mfg-machine-elements-mechanisms-and-tolerances

Manufacturing and Mechanical Engineering: Machine Elements, Mechanisms and Kinematics, and Tolerances and GD&T

Part 2 of 5 of the Manufacturing and Mechanical Engineering for Software Devs reference (plugin manufacturing-mechanical-engineering-for-software-devs), covering §6–§8. Sibling skills: mfg-mechanics-stress-fatigue-and-materials (§0–§5), mfg-process-families-machining-additive-and-moulding (§9–§16), mfg-dfm-metrology-plm-npi-and-what-transfers (§17–§24), mfg-reference (§25–§30). Section numbers are shared across the set; a reference written as §N → skill points into that sibling skill.

Currency: The mechanics is settled. Two areas moved. See §25 → mfg-reference for additive manufacturing after its correction, and industrial robotics adoption.

⚠️ Companion to a civil/industrial engineering reference. ⚠️ That one covers flow, constraints and safety systems; this one is about physically MAKING things — and the lessons are different.

⚠️ The governing asymmetry: in software the marginal cost of a change is near zero and in manufacturing it is enormous. ⚠️ Almost every practice in this document is a response to that one fact — tolerances, DFM, PLM, EVT/DVT/PVT gates, and the reason mechanical engineers appear conservative to software people.

⚠️ GOTCHA boxes mark the intuitions that don't survive contact with atoms.

The three ideas that organize this document:

1. ⚠️ NOTHING IS EXACT — everything is a tolerance (§8). The single biggest mental shift for software people: there is no equality in the physical world, only distributions, and a design that requires exactness is a design that fails.
2. ⚠️ The PROCESS determines the design, not the other way round (§9 → mfg-process-families-machining-additive-and-moulding, §17 → mfg-dfm-metrology-plm-npi-and-what-transfers). Geometry that's free in CAD can be impossible, or absurdly expensive, to make.
3. ⚠️ FATIGUE kills things that were never overloaded (§4 → mfg-mechanics-stress-fatigue-and-materials). Most mechanical failures happen far below the static strength, after many cycles — the physical analogue of a bug that only appears after a million requests.

---

§6. Machine Elements

⚠️ Most mechanical design is SELECTING standard components, not designing new ones.

⚠️ FASTENERS  ⚠️ a bolted joint works by CLAMPING — preload creates
   friction that carries the load, and the bolt should ideally not
   see shear directly. ⚠️ This is why TORQUE SPEC matters and why
   under-torqued bolts fatigue and fail
   ⚠️ Torque is a poor proxy for preload (friction scatter is large),
   which is why critical joints use angle or stretch measurement
⚠️ BEARINGS  rolling vs plain vs fluid film. ⚠️ Rated by L10 life —
   a STATISTICAL life at which 10% have failed, not a guarantee
⚠️ GEARS  ratio, module/pitch, ⚠️ backlash (lost motion on reversal —
   a real problem in positioning systems)
SPRINGS · SEALS (⚠️ dynamic seals are a common failure point) ·
COUPLINGS · belts and chains

⚠️ Standard parts exist and should be used: ⚠️ ISO/ANSI threads, standard bearing sizes, stock material thicknesses. ⚠️ A custom fastener is a supply chain liability forever (§22 → mfg-dfm-metrology-plm-npi-and-what-transfers). ⚠️ Threadlocking, prevailing-torque nuts and lock washers — ⚠️ and note that ordinary split lock washers have been shown to be largely ineffective, which is folklore that persists.

---

§7. Mechanisms and Kinematics

⚠️ Degrees of freedom, linkages (⚠️ four-bar being the workhorse), cams, and the distinction between kinematics (motion) and dynamics (forces causing it).

⚠️ GOTCHA — EXACT CONSTRAINT (kinematic) design is the concept software people find most surprising and most useful. ⚠️ A rigid body has six degrees of freedom; constrain each exactly once and the assembly is deterministic. ⚠️ OVER-CONSTRAINT — the intuitive "add more bolts and pins to make it solid" — forces parts to fight each other, transmits manufacturing variation into stress, and produces assemblies that bind or warp unpredictably. ⚠️ The classic example: a three-legged stool never rocks; a four-legged one does, because it's over-constrained. ⚠️ The software analogue is real: redundant sources of truth don't add robustness, they add contradiction.

⚠️ Vibration and resonance: ⚠️ every structure has natural frequencies; excite one and amplitude grows until damping or failure limits it. ⚠️ Modal analysis, damping, tuned mass dampers, and the practical rule to keep excitation frequencies away from natural ones.

---

§8. ⚠️ Tolerances and GD&T

⚠️ THE conceptual shift for software people. In the physical world there is no equality — only distributions. ⚠️ A "10 mm" hole is never 10 mm; it is 10 mm ± something, and the design must work across the whole range.

⚠️ TOLERANCE  the permitted variation. ⚠️ TIGHTER TOLERANCE COSTS
   MORE, often NON-LINEARLY — halving a tolerance can multiply cost
   several-fold by forcing a different process (§9) or added
   inspection. ⚠️ Over-tolerancing is the most common and most
   expensive novice error in mechanical design
⚠️ FITS  clearance · transition · interference (⚠️ press fits, where
   the parts are deliberately the "wrong" size)
⚠️ STACK-UP  ⚠️ tolerances ACCUMULATE across an assembly
   ⚠️ WORST CASE  sum of all tolerances — safe, expensive, and
      astronomically unlikely
   ⚠️ STATISTICAL (RSS)  root-sum-square — realistic, and it
      ASSUMES independence and centred distributions, which
      real processes often violate
⚠️ GD&T (ASME Y14.5 / ISO GPS)  ⚠️ specifies FUNCTION rather than
   just dimensions: form (flatness, straightness), orientation
   (perpendicularity, angularity), location (⚠️ TRUE POSITION),
   and runout — all relative to explicitly declared DATUMS
⚠️ MMC / BONUS TOLERANCE  ⚠️ as a feature departs from maximum
   material condition, extra positional tolerance becomes
   available — because the ASSEMBLY still works. This is
   tolerancing the FUNCTION, not the number

⚠️ Why GD&T exists at all: ⚠️ plus/minus dimensioning is ambiguous about what matters and creates square tolerance zones where the function wants round ones. ⚠️ GD&T states the design intent unambiguously, which is exactly what a specification should do. ⚠️ DATUMS are the key idea: ⚠️ they define how the part is referenced and therefore how it is measured and fixtured — and inconsistent datums between design, manufacture and inspection is a recurring source of parts that "measure fine" and don't fit.

---

PART II — MAKING THINGS