The $400 Check That Prevents the $40,000 Mistake

Steel is expensive to cut and worse to modify. A mold change to move a boss 2 mm costs one to three weeks and four figures. The same change caught in CAD, after assembling a set of printed parts on a bench, costs an afternoon. Yet the most common sequence in a first-time hardware program is CAD review, design freeze, tooling kickoff, and then a discovery at T1 that the PCB clears the boss by nothing and the enclosure will not close over the connector.

A full trial assembly in printed parts is the highest-return activity in the entire pre-tooling phase. It is cheap, it takes days, and it converts an argument about a screen model into a physical yes or no. The rule is simple: nobody cuts steel until every part of the product has been assembled at least once, by hand, with real fasteners and the real PCB.

What You Are Actually Checking

A trial assembly is not a general vibe check. Work a written list so nothing gets skipped because it looked fine.

  • Does the enclosure close? Halves that meet with no gap, no rocking, and no gap variation around the parting line. Measure the gap with feeler gauges at six points rather than eyeballing it.
  • Does every purchased component drop in? The actual PCB, not a printed dummy. The actual connectors, switches, battery, motor, lens, and speaker. Purchased parts have their own tolerances and they are frequently worse than yours.
  • Do the fasteners land? Every screw driven with a real driver at the real angle. You are looking for bosses that a driver cannot reach, screws that are 2 mm too long, and heat-set inserts with no clearance for the installation tip.
  • Do the ports and openings line up? Plug a USB cable in. Press each button. A hole positioned correctly in CAD can still be unusable if the wall is too thick for the connector's shroud.
  • Do the seals compress? Gland depth, gasket squeeze, and whether the closure force is achievable with the fasteners you specified.
  • Does the assembly sequence work? Build it in the order the factory will. If step four requires holding two parts while inserting a third and starting a screw, you need a fixture or a redesign, which is the whole subject of design for assembly.
  • Where does the stack-up bite? Printed parts reveal accumulation that a nominal CAD model hides, and any failure here should send you back to a proper tolerance stack-up analysis rather than to a quick nudge of one dimension.

Know What the Printed Part Is Not Telling You

The value of this exercise collapses if the team over-reads the results. A printed part is a geometry proxy, and it lies in specific, predictable ways.

Dimensions are off in a direction you can correct for. FDM parts typically come out with holes undersized and outer dimensions slightly oversized. Resin parts shrink during UV cure. SLS parts vary with position in the build volume. Measure the printed part before you conclude anything: if the boss came out 0.15 mm oversized, the interference you just found may be the printer, not the design. Working from realistic printed tolerances keeps you from chasing phantom problems.

Stiffness is wrong. A printed enclosure flexes differently than a molded one. A lid that seems to close acceptably because the printed wall bows may not close at all in glass-filled nylon, and a snap arm that feels right in resin will behave completely differently in polypropylene. Verify latch geometry against snap-fit design rules instead of trusting the feel of a printed one.

Threads and inserts behave differently. Self-tapping screws cut cleanly into printed material and strip in molded material, or the reverse. If the product relies on threaded assembly, print test coupons and pull them, using the practices in threads and inserts in printed parts.

Molding artifacts are absent. There is no sink over a thick rib, no gate vestige, no knit line, no draft. A printed part with vertical walls will assemble fine and its molded twin, with 1.5 degrees of draft on every face, will interfere at the bottom. Model the draft into the parts you print, and check wall sections against uniform wall thickness rules before you print rather than after.

Which Process to Print In

Match the print process to the question you are asking, and be willing to print the same assembly two ways.

For pure fit and clearance, FDM is usually enough and costs almost nothing. Print one round overnight, find the obvious collisions, fix them, print again.

For anything with fine features, small snap fits, thin bosses, threaded holes, or light pipes, print SLA or MJF. FDM resolution will invent problems that do not exist and hide ones that do.

For parts that must be handled repeatedly or flex, SLS or MJF nylon is the closest available proxy for a molded thermoplastic and survives dozens of assembly cycles without cracking.

How to Run the Rounds

Three rounds is the normal count and each one has a job. Round one is a collision check: print everything, assemble, expect several problems, and do not bother with cosmetics. Round two verifies the fixes and adds real purchased components and fasteners. Round three is a rehearsal of the production build, assembled by someone who did not design it, in the documented sequence, timed.

Discipline matters more than technique. Photograph every failure with a scale in frame. Log each finding with the part number, the measured value, and the intended value. Change one variable at a time. Keep every round's parts labeled with the CAD revision, because six weeks later nobody remembers which grey box was rev C. And put a non-designer on the assembly, since the person who modeled it unconsciously handles it correctly.

When You Are Done

You are ready for tooling when a complete unit assembles in the production sequence with no shims, no filing, no unexplained interference, and no verbal instructions, and when every deviation you accepted is written into the drawing as a tolerance rather than remembered by one engineer. Everything after this point gets checked against real molded parts in the first mold trial, and the fewer surprises you carry into T1, the fewer weeks you spend in tooling revisions.

Running the Assembly Review Before You Commit

Projects House runs pre-tooling fit and assembly reviews for clients: print plan, checklist, measured findings, and a consolidated change list your CAD team or ours can execute in one revision. Send your assembly model and target tooling date through our contact form.