Inside a 30-Printer Plant Pot Farm

At 11 o’clock on most nights, one of us is standing in a garage in Pennsylvania with a spool of filament under one arm, going down the rows and swapping out the ones that won’t survive until morning. You learn to judge a spool by weight. Somewhere under half a kilo left and it isn’t finishing an overnight print, so it comes off now, even if there’s technically enough for one more pot. That judgment call, made about forty times a week, is most of what running a print farm actually is.

Jill and I are nurses. We started this in 2018 after a plant fair where we sold out of planters we’d printed on one machine, for fun, in a weekend. We came home, bought a second printer, and then we kept going. There are more than thirty of them now, and they have taken over the garage completely.

People ask what a 3D print farm looks like. Mostly it looks like laundry. Here’s the real version.

On this page

• What a print farm actually is
• The numbers, honestly
• What a day looks like
• Why we print to order instead of stocking shelves
• What goes wrong, and how often
• The garage is the hardest variable
• The filament, and what it is
• What we check before a pot goes in a box
• Would we do it again

What a print farm actually is

A print farm is a lot of 3D printers running the same kind of job at once, managed as one unit instead of as thirty hobbies. That’s the whole definition. There’s no special machine. Ours are the same class of desktop printer a person might have one of at home. We just have thirty-something of them on shelving, on a concrete floor, with a lot of extension cord.

What changes at scale isn’t the printing. It’s everything around the printing. One printer, you babysit. Thirty, you cannot. So you stop optimizing for the best possible print and start optimizing for the print that succeeds unattended while two people are asleep or at work. Those are different goals and they pull in opposite directions.

Practically, that means slightly thicker walls than strictly necessary, slightly slower first layers, geometry that doesn’t need support material, and a refusal to design anything whose success depends on the machine having a good night. If a model fails twice out of twenty, it doesn’t ship. We’ve killed designs we loved over that number.

The numbers, honestly

• 30+ printers. The count moves, because at any given time one or two are apart on the bench.
• About four hours a pot. Small ones come in under three, the bigger designs push past five.
• 0.4mm nozzle, 0.2mm layers on nearly everything. Finer layers look marginally nicer and double the print time, which is not a trade we can make thirty times over.
• Roughly one spool per machine per few days at full tilt, which is why the 11pm walk exists.
• Thousands of orders shipped since 2018. Every single one of them printed after somebody clicked buy.
• Two business days from order to shipped, which is the constraint everything else is built around.

That last number is the one that makes this hard. Four hours a pot and two days to ship means a single machine can realistically produce a handful of pots inside the window. The only way the math works is parallel machines. Thirty printers isn’t a flex. It’s arithmetic.

What a day looks like

Morning: walk the rows with coffee. Every bed gets cleared, every finished pot gets pulled and set on the table, every failure gets noted. This takes about forty minutes if nothing went sideways overnight, longer if something did.

Then the queue. Orders that came in overnight get grouped by design and assigned to machines, biggest and slowest first so they’re running while the day is long. Small stuff fills the gaps.

Midday is packing, which happens at the kitchen table because the garage is full of printers and the kitchen has better light. Then whichever of us is on shift leaves for the hospital, and the other one keeps an eye on things.

Evening is the second load. And then the 11pm spool walk, which is the ritual that defines the whole operation. Cold garage in February, warm garage in August, same walk, same decision at every machine: does this finish, or do I change it now?

Working nights makes this easier than it sounds, weirdly. A nurse coming off a 7pm-to-7am shift is awake at hours when normal businesses are closed, and a printer does not care what time it is.

Why we print to order instead of stocking shelves

The obvious alternative is to batch-print a few hundred of the popular designs, box them, and ship from stock. Faster shipping, simpler mornings. We’ve thought about it every year and we’ve never done it.

Part of it is space: thirty printers and a few hundred boxed pots do not both fit in a garage. Part of it is that made-to-order means we can change a design the day we find a flaw, rather than shipping four hundred of the bad version first. Our early boot planter had a rim that gripped nursery pots too tightly, and we fixed it in a week because there was no inventory arguing against the change.

And part of it, honestly, is that we like it. A pot that exists because you specifically asked for it is a different object than a pot that sat in a bin for eight months. That’s not a business argument. It’s just true.

