Electronics · Math shelf · Optional — for the curious · build 2026.09.23-1644

Tolerance & Error

The idea

Every component is a promise with a plus-or-minus, and every measurement has its own wiggle. Engineering isn't expecting exact numbers — it's knowing how far off is fine.

The gold band is a promise

The fourth stripe on a resistor states its tolerance — how far the true value may sit from the labeled one. Gold means ±5% (silver ±10%, brown ±1% on fancier parts). Our kit is gold, so a "220 Ω" resistor is really a promise: somewhere between 209 and 231 Ω.

Labeled (nominal)±5% is…The promise
220 Ω±11 Ω209 – 231 Ω
1 kΩ±50 Ω950 – 1,050 Ω
4.7 kΩ±235 Ω4,465 – 4,935 Ω
10 kΩ±500 Ω9,500 – 10,500 Ω

The labeled value is called the nominal value — the name-value. Ohm a handful of "1 kΩ" resistors from the assortment and you'll read something like 987, 1,004, 992: all different, all keeping the promise.

Percent difference — the honest comparison

% diff = (measured − nominal) ÷ nominal × 100

Worked: you measure 1,032 Ω on a nominal 1 kΩ. (1,032 − 1,000) ÷ 1,000 = 0.032 → +3.2%. Inside ±5%: the part is healthy. This one formula turns "is that close?" from a feeling into a number.

The meter wiggles too

The meter's spec sheet gives its own honesty range — hobby meters typically promise about ±(0.5% + 2 digits) on DC volts. The "2 digits" means the last displayed digit may be off by two counts: a true 5.00 V may read 4.97 to 5.03. So the final digit of any reading is always a little soft — watching it flutter is normal, not a bug.

When disagreement is a bug

Measured vs. expectedVerdict
within a few %normal — tolerance and meter wiggle
off by 2×not tolerance: wrong resistor grabbed, or a circuit bug
off by ~1,000×read the display's unit again — vs Ω (the Milli, Kilo, Mega sheet)

Tolerance never explains a factor of ten. Small disagreements are physics; big ones are information — that's the drop check's whole philosophy on the Debugging Loop card.

Errors travel through formulas

Build a divider from two ±5% parts and the prediction inherits their spread. Worst case for a 1 kΩ / 1 kΩ divider: one part at 950, the other at 1,050 puts the tap anywhere from 2.38 to 2.63 V instead of exactly 2.50. In practice both parts usually sit much closer — but this is why a measured 2.46 V against a predicted 2.50 is a success, not a miss.

Which sets the reporting rule: don't write more digits than your parts can back up. Predicting "2.51694 V" from ±5% resistors is fiction past the first three digits; "≈ 2.5 V" is the honest answer.

The math, for the curious — why kit values look so odd

1.0 — 2.2 — 4.7 — 10: strange numbers, until you see the design. Standard resistor values are spaced by equal ratios, not equal gaps, so the ±-ranges tile the number line with no orphan values. The full E12 series puts twelve steps in each ×10 decade, each step ×1.21 — the twelfth root of 10 — exactly the way a piano splits each octave into twelve equal-ratio semitones. Resistor drawers and keyboards are tuned the same way.

Practice

Answers at the bottom — and problems 1–3 are checkable against real parts with the meter.

  1. What range does a gold-band 470 Ω promise?
  2. You measure 4.55 kΩ on a nominal 4.7 kΩ. Percent difference — and is the part healthy?
  3. You measure 2.2 kΩ on what you thought was a 220 Ω. Tolerance?
  4. A nominal 10 kΩ measures 10,443 Ω. Percent difference — healthy?

Answers

1. ±23.5 → 446.5 – 493.5 Ω. 2. (4,550 − 4,700) ÷ 4,700 ≈ −3.2% — healthy. 3. No — that's 10×. Either the display said kΩ and you wrote Ω, or the stripes are red-red-red (2.2 kΩ), not red-red-brown. Check both before blaming the part. 4. +4.4% — inside the gold promise, though barely; a part that close to the edge is still a kept promise.