The term measurement accuracy is used frequently in manufacturing and inspection, but it is often confused with precision, error, uncertainty, resolution, and tolerance. Misunderstanding these concepts can lead to unsuitable instrument selection and inconsistent pass\/fail decisions.<\/p>\n\n\n\n
This article explains measurement accuracy from a practical metrology perspective, including its relationship with tolerances, major sources of error, evaluation methods, how to read instrument specifications, and ways to improve measurement reliability in the workplace.<\/p>\n\n\n\n
Before discussing measurement performance, it is essential to distinguish the basic terms used in metrology. Saying that accuracy is poor does not identify whether the average is biased, the readings are scattered, or the conditions change from one measurement to the next.<\/p>\n\n\n\n
Accuracy describes closeness to a true or reference value, while precision describes the consistency of repeated results. Error is the difference between a measured value and a true or accepted reference value. Uncertainty expresses the range within which the true value is considered likely to lie.<\/p>\n\n\n\n
Accuracy indicates how close the average measured value is to the true value. Incorrect calibration, an unsuitable reference, or a missing compensation value can shift all results in the same direction.<\/p>\n\n\n\n
Precision indicates how closely repeated measurements agree under the same conditions. Differences in positioning, applied force, vibration, electrical noise, and surface condition can increase scatter even when the average value is correct.<\/p>\n\n\n\n
A process can be accurate but not precise, or precise but not accurate. The first condition requires reduction of variation; the second generally requires calibration, correction, or review of the reference.<\/p>\n\n\n\n
Systematic error produces a consistent bias. Typical causes include zero offset, scale-factor error, uncompensated temperature effects, and misalignment of a fixture or reference surface. Calibration and compensation are often effective countermeasures.<\/p>\n\n\n\n
Random error changes from one measurement to another. It may result from reading variation, small changes in contact force, electrical noise, vibration, or uneven surface properties. Standardized procedures and stable environments are used to reduce it.<\/p>\n\n\n\n
Measurement uncertainty represents the range associated with a measurement result rather than a single error value. It is closely related to product acceptance and quality assurance because it describes the confidence that can be placed in a result.<\/p>\n\n\n\n
For example, a result of 40.000 mm with an uncertainty of \u00b10.030 mm means that the actual value may lie between 39.970 and 40.030 mm. A result inside the tolerance is not automatically safe when it is close to the limit.<\/p>\n\n\n\n
Uncertainty should include not only the instrument specification but also the effects of the procedure, fixture, environment, workpiece, data processing, and operator.<\/p>\n\n\n\n
Measurement accuracy is often used as a general expression for how correctly something is measured. In metrology, however, it becomes meaningful only when the reference, conditions, evaluation method, and acceptable level are clearly defined.<\/p>\n\n\n\n
A useful practical definition is the overall degree to which a measurement result obtained under specified conditions can be regarded as close to the true value. Measurement results always involve uncertainty, so accuracy must be considered as part of a complete measurement system.<\/p>\n\n\n\n
Catalog specifications usually describe instrument performance under conditions defined by the manufacturer. Actual measurement-system performance also depends on the sensor, display unit, fixture, mounting, measurement point, workpiece orientation and deformation, environment, and filtering or averaging settings.<\/p>\n\n\n\n
Consequently, an instrument that appears adequate on paper may not provide sufficient performance in the actual process. Verification with representative workpieces and operating conditions is therefore important.<\/p>\n\n\n\n
In practice, accuracy may be used to mean indication error, repeatability, resolution, or linearity. These are different characteristics. Specifications and discussions should state exactly whether the requirement concerns indication error, repeatability, reproducibility, linearity, or uncertainty.<\/p>\n\n\n\n
Tolerance is the permitted range of a dimension or other characteristic and is the basis for deciding whether a product passes or fails inspection.<\/p>\n\n\n\n
For example, a nominal value of 40 mm with a tolerance of \u00b10.1 mm permits values from 39.9 to 40.1 mm. Tolerances may apply to dimensions, thickness, angle, temperature, current, film thickness, or any other controlled characteristic.<\/p>\n\n\n\n
Manufacturing always produces variation because of machine resolution, tool wear, material differences, and temperature. Measurement also produces variation. Tolerances provide a practical balance among function, quality, manufacturability, and cost.<\/p>\n\n\n\n
The important factor in instrument selection is not the catalog number alone, but how small the measurement-system uncertainty is relative to the tolerance width.<\/p>\n\n\n\n
If uncertainty is too large, decisions near the tolerance limits become unstable. Good parts may be rejected, and nonconforming parts may be accepted. Reinspection, sorting, rework, and delayed shipment can increase the total process cost.<\/p>\n\n\n\n
A guard band narrows the acceptance range inside the specified tolerance by an amount related to measurement uncertainty. This reduces the risk of accepting a nonconforming part, although it can increase the risk of rejecting a conforming part.<\/p>\n\n\n\n
