Lab note · Oct 28, 2026
How we find the perfect fit in one print
Push-fit parts need a gap of a few hundredths of a millimetre. Here is how we find it with one test print instead of ten.
- Test prints1
- Sizes6, two of each
- Step0.02 mm
The question
Our coaster holders have small charms that press into + shaped sockets. Too tight and the charm will not go in; too loose and it falls out. The window between the two is about 0.05 mm, smaller than the difference between two printers.
What went wrong first
We guessed: one size, print, try, adjust, print again. Three rounds later we still had no answer, and the parts got mixed up the moment they came off the plate: nobody could tell a 0.18 from a 0.20.
Worse, the same charm went into one socket and not the next. One size was never going to fit them all.
The method: a numbered sweep
One plate, six sizes, two of each, from "surely too tight" to "surely too loose", 0.02 mm apart (the gap per side between the peg and the socket).
The secret is the marking: size 1 gets one small dot on the back, size 2 gets two dots, and so on. The dots are hidden once the charm is in, but you can always tell the parts apart.
Then try every socket and write down which dot count goes in by hand AND stays when you turn the holder upside down.
What we found
On a holder with flat walls, one size fit every socket. On our rounder holders, the sockets came out different sizes depending on where they sit: the two in the middle of the curve printed tighter, the ones at the ends looser. The same pattern showed up on two different holders.
And one end socket was tighter than its mirror twin on the other side, with nothing inside it. The likely cause is airflow: the cooling fan blows from one side, and plastic that cools faster sags less, so that hole came out a hair smaller.
| Dots | Gap per side |
|---|---|
| • | 0.18 mm |
| •• | 0.20 mm |
| ••• | 0.22 mm |
| •••• | 0.24 mm |
| ••••• | 0.26 mm |
| •••••• | 0.28 mm |
| Socket | 1 | 2 | 3 | 4 | 5 | 6 |
|---|---|---|---|---|---|---|
| Dots | • | •• | ••• | ••• | •• | • |
What we do now
- Mark every test variant. Dots on a hidden face cost nothing and end the guessing.
- Sweep wide enough in one print to go from "won't go in" to "falls out".
- On curved parts, test every position: sockets differ by where they sit.
- Keep the seam away from fits (Seam position: Back) and ship one extra size for the odd tight socket.
- In the final file, mark the sockets too, with the same dots: assembly becomes "match the dots".