---
title: "What Is DFM? The Design Rules That Cost Real Money"
description: "DFM means designing a part so it is cheap and reliable to make. Real examples of features that multiply cost, and how to get feedback early."
canonical: https://openspindle.com/blog/what-is-dfm
author: "Tom"
datePublished: 2026-08-24T00:00:00.000Z
dateModified: 2026-08-24T00:00:00.000Z
category: "Process & DFM"
---

# What Is DFM?

Design for manufacturability, or DFM, is the practice of designing a part so that it is straightforward, fast and repeatable to produce. In machining, most of the cost sits in a small number of design choices: how tight the tolerances are, how many setups the part needs, how deep and how sharp the internal corners are, and how thin the walls get. Change those four things and the quote moves substantially.

## The cost drivers, ranked

| Design choice | Why it costs | Cheaper alternative |
| --- | --- | --- |
| Tight tolerances everywhere | Slower cuts, more inspection, more scrap | Tighten only the features that mate |
| Multiple setups | Every re-fixture is labor and a stack-up risk | Design so features share as few faces as possible |
| Sharp internal corners | A round cutter cannot make a sharp inside corner | Add a radius, ideally larger than the smallest cutter |
| Deep, narrow pockets | Long thin tools chatter and must run slow | Keep depth under about 4x the pocket width |
| Thin walls | Vibration, deflection, scrap parts | Keep metal above 0.8 mm, plastic above 1.5 mm |
| Undercuts | Needs special tooling or a second operation | Redesign as an assembly or open the geometry |
| Cosmetic finish on all faces | Extra operations and handling | Specify finish only where it is visible |

## Internal corner radii, the most common miss

End mills are round. That means any internal corner produced by milling carries a radius equal to the cutter radius. A truly square internal corner is achievable, but not by milling: it takes wire or sinker EDM, broaching, or a corner relief cut that removes the material behind the corner. Each of those is a separate operation with its own cost, which is why a square internal corner on a milled part is a DFM problem rather than an impossibility. If your model shows a sharp internal corner and says nothing else, the shop has to interpret it, ask you, or reach for the smallest cutter that will fit.

Small cutters are the problem. A 2 mm end mill removes material far slower than a 12 mm end mill and breaks far more easily. Specifying a small corner radius quietly forces the shop into a slow, fragile tool for the whole pocket.

The practical rule is to make the internal corner radius at least one third of the pocket depth, and to use the largest radius the design tolerates. Going from a 2 mm to a 6 mm corner radius on a deep pocket can cut the machining time meaningfully.

## Tolerances are not free

Standard machining tolerance is around ±0.005 inch, or ±0.13 mm. Most shops hold that without special effort. Tightening to ±0.001 inch means slower finishing passes, temperature control, more frequent inspection and higher scrap. It is not twice the cost, but it is a real multiplier, applied to every feature you tighten.

The discipline is to identify the features that actually mate with something and tighten only those. Everything else gets the general tolerance in the title block. A drawing where every dimension carries a tight tolerance tells the shop you have not decided what matters, and they will quote defensively.

## Setups are the hidden multiplier

Every time a part comes out of the vise and goes back in a different orientation, someone has to re-fixture it, re-probe it and re-establish the datum. That is labor, and it introduces a tolerance stack between features cut in different setups.

A part that can be machined in one setup is dramatically cheaper than the same part with features on five faces. Early in a design, it is often nearly free to move a hole or a boss to a face that is already being machined. Late in a design it is not free at all, which is why the DFM conversation belongs at the concept stage.

## What a good DFM review actually gives you

A DFM review is not a shop saying "no problem" to your drawing. It is a specific, itemized list of the features that are driving your cost and your risk, with an estimate of what changing each one would save.

A useful review names the feature, says why it is expensive, proposes a specific alternative, and quantifies the benefit. A review that says only "tolerances are tight" is not actionable. One that says "the ±0.001 inch callout on the bore forces a grinding operation; ±0.003 would let us hold it on the mill and remove a setup" is.

If you receive a quote on a complex part with no manufacturability comments at all, that is a signal, and usually not a good one.

| A weak DFM response | A useful one |
| --- | --- |
| "Looks fine, we can make it" | Names each cost-driving feature specifically |
| "Tolerances are tight" | Identifies which callout, and what it forces |
| "This will be expensive" | Estimates the saving from each proposed change |
| Silence on a complex part | Flags what it could not interpret and asks |

## How to get DFM feedback before you are committed

The highest-leverage hour in a hardware program is a manufacturing engineer reading your drawing before anything is cut. It is also the hour teams skip most often, because the schedule is tight and the review feels like a delay.

