How to Reverse Engineer an Obsolete Part When There Are No Drawings

A part breaks. The manufacturer stopped making it years ago, or went out of business, and nobody has a drawing. The machine it belongs to still works fine apart from that one piece. This is the situation reverse engineering exists for.

Reverse engineering means working backward from a physical part to the information needed to make a new one: a CAD model and a drawing a machine shop can quote and build from. Done well, it also fixes whatever made the original fail.

Check the cheaper options first

Reverse engineering costs money. Before paying for it, spend an hour ruling these out:

• Old stock. Search the part number on eBay, surplus sellers, and dealers who carried the brand.
• A cross-reference. Many "obsolete" parts are standard components with a brand's number stamped on them. Bearings, seals, and fasteners usually have a current equivalent.
• The manufacturer or its successor. Companies that were bought out sometimes still hold drawings for old product lines.
• Owner forums and clubs. Someone may already have made a replacement, or have the drawing.

If all of those come up empty, reverse engineering is the next step.

Why copying the shape is not enough

The obvious approach is to measure every surface of the old part and draw exactly what you measured. That approach has three problems.

The part you are holding is worn, and it may be broken. Measure a worn bore exactly and you have drawn a worn bore. The new part comes back loose on day one.

The original was made to a tolerance, not to a single number. Your sample is one part from somewhere inside that range. The drawing has to say which dimensions matter and how much they are allowed to vary, or the shop has no way to know. That is the job of a proper drawing package.

The part has to fit something. A dimension that looks arbitrary on its own usually exists because of a mating part: a shaft, a bolt pattern, a gasket face. Those relationships are where a tolerance stack-up goes wrong if nobody checks them.

So the real work is working out what each feature does, and drawing the part as it was meant to be, not as it is now.

How the process works

1. Understand what the part does. Where it mounts, what it touches, what load it carries, and how it failed. If possible I want to see the mating parts or the assembly, not just the broken piece.

2. Measure it. I measure with calipers, micrometers, and gauge pins. Gauge pins are the reliable way to size a hole: you find the largest pin that passes through, and that is the hole. For parts made of flat faces, bores, and slots, which covers most machined parts, these tools are accurate enough to build a production drawing.

3. Work out design intent. A bore that measures 0.5007" was almost certainly designed as 0.500". A bolt pattern that measures slightly off a round number almost certainly was a round number. Where wear has changed a surface, I take the size from an unworn area or from the part it mates with.

4. Model it and draw it. The part is built as a parametric model in SolidWorks, then detailed on a drawing with the tolerances that actually matter for fit and function.

5. Fix the failure while you are there. If the original broke because of a weak section, a poor material choice, or a sharp corner, this is the cheapest moment to change it.

Example: a rototiller fuel bowl

The fuel bowl on a rototiller was made from plastic, and it kept cracking. Replacing it with another plastic bowl would have meant replacing it again.

I reverse engineered the bowl and redesigned it to be machined from steel. Same fit on the machine, far longer life. The replacement fixed the reason the part kept failing instead of copying it.

That is the difference between a copy and a reverse engineered part. A copy reproduces the original, flaws included.

What you get back

• A native SolidWorks model, plus a STEP file any shop can open.
• A dimensioned drawing with tolerances, ready for a machine shop to quote.
• A material recommendation, especially if the original material was part of the problem.

You own the files. If the part fails again in ten years, you send the drawing to a shop instead of starting over. If you need help getting it made, I can also handle prototyping and fabrication through a Treasure Valley shop network.

What it costs

My rate is $125 an hour, with a 4-hour minimum per project.

That covers the engineering: measuring, modeling, and the drawing. Making the part is a separate cost from the machine shop. Every project starts with a capped estimate, so you know the maximum before any work begins.

The biggest cost drivers are the number of features, how many of them have to fit other parts, and whether the part needs redesigning rather than just recreating.

Is it legal to reverse engineer a part?

This is general information, not legal advice. In the US, measuring a part you own to make a replacement for your own equipment is generally considered acceptable. The risks increase in three situations:

• The part is still under patent. Reverse engineering is not a defense to patent infringement.
• You signed something. A contract or license that prohibits reverse engineering can apply even when the law otherwise would not stop you.
• You plan to sell the parts. Making replacements to sell carries more risk than making one for your own machine. Talk to a lawyer first.

I do not take on parts that are still under patent.

When not to reverse engineer a part

• Life-critical parts. Aircraft components, lifting and rigging hardware, brakes, and anything where a failure could hurt someone need certification and testing that a copied part does not have. Go to the manufacturer or a certified supplier. I do not take these on.
• Freeform or sculpted surfaces. Cast and molded parts with organic curves are better captured with a 3D scanner. My work is measured with hand tools, which suits machined geometry, so I will tell you if a scan-based provider is the better fit.
• A current part is available. If the cross-reference search turns up a replacement, buy it.

If you are not sure whether your part is a good candidate, send me photos and a description of what it does and how it failed. I will tell you whether it is worth reverse engineering, and roughly what it would cost.