Modify the Prototype or Rebuild It

The prototype came back from testing with a cracked bracket, a mounting boss in the wrong place, and a note saying the button should be 8 mm higher. The instinct is to fix the CAD and print a whole new one. Sometimes that is right. Often it is a week of lead time and several hundred dollars to change three features on a part where everything else was correct — and it throws away an object already assembled, wired, calibrated, and validated in every respect except the three that changed.

Model makers have modified rather than rebuilt for as long as models have existed. Done properly the result is indistinguishable from a fresh build. The skill is knowing which the situation calls for.

The decision, in one table

Two rules override the table. Never modify a model used for dimensional verification — patched geometry is not trustworthy for measurement, and dimensional accuracy in a model is exactly what a repair compromises. And count the modifications: a part cut and patched three times carries enough unknown internal geometry that the next change should start from a fresh print.

Repairs and modifications that actually hold

Adding material

For printed thermoplastics, the strongest patch is welding with the same material: a flat-tipped soldering iron, or a welding gun with filler rod cut from the same spool. The join is fused rather than glued, and on ABS a brush of acetone slurry made from ABS scrap both bonds and fills. For resin parts, a UV-curable resin of the same family works well. Two-part epoxy is the universal fallback: strong, gap-filling, sandable, but a different material that finishes differently.

For structural additions, a hybrid beats a glued blob: print or machine a small insert with a mechanical interlock — a dovetail, a stepped shoulder, a keyed pocket — mill a matching pocket in the original, and bond the two. This is a small-scale version of combining printing and machining in one prototype, and the mechanical engagement is what makes the repair survive load.

Removing material

Cutting is easy; the risk is cutting the wrong thing. Scan or photograph and record dimensions first, so the original geometry is recoverable in CAD even if the part is not. Use a rotary tool for pockets, a small mill for anything that must be square, and a step drill for round openings. Support thin walls from behind or they will flex and tear.

Threads and fastening points

A stripped screw boss is the most common prototype failure. Drill it out and install a heat-set insert; the repaired joint is usually stronger than the original and survives many assembly cycles. Do it preemptively on any boss that sees repeated assembly during testing — the whole argument for using inserts rather than tapped plastic. For a boss in the wrong place, epoxy a machined aluminum block into the void and drill and tap it where you need it.

Surface and cosmetics

Filler primer, spot putty, and progressive sanding hide almost any repair, which is why a well-patched model still looks production-ready. The standard sequence for finishing a printed prototype applies unchanged — the only extra step is checking that patch and base material sand at similar rates, or you will create a low spot.

Electronics and wiring

Modifying a board is its own craft: cut traces with a scalpel, add bodge wires in fine enameled wire, tack a piggyback board onto the pads of a removed part. Photograph every modification and mark it on a printed schematic immediately, because an unmarked bodge wire is a bug waiting to be found six months later. A board with more than a handful of cuts and jumpers has stopped being a reliable test article.

Change control, even on a prototype

The real cost of modifying models is informational, not structural. A physical part that no longer matches the CAD is a trap: the next print comes out of the file without the changes, and nobody remembers what was different. Three habits prevent that:

1. Update the CAD the same day. The physical modification is the experiment; the CAD is the record. If you cannot model it immediately, add a dated note and photographs to the file.
2. Label the physical part. Write a revision on it in permanent marker. Two identical-looking prototypes on a bench, one modified and one not, get mixed up within a day.
3. Keep a modification log — date, what changed, why, who did it, what happened after. This is ordinary prototype documentation, and it lets a change graduate cleanly into the engineering change process at release.

The economics

A modification session at a model shop is typically a few hours of skilled labor — call it $150–$600 — versus a reprint of a moderately sized housing at $200–$900 plus finishing, reassembly, and lead time. The gap widens sharply for large parts and anything that took a day to wire, and narrows to nothing for a small printed bracket.

The strategic version of the argument is learning rate. A team that can change a prototype in an afternoon runs four experiments a week; a team that reprints runs one. That is worth far more than the per-part savings — the same reasoning behind keeping the overall cost of prototyping low enough that you are never afraid to test something.

There is a limit. When modifications have piled up to the point where nobody can say with confidence what the object is, stop, model everything you have learned, and build one clean prototype as the new baseline. Teams that grew a model out of a self-built first version usually hit this point exactly once, and hitting it is a sign of progress rather than failure.

Projects House repairs, modifies, and upgrades existing prototypes — including models built elsewhere — and keeps the CAD in step with the physical part. Send us photos of what you have through the contact form.