an I²C multiplexer so I can run several on the same bus and cross-check them against each other.
- A camera, so I can look in and say hello.
What actually happened
Part I ended with the promise of Part II — decorating and going bioactive — and Part III, the full tech walkthrough with the code and the still suit.
Neither got made. The video landed to a quiet reception, then Open Sauce happened, then I fell down a brushless-motor rabbit hole, then Halloween arrived with a hard deadline and Ghost Rider ate everything.
The enclosure itself got finished and Nacho lives in it. It’s the documentation that stopped, which is a slightly embarrassing thing to admit on a page whose entire job is documentation. So consider this the belated Part I write-up, and the tech walkthrough still owed.
If you’re building one
- Build the flat reference surface first. Not after your welded sides refuse to meet.
- Tap the steel, don’t weld tabs. Being able to disassemble a 5-foot enclosure is worth the drilling.
- Aluminium channel on polycarbonate edges turns a floppy panel into a rigid one for a few dollars.
- Nylon runners beat rollers for big sliding doors — fewer parts, nothing to jam.
- Put the water tank on a drawer. Whatever you build will leak once, and you will need to reach it.