Why do electronic components have such odd values? (2021)

Original Article ↗ Hacker News Discussion ↗

Clarifying the 70 Ω example

  • Several commenters note the article’s last example seems numerically wrong: 33 Ω + 47 Ω = 80 Ω, not 70 Ω.
  • Common view: it’s likely a typo and should be 22 Ω + 47 Ω ≈ 69 Ω.
  • Others joke that 80 Ω is within common loose tolerances anyway, as are 68 Ω and 75 Ω.

Tolerance, statistics, and design practice

  • Formal derivation: when adding resistors with the same percentage tolerance in series, the resulting percentage tolerance stays the same; absolute error adds, denominator (nominal resistance) also grows.
  • Parallel combinations with identical percentage errors also retain that percentage; mixed high/low errors can partially cancel.
  • Strong debate over “tolerance vs statistics”:
    • One side stresses tolerance is a worst‑case contractual bound; good engineering designs to that, not to probability.
    • Others note that random errors statistically average out (central limit theorem), so large series/parallel networks can have smaller expected relative error, though real distributions may be skewed or bimodal.
  • Distinction drawn between tolerance (spec) and measurement uncertainty (instrument error).

Manufacturing, binning, and resistor technologies

  • Discussion of resistor types: carbon film, metal film, wirewound, thin‑film, foil; choice affects accuracy, tempco, inductance, and cost.
  • Some recall the idea that “5% parts are 1% parts that failed binning”; others argue this is usually uneconomic for very cheap parts and often a myth today.
  • For tight tolerances and low tempco, manufacturers use different materials, geometries, and laser trimming; very high‑precision parts become extremely expensive.

E‑series preferred numbers and Renard numbers

  • Commenters restate that E‑series values form geometric/logarithmic sequences rounded to two digits.
  • The key idea praised: “tolerance overlap” — adjacent preferred values’ ±tolerance bands just touch, ensuring any target value is within a fixed relative error of some series value.
  • Some point out the article blurs “usage tolerance” (acceptable design error) with fabrication tolerance of real parts.
  • The Renard cable story is recognized as the same geometric / overlap principle applied to mechanical sizes.

Practical usage and power ratings

  • Many say E12 (or even E6) plus 1% resistors is enough for most work, especially digital; analog and metrology may need finer series and matched networks.
  • Trimmers and digital calibration are common where precision beyond basic resistors is needed.
  • Power rating is usually inferred from package size; designers often standardize on the highest rating they expect to need in a footprint.