Designing Press-Fit Bearing Pockets for 3D Printed Parts: 608 and 625 Bearing Tolerances

A printed part that needs to spin freely around a shaft almost always relies on a cheap skate/skateboard-style 608 bearing or its smaller cousin the 625, pressed directly into a printed pocket rather than bolted down with a separate bracket. Done right, the bearing seats with a firm press, stays put under normal use, and spins smoothly for years. Done wrong, it either falls out the first time the part is handled or, just as commonly, never goes in cleanly in the first place because the pocket was modeled at the bearing's nominal dimension instead of accounting for how FDM printers actually produce circular holes.

Know Your Bearing's Real Dimensions

A 608 bearing (used in skateboards, rollerblades, and endless maker projects) has a nominal outer diameter of 22mm, inner diameter of 8mm, and width of 7mm. A 625 bearing runs smaller: 16mm OD, 5mm ID, 5mm width. These numbers are standardized closely enough across manufacturers that designing to them is safe, but "nominal" still means a real bearing might measure a few hundredths of a millimeter off in either direction — always measure the actual bearing you're using with calipers before finalizing a tight-tolerance pocket, especially if you bought bearings from an unbranded bulk listing rather than a known brand.

BearingNominal ODNominal IDWidthCommon Use 60822mm8mm7mmWheels, rollers, larger rotating assemblies 62516mm5mm5mmSmaller mechanisms, lighter-duty rotation Flanged variants (608Z, 625-2RS with flange)Same bore/width + flange lip——Self-locating in a through-hole, no shoulder needed

Why Printed Circles Aren't Round

FDM printers trace a circle as a series of short straight segments, and more importantly, a horizontal circular hole prints slightly undersized compared to its modeled diameter because of how the nozzle's extrusion width and the part's cooling behavior interact at the hole's edges — this is a well-documented FDM quirk often called "hole shrinkage" or "circle undersizing," and it gets worse as the hole gets smaller relative to the nozzle diameter. For a bearing pocket, this means modeling the pocket at the bearing's exact measured OD will almost always print undersized and too tight, or in borderline cases, just right on one printer and too tight on another with different cooling or calibration. The fix is deliberate, printer-specific compensation, not avoiding the issue.

Sizing the Pocket

Start with a horizontal test print: model a small test block with 3-4 pocket diameters spaced 0.05-0.1mm apart, bracketing your bearing's actual measured OD (for example, if your 608 bearing measures 21.95mm, test pockets at 21.95, 22.00, 22.05, and 22.10mm). Press the actual bearing into each pocket by hand and note which one gives a firm, confident press-fit without needing a hammer or a vise — that value, not a generic rule of thumb, is your real compensated pocket diameter for that printer, material, and bearing combination. Once established, this value is remarkably consistent across prints on the same machine as long as your slicer settings, filament, and nozzle stay the same, so it's worth recording per-printer rather than re-discovering it every project.

• PETG and ABS compress slightly more than PLA under press-fit force, which can make a pocket sized for PLA feel loose in PETG — recheck your compensation value when changing material, don't assume it carries over.
• Pocket depth should match or very slightly exceed the bearing's width (adding 0.1-0.2mm of depth margin avoids the bearing's face binding against the pocket floor before it's fully seated).
• For a through-hole design (shaft passes all the way through), add a small retention lip or shoulder on one side so the bearing has a positive stop and doesn't get pushed through during assembly or use — printing this as a stepped bore (two diameters, the smaller one undersized to the bearing's bore, the larger one the pocket) is straightforward in any parametric CAD tool.

Print Orientation Matters

Print bearing pockets with their axis vertical (hole printed as a horizontal circle, built up layer by layer) rather than lying on its side whenever your part geometry allows it — a vertically-printed hole benefits from the circle-compensation approach above and keeps the pocket wall as a single continuous printed surface. A pocket printed with its axis horizontal (so the hole is formed by the part's side profile across many layers) is far more prone to faceting and layer-stepping inside the bore, which both looks worse and fits less predictably than a cleanly printed vertical hole.

When to Skip Press-Fit Entirely

For a bearing that will see real load, frequent disassembly, or needs perfectly repeatable fit across many printed copies (a kit sold to other people, for instance), consider a flanged bearing seated against a printed shoulder and secured with a light retaining ring or a printed snap-fit tab instead of relying on friction alone — press-fit retention degrades gradually with repeated insertion/removal cycles as the printed pocket walls compress and don't fully spring back, which matters less for a one-off project and more for anything meant to survive years of use or multiple rebuilds.

Once you've found your printer's real compensation value for a given bearing and material, bearing pockets stop being a guessing game — they become a known, repeatable number you plug into every future design, and that one small calibration step quietly saves a disproportionate amount of redesign-and-reprint frustration down the line.