Why DfM rules are usually broken
Design-for-manufacturing as taught in university is a list of rules: minimum wall thicknesses, draft angles, hole-to-edge distances. Each rule is correct. The way they get applied in practice — at the end of design, by a CAD operator who isn't the original designer, against a checklist — produces parts that "follow the rules" but are still unmanufacturable in detail.The patterns below are what we've learned from being on the receiving end of designs that "passed DfM review" and still couldn't be made.
Rule 1: design with the process in mind, not against a checklist
A part designed for milling looks different from a part designed for casting, even if both are manufacturable. The design intent should commit to a primary process early. Common failure: a "process-agnostic" design that ends up cast (because casting won the cost negotiation) but with sharp internal corners that should have been machined out.Pick the process. Design to its strengths and constraints. Re-process if the supplier mix changes.
Rule 2: tolerance budget vs. tolerance wishlist
Every dimension on the drawing needs a tolerance, but most of them don't matter. The discipline:- Identify the 5-10 dimensions that are functionally critical. Tight tolerance on those.
- Loose tolerance everywhere else (often "general tolerance per ISO 2768-mK" is enough).
- Stack-up analysis on the critical chain.
The drawing with every dimension at ±0.05 mm is unmanufacturable at any reasonable cost. The drawing with the right dimensions at ±0.05 mm and the rest at ±0.5 mm costs a tenth as much and works as well.
Rule 3: the manufacturer is part of the design team
Send the part to the manufacturer at concept stage, not after the design is "frozen". A 30-minute conversation with the foreman / programmer / toolmaker before the design is finalised saves 3 weeks of rework after.What we look for in this conversation:
- "What would you change to make this cheaper?"
- "Where do you expect to have problems?"
- "What's the longest lead-time item in this design?"
- "What's the highest-risk feature?"
The answers reshape the design more than internal review ever does.
Rule 4: tools, not just parts
A design review that doesn't consider the tooling, fixturing, and gauging is half a review. Questions:- Can this part be held in a vise / chuck / fixture without distortion?
- Are there features that need a custom fixture? Has it been quoted?
- How is the part inspected? Manually with a calliper, with a CMM, or in-line with a probe?
- What's the gauge that confirms the critical dimensions? Has it been built and validated?
We've shipped designs where the part was elegant and the tooling cost twice the part cost over a production run. The design review must include the tooling cost.
Rule 5: tolerances on the assembly, not just the parts
Two parts each within tolerance can stack into an assembly outside tolerance. The classic stack-up problem. The designer who tolerances the assembly first and the parts second produces designs that fit together.Run the stack-up at design freeze. If it fails, the design isn't done. If it passes with margin, the part tolerances can usually be loosened (= cheaper) without affecting the assembly.