What goes wrong, and how often

Failures cluster. A good week, you lose a couple of prints across the whole farm. A bad week, one machine develops a habit and eats six.

First-layer adhesion is the top cause by a distance. If the very first layer doesn’t stick evenly to the bed, the print either peels up at a corner or detaches entirely, and you come down in the morning to a tangled nest of filament the size of a grapefruit. The fix is boring maintenance: clean the bed with soap and water, not alcohol, and re-level after anything gets bumped.

Nozzle clogs are second. A partial clog is worse than a full one, because the machine keeps going and produces a pot with thin, gappy walls that looks almost fine until you hold it up to a window.

Filament tangles on the spool are third and the most infuriating, because the print was going perfectly and a knot forty layers back ends it. This is why we’re picky about brands and why we stopped buying whatever was cheapest that month.

Then there’s the category we call “it was fine yesterday,” which covers belts, fans, thermistors and the slow drift of a machine that’s printed continuously for two years. Every printer has a personality by month six. Number 14 has never once made a clean bottom layer and we’ve stopped asking it to.

The garage is the hardest variable

A factory has climate control. We have a Pennsylvania garage, which in January is very cold and in July is very much not.

Cold is the bigger problem. PLA shrinks as it cools, and when the ambient temperature drops, the bottom layers of a tall print contract at a different rate than the layers going down on top of them. The corners lift. Last February the garage got down into the low 40s overnight and an entire back row warped at the base, six pots, all of them scrap. We now run enclosures on the machines in the coldest corner and keep a space heater on a thermostat, which solved about ninety percent of it.

Humidity is the quieter one. PLA absorbs moisture out of the air, and wet filament prints badly: you hear it popping and hissing at the nozzle and the surface comes out rough and brittle. Spools live in sealed bins with desiccant, and anything that’s been open more than a few days goes through a dryer before it goes near a machine. In a humid Lancaster County August this matters more than anything else we do.

Also, it’s loud. Thirty stepper motors is not a quiet room. You get used to it and then you notice immediately when one of them changes pitch, which is genuinely useful.

The filament, and what it is

Everything we print is PLA, which is a plant-based plastic, usually made from corn starch, and made in the USA for what we buy. It prints at a low enough temperature to be practical in a garage and it holds fine detail better than the alternatives, which matters when a design has a face on it.

We get asked constantly whether it’s safe to have a plant living in it, and whether the material itself is a concern in a house with kids and animals. Those are two different questions and we’ve answered both at length: is PLA safe for plants covers what thirty printers’ worth of pots have taught us about roots, soil and water sitting against printed walls, and is PLA toxic is the plain-English version of the material question.

What we will not tell you is that it’s biodegradable in your backyard, because in your backyard, it isn’t. PLA breaks down under industrial composting conditions that your yard does not have. We’d rather say that plainly than get cute about it.

What we check before a pot goes in a box

Every pot gets picked up and looked at. That sounds soft, but hands catch things a checklist doesn’t. Specifically:

• The drainage hole. Clear, round, no stray strands across it. Every pot we make has one.
• The bottom layer. Held up to the light. Any gap you can see through is a pot that will weep in somebody’s living room.
• The inside diameter, checked against the nursery pot size it’s meant to take. A pot that’s two millimetres tight is a pot that gets an email about it.
• Layer consistency on the visible face. A band of slightly different texture usually means the machine had a moment. It goes in the scrap bin.
• It sits flat. On the kitchen table, which is dead level and has adjudicated hundreds of these.

Scrap rate is real and we don’t pretend otherwise. Some percentage of everything we print never leaves the garage.

Would we do it again

Yes, with one change: we’d have gone from one printer to six much faster, and from six to twenty much slower. The jump from a handful of machines to a room full of them is where you stop being a person who prints things and start being a person who maintains a small factory, and nothing about that transition is in any tutorial.

The advice we’d give anybody thinking about it: the printers are the cheap part. The expensive parts are your evenings, your garage, and the surprisingly emotional experience of throwing away a pot that took four hours because there’s a pinhole in the base. Also buy better filament than you think you need. We learned that one the hard and lumpy way.

If you want to see what all of that produces, the whole lineup lives here, and every one of them was a spool and a decision at 11 o’clock at night not so long ago.