Insufficient measurement performance therefore appears not only as an inability to measure, but also as a reduction in the range where a confident decision can be made.<\/p>\n\n\n\n
Consider a nominal dimension of 40 mm with a tolerance of \u00b10.1 mm, giving a specification range of 39.9 to 40.1 mm. If the instrument error is treated conservatively as \u00b1E, the acceptance range is reduced by E at both limits.<\/p>\n\n\n\n
With an accuracy of \u00b10.001 mm, the acceptance range becomes 39.901 to 40.099 mm. At \u00b10.01 mm it becomes 39.910 to 40.090 mm, and at \u00b10.03 mm it becomes 39.930 to 40.070 mm.<\/p>\n\n\n\n
As measurement error increases, more products near the limits become difficult to classify confidently, even though the product tolerance itself has not changed.<\/p>\n\n\n\n
Higher-performance instruments generally cost more to purchase and maintain, but they may reduce scrap, rework, repeated inspection, shipment delays, and customer claims. Selection should be based on the total cost of measurement and incorrect decisions.<\/p>\n\n\n\n
A practical review should include tolerance width, the frequency of products near the limits, the financial impact of incorrect decisions, required throughput, and estimated measurement-system uncertainty.<\/p>\n\n\n\n
When results are unstable, replacing the instrument is not always the best first action. Causes should be investigated systematically in four groups: the instrument, environment, measurement method, and people.<\/p>\n\n\n\n
Important factors include resolution, indication error, linearity, drift, warm-up behavior, wear, play, sensor characteristics, and noise. Fine resolution does not guarantee accurate results.<\/p>\n\n\n\n
For contact instruments, worn contact tips, dirty measuring surfaces, and mechanical play can cause error. For optical and other non-contact sensors, surface roughness, gloss, color, reflectivity, working distance, and target angle can directly affect the signal.<\/p>\n\n\n\n
Temperature is one of the most important influences. The workpiece, fixture, and instrument expand or contract with temperature. Differences between the measuring room and production area, as well as local heating from handling, can be significant.<\/p>\n\n\n\n
Vibration from presses, conveyors, air-conditioning airflow, or an insufficiently rigid stand can disturb readings. Dust and oil films alter reference surfaces, while electromagnetic noise, poor grounding, and cable routing affect electrical measurements.<\/p>\n\n\n\n
Small differences in measurement position may capture taper, waviness, or local shape variation. The measurement point should therefore be defined by drawings, stops, guides, or dedicated fixtures.<\/p>\n\n\n\n
Contact force can deform thin or polymer workpieces. In non-contact measurement, changes in distance and angle alter the optical condition. Zeroing, reference-setting, measurement speed, averaging, and filtering must also be standardized and validated.<\/p>\n\n\n\n
The way a caliper or micrometer is applied, viewing angle, applied force, and selection of the reference surface may differ between operators. Large operator-to-operator differences usually indicate that the process requires better standardization rather than simply more individual skill.<\/p>\n\n\n\n
Work instructions should define measurement points, handling, the number of measurements, treatment of borderline results, remeasurement conditions, and escalation procedures.<\/p>\n\n\n\n
Measurement performance should be made visible using numerical indicators. Repeatability, reproducibility, and linearity are particularly useful for identifying whether variation comes from the instrument, setup, operator, or measurement range.<\/p>\n\n\n\n
Repeatability is the variation observed when the same operator measures the same workpiece repeatedly using the same instrument, fixture, method, and environment.<\/p>\n\n\n\n
It is commonly evaluated using standard deviation or range. Poor repeatability suggests influences such as positioning, contact force, vibration, electrical noise, or surface condition. Non-contact measurements should be evaluated using the actual target material whenever possible.<\/p>\n\n\n\n
Reproducibility describes agreement when conditions such as the operator, day, equipment, or fixture change. It is important in production environments with shift changes or multiple measuring stations.<\/p>\n\n\n\n
Measuring the same reference with several operators and on different days helps reveal where differences occur. A more systematic study leads to the concepts used in Gage R&R analysis.<\/p>\n\n\n\n
Linearity indicates how measurement error changes across the measuring range. Calibration at one point does not guarantee the same performance over the full range.<\/p>\n\n\n\n
Evaluation should use reference standards at several points, such as the lower, middle, and upper portions of the actual operating range. The results can support range selection, compensation, or an improved calibration method.<\/p>\n\n\n\n
Specification tables contain several performance terms. Comparing only the smallest number can result in selecting an instrument that does not meet the real requirement.<\/p>\n\n\n\n
Resolution is the smallest increment that an instrument can display or detect. A resolution of 0.001 mm means that the display changes in 0.001 mm steps; it does not mean that every displayed value is correct within 0.001 mm.<\/p>\n\n\n\n
An instrument may repeatedly display a finely resolved but biased value. Resolution should be checked together with indication error, repeatability, linearity, temperature characteristics, and uncertainty.<\/p>\n\n\n\n
