Is It Within Tolerance?#

Comparing what was measured to what the part should be. A resistor drifted 15% from its marked value, a capacitor that’s lost half its capacitance, an inductor whose core is partially saturated — these parts “sort of work” and cause subtle, maddening failures.

Tolerance Specifications#

Resistors#

Most modern resistors are ±1% or ±5%. If a ±1% resistor measures outside ±5%, something is wrong.

Capacitors#

Inductors#

Tolerance Verification#

Identify the rated value and tolerance, measure out-of-circuit, calculate deviation:

(measured - rated) / rated × 100%

Example: A 4.7 kΩ ±5% resistor measures 4.52 kΩ:

• Deviation: (4520 - 4700) / 4700 × 100% = -3.8%
• Within ±5% tolerance — part is fine

Same resistor measuring 4.10 kΩ:

• Deviation: -12.8%
• Outside ±5% — resistor has drifted or been damaged

When Tolerance Alone Isn’t Enough#

Some circuits depend on precise values or matched ratios.

Voltage dividers and reference circuits: If both resistors are ±5% but one drifted high and the other low, the ratio error can be up to 10%. For ADC reference dividers, regulator feedback networks, and amplifier bias networks, the actual ratio matters more than individual values.

Filter circuits: RC and LC corner frequencies depend on component values. A ±20% capacitor shifts the corner frequency by ±20%. Calculate actual corner frequency: f = 1 / (2π × R × C)

Oscillator circuits: RC oscillators depend on component values. ±10% drift in R or C shifts frequency by up to ±10%.

Matched pairs: Differential amplifier input resistors and instrumentation amp gain resistors need to match each other. Two ±1% resistors could differ by up to 2% from each other.

Comparative Testing#

When the datasheet isn’t available or the marking isn’t trusted, compare against a known-good component of the same type. Measure both under the same conditions — they should be within the tolerance band of each other.

Tips#

• Use a meter at least 3–4x more accurate than the tolerance being verified — a ±0.5% DMM can’t meaningfully verify ±1% tolerance
• Measure at room temperature for meaningful comparison to rated values
• For ratio-dependent circuits, measure both components and calculate the actual ratio
• Calculate expected filter corner frequencies from measured values when debugging filter behavior

Caveats#

• DMM accuracy specification overlaps with tight tolerances — measurement uncertainty may exceed the tolerance band
• Temperature affects the reading — resistors have tempco, ceramics (X7R, Y5V) change value with temperature
• A 1% resistor at 3% drift isn’t “failed” by the 5% standard but is degraded beyond its original grade
• Temperature tracking matters in precision circuits — matched components should have the same tempco so they drift together
• Aging: components drift over time, especially in high-temperature environments; film resistors age less than carbon composition
• The “known-good” reference is only as trustworthy as its provenance — parts from original packaging are reliable; parts from a tape-labeled bin are not

In Practice#

• Resistor measuring outside its tolerance band indicates drift from aging, thermal stress, or physical damage
• Capacitor measuring low but within its wide tolerance (-20/+80%) may still be failing — ESR is often a better indicator than capacitance
• Filter cutting off at wrong frequency suggests R or C has drifted — measure components and calculate actual corner
• Amplifier gain that’s off by a few percent suggests feedback resistor drift — measure the ratio, not just individual values
• Regulator output voltage that’s off suggests feedback divider ratio has shifted — both resistors may have drifted in opposite directions
• Matched pair that differs by more than expected from each other (even if both are “within tolerance”) will cause offset or gain errors in precision circuits
• A precision measurement that drifts after the system has been running for several minutes commonly appears when a nearby power-dissipating component is thermally coupling into the measurement circuit — the drift follows the thermal time constant of the PCB, not any electrical time constant.
• A system that works correctly at room temperature but produces intermittent errors at elevated temperature is frequently showing a failure that propagates upward only when temperature pushes a marginal parameter past its threshold — the component-level drift is within spec at room temperature but exceeds the margin available at the subsystem level when thermal shifts are added.
• A replacement component that “fixes” the problem on one board but not another is frequently showing that the component wasn’t the root cause — the variation between boards (layout, soldering, component placement) is the actual mechanism, and the replacement component happened to shift a margin enough to work on the first board.
• A device that works at room temperature but fails in an enclosure reveals a thermal integration problem — the enclosure restricts airflow, and some subsystem is exceeding its thermal limits.
• A measurement that drifts during the first 20 minutes after power-up and then stabilizes often shows up when the circuit’s thermal environment is changing as it heats up — the system reaches thermal equilibrium and the drift stops. In the field, this warmup drift may be longer (cold start) or may never reach equilibrium (if the load varies continuously).
• A subsystem that meets its specification under test conditions but fails in the device commonly appears when the device-level conditions (supply noise, thermal coupling, loading) push the subsystem’s operating point to a region where its specification is no longer met — the subsystem was healthy in isolation and unhealthy in context.
• A circuit that requires specific component sourcing to work reliably is exhibiting an abstraction smell — the circuit’s behavior depends on component characteristics (parasitic properties, exact tolerance position, manufacturing process) that aren’t part of the component’s specified abstraction. The dependence on a specific supplier means the design margin is too thin for the specification tolerance.
• An IC that works in one circuit but not another, despite identical connections, often indicates that the IC is being treated at the wrong abstraction level — the second circuit has different operating conditions (load, temperature, input range) that violate assumptions invisible when the IC is treated as a simple component.