A 3D printer arrived just before Christmas, and I have spent the gap between then and now producing, by weight, considerably more failed plastic than successful objects. I want to write this down while the failures are fresh, because every guide online shows you the finished benchy gleaming on the bed and almost none of them show you the bin next to the printer, which in my case filled up first.
I went in expecting the hard part to be the design, the CAD, the modelling. It is not. The hard part, at least at the start, is getting the very first layer of plastic to stick to the bed and stay stuck, and almost everything that went wrong in my first fortnight came back to that one unglamorous problem.
failure one: nothing sticks
The first half-dozen prints did not fail dramatically. They failed by simply not happening. The nozzle would lay down its first pass and the plastic would come straight back up with it, dragged around the bed in a sad little ball. No adhesion at all.
The cause was the thing everyone tells you and you ignore because it sounds too simple: the bed was not level, or more precisely the gap between nozzle and bed was wrong, and it was wrong in a way I could not see by eye. Too far and the plastic has nothing to press against. The fix is the famous paper trick, sliding a sheet of paper under the nozzle at each corner and adjusting until you feel a slight drag, and then doing it again because the first pass is never quite right.
What nobody quite says plainly is that "level" is the wrong word. The bed being flat is necessary but not sufficient. What you are actually setting is the nozzle height, the squish of that first layer, and getting it consistent across all four corners. Once I stopped thinking "is it level" and started thinking "is the first layer being pressed down evenly everywhere", the adhesion problem mostly went.
failure two: the corners lift
With things finally sticking, I hit the next classic. Larger prints would start fine and then, an hour in, the corners would peel up off the bed and curl, warping the whole base and sometimes knocking the print loose entirely. Hours of printing, ruined by a lifting corner.
This is warping, and it is physics rather than a fault. The plastic shrinks slightly as it cools, the outer parts cool faster than the centre, and the contraction pulls the corners up. The bigger and flatter the part, the worse it is. The fixes are all about controlling how the plastic cools:
- A heated bed kept warm enough that the lower layers do not cool too fast.
- No cold draught across the printer, which in a January workshop is not nothing.
- A "brim", a thin skirt of extra plastic around the base that gives the corners more grip to fight against.
The brim was the one that actually rescued the larger prints. It is wasteful, you peel it off and bin it afterwards, but a print that finishes with a brim beats a perfect-looking one that lifts off at layer two hundred.
failure three: the stringy mess
The third failure was cosmetic but maddening. Prints came out covered in thin wispy strings of plastic, like cobwebs, stretched between the parts of the model where the nozzle had travelled over a gap. The object underneath was fine; it just looked as though a spider had got to it.
This is stringing, and it is about the nozzle oozing plastic while it moves between points rather than printing. The cure is "retraction": pulling the filament back a little whenever the nozzle travels, so it is not under pressure and does not dribble. My slicer had retraction on by default but tuned for a different material, and getting the distance and speed right for the filament I was actually using took a few test prints of the little tower-and-bridge models people use for exactly this. Tedious, but it turned the cobwebs into clean travel moves, and there is a real satisfaction in pulling a print off the bed that needs no cleaning up at all.
the pattern underneath the failures
Here is the thing that surprised me, and it is the reason I am writing this as an engineer rather than a hobbyist. Almost none of my failures were the machine breaking. The printer did what it was told, faithfully, every time. The failures were upstream of it: the bed height I set, the slicer settings I chose, the cooling environment I had not thought about. The machine was the most reliable part of the whole pipeline. I was the unreliable part.
That is a familiar shape. It is the same shape as a flaky deploy that turns out to be a config you got wrong, or a "broken" service that is faithfully serving exactly the bad input you handed it. The tool executes your instructions precisely, and when the output is wrong the instructions are usually where to look first. I have spent a career relearning that in software, and apparently I get to relearn it again in plastic.
If you are starting out, the single thing I would pass on is to keep the failures. Do not bin them straight away. Line them up on the shelf and look at how each one went wrong, because the failure mode is legible if you let it be: a lifted corner, a stringy web, a base that never gripped. Each one is the machine reporting, very honestly, exactly which of my settings or my assumptions was off. Read in that order, the bin of spaghetti is not a record of failure at all. It is a fortnight of fairly cheap, fairly fast lessons, each one pointing back at me rather than the printer.
So the bin is full and I am not discouraged. Every failed print taught me something specific and physical that no amount of reading would have. The first layer is everything. Corners want to lift. Draughts are real. And the machine, bless it, was right all along, which means the next batch of failures will be mine to own too, and that is exactly how I would want it.