On this page
A printed-resistor reading near a tolerance boundary can be inside the drawing limits while its measurement uncertainty extends outside them. A guard band changes the acceptance interval used for the measurement decision; it does not change the specified resistor tolerance. The rule must be selected deliberately, applied to both boundaries where relevant and reported without confusing non-acceptance with a proven physical defect.
Measurement purpose
Apply and verify an explicitly agreed uncertainty-based acceptance rule for a defined printed-resistor measurement.
Specimens and conditions
- Resistance state
- Defined terminals, temperature, excitation, stabilization and conditioning history
- Acceptance basis
- Drawing limits and applicable uncertainty statement with authorized rule version
Equipment and records required
- Resistance method: Suitable calibrated measurement chain and justified method-specific uncertainty
- Rule verification: Independent calculations and numerical boundary challenge cases
Method sequence
- Definition
Separate tolerance, process variation and measurement uncertainty
Record: Measurand and rule inputs
- Calculation
Construct both acceptance boundaries under the approved rule
Record: Acceptance interval and feasibility check
- Disposition
Apply precision and category rules consistently
Record: Value, uncertainty and conformity outcome
Decision and uncertainty
Apply only the authorized guard-band equation and reporting categories; non-acceptance under a band is not automatically proof of physical nonconformity.
Coverage assumptions, range dependence and correlated contributions must be justified before assigning numerical risk.
The measurement owner establishes uncertainty; the customer and designated quality authority approve the decision rule.
Traceable outputs
| Record | Required contents |
|---|---|
| Decision calculation | Tolerance limits, uncertainty, acceptance limits, retained value and comparison result |
| Implementation evidence | Equality, rounding, unit and variable-uncertainty challenge outcomes |
Method review decisions
- Keep tolerance limits and measurement acceptance limits separate.
- Use uncertainty for the actual resistance measurement state.
- Define equality, rounding and unresolved-result handling before testing.
Identify the resistance quantity before drawing a guard band
Specify the measured terminals, specimen temperature, excitation, stabilization and relevant conditioning history. A guard band cannot reconcile resistance measured on different electrical boundaries or at different physical states. If the drawing specifies a normalized resistance, the approved normalization and its uncertainty belong to the measurement definition.
Distinguish drawing tolerance from process spread and measurement uncertainty. Tolerance is the permitted product interval. Process variation describes the manufactured population. Measurement uncertainty describes the knowledge of a particular measured quantity under the method. A control-chart limit or the standard deviation of recent production is not automatically the uncertainty to use in an acceptance rule.
Define a two-sided acceptance interval explicitly
Let the drawing permit resistance from L to H. One possible agreed rule accepts only when the reported value y, expanded by a specified uncertainty U in both directions, lies within that interval. Under a symmetric uncertainty model, this gives an acceptance interval from L plus U to H minus U.
This is an illustrative rule, not a universal requirement. Other decision rules may be specified by the customer, method or governing standard. Record why the selected rule is appropriate, which uncertainty statement it uses and what happens outside the acceptance interval. Do not select the rule after inspecting a borderline lot.
Accept when L + U ≤ y ≤ H − U
- L and H are the lower and upper drawing tolerance limits in ohms.
- y is the reported resistance estimate under the specified measurement state.
- U is the expanded measurement uncertainty used by this particular agreed rule.
Symmetric uncertainty interval y ± U and a rule requiring that interval to lie within both tolerance limits; other rules require their own equations.
Work both limits in a 100-ohm illustration
Assume a hypothetical resistor requirement of 100 ohms plus or minus one ohm, so L is 99 and H is 101. If the applicable U is 0.2 ohms, the stated rule accepts measured values from 99.2 to 100.8 ohms inclusive. The drawing interval remains 99 to 101; only the decision interval has narrowed.
