Reverse engineering from a sample

The four stages

• Measurement — calipers and micrometers for most work, a CMM or 3D scan for complex geometry or where a surface profile matters.
• Material identification — spectrometry where it matters, hardness testing, or inference from the application when the budget does not justify testing.
• Drawing production — a dimensioned drawing with tolerances and a material callout. This is the deliverable a shop quotes against.
• First article — one piece made and measured against the drawing before the rest of the run.

Deciding what is critical

On most parts, three to six dimensions carry the function: bore and shaft fits, sealing faces, mounting hole positions, and anything that sets a running clearance. Everything else can be held to a general tolerance and cost almost nothing.

If you do not make that call, the shop will — either by tightening everything, which is expensive, or by loosening everything, which produces a part that looks right and does not work.

Material, and how to avoid guessing wrong

Spectrometry on a sample settles an alloy question for a modest fee and is worth it on anything structural or heat-treated. For a bracket, inference from the application is fine.

Hardness is where reverse-engineered parts most often fall short: the geometry copies easily and the heat treatment does not. If the original was hardened, specify the process and the target hardness, and ask for the test result with the first article.

Cost, roughly

• Measurement and drawing from a simple sample: a few hundred renminbi, often absorbed into the quote.
• 3D scan and model for complex geometry: higher, quoted per part.
• Material analysis: modest per sample, and cheap relative to making the wrong part.
• First article with an inspection report: one piece at setup cost, which on a small run is a real fraction of the total.
• Production: the per-unit price, which is usually the smallest number in this list on a low-volume job.