It is not a delay. Every change identified before tooling is cheap; the same change after tooling costs a tool. On a molded part the difference can be tens of thousands.

Send the model early, in a native or STEP format, along with what the part actually has to do. Context matters: a shop that knows the bore is a bearing fit will treat it differently from one that just sees a tight tolerance. Say what your volumes will be, because the right design for fifty parts is not the right design for fifty thousand. And ask explicitly for manufacturability notes rather than only a price.

## DFM outside machining

The principle is identical in every process. The specific rules are not, and applying machining habits to a molded or fabricated part produces expensive surprises.

| Process | The features that drive cost | The cheap fix |
| --- | --- | --- |
| CNC machining | Tight tolerances, setups, sharp internal corners, deep pockets | Fewer setups, larger corner radii, general tolerances |
| Sheet metal | Tight bend tolerances, features close to bends, many unique bends | Keep holes away from bend lines, reuse one bend radius |
| Injection molding | Undercuts, thick sections, no draft, sharp corners | Add draft, keep walls uniform, design out side actions |
| 3D printing | Unsupported overhangs, trapped supports, tiny features | Orient for self-support, give supports access |
| Casting | Non-uniform sections, thin ribs, tight as-cast tolerances | Uniform wall thickness, machine only the critical faces |
| Turning | Long slender parts, deep small bores, tight concentricity | Shorter length-to-diameter, relieve non-critical diameters |

*The most expensive mistake across all of them is the same: designing the part before deciding the process.*

## The wall thickness rule nobody applies consistently

Uniform wall thickness is the single most transferable DFM principle, and it matters for a different reason in every process.

In injection molding, thick sections cool more slowly than thin ones, and the difference pulls the part out of shape as it shrinks. That is what sink marks and warp actually are. A part with a 4 mm boss on a 2 mm wall will usually show a dimple on the opposite face. Process tuning, gas assist and a textured finish can reduce or disguise it, but they are working against the geometry. The reliable fix is to core out the boss so the wall stays uniform.

In casting, the same physics produces shrinkage porosity in the thick section, which is worse because it is internal and invisible.

In machining, uniform walls matter for a different reason: residual stress. Machining away material unevenly lets locked-in stress redistribute, and the part moves after it comes out of the vise. On large plates this is why a part measures correct on the machine and wrong the next morning.

In sheet metal the equivalent rule is a single material thickness and a single bend radius wherever possible, because each additional radius is a tooling change.

## Related reading

- [CNC machining tolerance chart](https://openspindle.com/reference/cnc-machining-tolerance-chart.md)
- [GD&T symbols reference](https://openspindle.com/reference/gdt-symbols-reference.md)
- [Tolerance stack-up analysis](https://openspindle.com/blog/tolerance-stack-up.md)
- [CNC machining capabilities](https://openspindle.com/capabilities/cnc-machining.md)

## Frequently asked questions

### When should DFM feedback happen?

Before tooling, and ideally at the concept stage. The single highest-leverage hour in a hardware program is a manufacturing engineer reading the drawing before anything is cut. Late DFM feedback means either accepting the cost or redoing the design.

### Does DFM apply to 3D printed parts?

Yes, with different rules. Printing cares about overhang angles, support removal access, minimum feature size and build orientation rather than setups and cutter radii. The principle is identical: the design decides the cost.

### How do I get DFM feedback if I do not have a shop yet?

Send the model to whoever will quote it and ask explicitly for manufacturability notes. A shop that returns a quote with no comments on a complex part has probably not read it closely.

### What should I send to get useful DFM feedback?

The model in a native or STEP format, plus what the part actually has to do and what your volumes will be. Context changes the advice: a shop that knows a bore is a bearing fit treats it differently from one that just sees a tight number, and the right design for fifty parts is not the right design for fifty thousand.

### Why do internal corners have to be rounded?

End mills are round, so any corner produced by milling carries a radius equal to the cutter radius. A square internal corner is not impossible, it just needs another process such as wire EDM or broaching, or a corner relief cut, and each adds cost. Specifying a small radius quietly forces the shop into a small, slow, fragile tool for the entire pocket, so using the largest radius your design tolerates is one of the cheapest cost reductions available.

### Does DFM matter at prototype quantities?

Less than at volume, but it still matters, and it matters most for what happens next. A prototype designed with no thought to manufacturing frequently cannot be scaled without a redesign, which means the tooling you eventually buy is for a part you have just changed.