A fixed specification such as \u00b11.0 \u00b0C is straightforward. A proportional expression such as \u00b1(3% + 5) \u03bcm must be calculated at the intended measurement value. At 50 \u03bcm, the error is \u00b1(50 \u00d7 0.03 + 5) = \u00b16.5 \u03bcm.<\/p>\n\n\n\n
Specifications such as \u00b1(2.0% rdg + 5 dgts) combine a percentage of the reading with a number of least-significant display digits. Always calculate the actual error range at the values used in your process.<\/p>\n\n\n\n
Linearity may be specified as \u00b1% of full scale. For a 10 mm range with \u00b10.5% F.S., the maximum deviation is \u00b10.05 mm, regardless of whether the measured value is close to 1 mm or 10 mm.<\/p>\n\n\n\n
Using a range much wider than necessary can therefore produce an unnecessarily large error contribution. Select a model and range suited to the actual operating region.<\/p>\n\n\n\n
Specifications may depend on temperature, warm-up time, averaging, sampling, working distance, target material, surface condition, and calibration status. Numbers from different products are comparable only when their conditions are aligned.<\/p>\n\n\n\n
Where possible, verify performance using actual workpieces and the intended fixture. Confirm repeatability, drift, response time, and sensitivity to expected environmental changes.<\/p>\n\n\n\n
Significant improvement is often possible before purchasing a more expensive instrument. Calibration, fixturing, standardization, and temperature management are especially cost-effective.<\/p>\n\n\n\n
Calibration determines the instrument’s deviation from a traceable reference and helps control systematic error. A calibration certificate alone is not sufficient; performance should also be checked under conditions close to actual use.<\/p>\n\n\n\n
Routine zero checks and measurements of a reference standard can reveal abnormal drift early. Calibration intervals should reflect usage frequency, past drift, required tolerance, and environmental conditions rather than being selected arbitrarily.<\/p>\n\n\n\n
Fixtures reduce variation by defining the measurement point, workpiece reference, instrument orientation, stroke, and applied force. Even simple stops and guides can significantly improve reproducibility.<\/p>\n\n\n\n
Work instructions should include diagrams of measurement points, positioning, measurement count, data-processing settings, and actions to take when results are abnormal or close to the acceptance limit.<\/p>\n\n\n\n
Dimensional measurement is commonly referenced to 20 \u00b0C. The workpiece, fixture, and instrument should be allowed to reach a stable and similar temperature before measurement.<\/p>\n\n\n\n
Useful practices include providing acclimatization time, avoiding direct sunlight and heat sources, minimizing prolonged handling with bare hands, and recording ambient temperature.<\/p>\n\n\n\n
Temperature compensation may be calculated from the material’s coefficient of thermal expansion, but it cannot fully correct nonuniform or local heating. Preventing temperature differences is generally more reliable than compensating for them afterward.<\/p>\n\n\n\n
Reliable measurement requires clear use of the terms accuracy, precision, error, uncertainty, resolution, and tolerance. Ambiguous specifications lead to unsuitable instruments and inconsistent decisions.<\/p>\n\n\n\n
Instrument selection should be based on measurement-system uncertainty relative to the tolerance, together with the cost of incorrect decisions and the required throughput.<\/p>\n\n\n\n
Measurement performance is determined by the entire system, including the instrument, fixture, environment, procedure, target, data processing, and operator. Repeatability, reproducibility, and linearity make the current condition visible.<\/p>\n\n\n\n
By combining traceable calibration, routine checks, effective fixturing, standardized procedures, and temperature control, manufacturers can improve both the reliability of pass\/fail decisions and the total cost of inspection.<\/p>\n","protected":false},"excerpt":{"rendered":"
The term measurement accuracy is used frequently in manufacturing and inspection, but it is often confused wit<\/p>\n","protected":false},"author":2,"featured_media":5357,"comment_status":"closed","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"_locale":"en_US","_original_post":"https:\/\/www.katsura-opto.com\/?p=4760","footnotes":""},"categories":[18],"tags":[],"class_list":["post-5470","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-column","en-US"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.katsura-opto.com\/wp-json\/wp\/v2\/posts\/5470","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.katsura-opto.com\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.katsura-opto.com\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.katsura-opto.com\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.katsura-opto.com\/wp-json\/wp\/v2\/comments?post=5470"}],"version-history":[{"count":1,"href":"https:\/\/www.katsura-opto.com\/wp-json\/wp\/v2\/posts\/5470\/revisions"}],"predecessor-version":[{"id":5471,"href":"https:\/\/www.katsura-opto.com\/wp-json\/wp\/v2\/posts\/5470\/revisions\/5471"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.katsura-opto.com\/wp-json\/wp\/v2\/media\/5357"}],"wp:attachment":[{"href":"https:\/\/www.katsura-opto.com\/wp-json\/wp\/v2\/media?parent=5470"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.katsura-opto.com\/wp-json\/wp\/v2\/categories?post=5470"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.katsura-opto.com\/wp-json\/wp\/v2\/tags?post=5470"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}