A reading of 100.85 ohms lies inside the drawing interval but is not accepted by this rule. Calling it physically above 101 ohms would be incorrect. Its treatment must follow the agreed reporting categories and disposition process. The example provides arithmetic, not a company resistor tolerance, uncertainty capability or measured lot result.
| Measured resistance | Uncertainty interval | Decision under the stated rule |
|---|---|---|
| 99.10 Ω | 98.90 to 99.30 Ω | Not accepted; interval extends below L |
| 99.20 Ω | 99.00 to 99.40 Ω | Accepted with inclusive endpoints |
| 100.50 Ω | 100.30 to 100.70 Ω | Accepted |
| 100.80 Ω | 100.60 to 101.00 Ω | Accepted with inclusive endpoints |
| 100.85 Ω | 100.65 to 101.05 Ω | Not accepted; interval extends above H |
| 101.30 Ω | 101.10 to 101.50 Ω | Not accepted; entire interval is above H |
Use the method’s uncertainty, not a convenient instrument number
The applicable estimate may include calibration, temperature knowledge, contact boundary, excitation, repeatability and data reduction. Which contributions matter depends on the resistance range and procedure. Obtain a justified method-specific uncertainty estimate rather than copying the display resolution or the most favourable accuracy line from an instrument specification.
If uncertainty varies by range or measured value, the acceptance boundary may not be a fixed subtraction using one constant U. Evaluate the actual rule consistently. Do not reduce U because a part is close to rejection, and do not average repeated readings to reduce contributions that are shared by every reading, such as a common temperature error.
Recognize when the rule leaves no useful acceptance region
For the symmetric example, the acceptance interval has width H minus L minus twice U. If U equals half the tolerance width, only the centre remains under inclusive arithmetic. If U is larger, no measured value can satisfy the interval-containment rule. That is a measurement-and-decision incompatibility, not evidence that every manufactured resistor is defective.
In that situation, review whether the uncertainty estimate, measurement method and selected rule are appropriate to the required decision. Improving the method can be more useful than repeatedly measuring the same parts. Changing the product tolerance requires design authorization; changing the rule requires the relevant agreement. Neither should be hidden inside an inspection spreadsheet.
Keep rounding from moving the decision boundary
Apply the agreed numerical procedure to retained values with sufficient precision. Specify whether rounding occurs before or after the comparison and how equality is handled. A display rounded to 100.8 ohms could represent a retained value above the 100.8 acceptance limit, so display appearance alone must not control the result.
Challenge the implementation with values exactly on, just below and just above both limits. Include any supported uncertainty changes and unit conversions. Preserve the rule version with the result. This numerical check verifies implementation only; it does not establish that the chosen uncertainty model or risk policy is technically suitable.
Do not attach a universal confidence percentage to the band
A band equal to one expanded uncertainty does not automatically establish the same false-acceptance probability for every measurement and production population. Interpretation depends on the uncertainty model, coverage, distribution and the distinction between individual-result risk and average population risk. Keep those assumptions with any numerical risk statement.
A narrower acceptance interval generally changes the balance between accepting nonconforming items and rejecting conforming ones under the applicable model. The engineering choice must consider the actual consequence and measurement quality. Avoid describing guard-banding as a way to make uncertain measurements certain; it is a specified decision procedure using uncertainty information.
Report the value, rule and disposition together
Retain the measured resistance, units, uncertainty statement, tolerance limits, acceptance rule and resulting category. Where a nonbinary rule includes an unresolved region, preserve that category instead of forcing it into an undocumented pass or fail. A commercial concession should be recorded separately from the measurement’s conformity result.
ChipSimple can review resistance acceptance requirements against the proposed drawing and verification method. The guard band belongs to that specific measurement decision, not a blanket manufacturing capability claim. Keep method improvements, product tolerance changes and authorized exceptions visible so future lots can be judged by the same documented logic.
Define printed-resistor acceptance at the boundary
Provide the actual measurement and decision requirements before selecting a guard band.
- Nominal resistance and upper/lower limits
- Measurement terminals, temperature and conditioning state
- Available method uncertainty and coverage basis
- Required acceptance rule and unresolved-result categories
- Rounding, reporting and authorized exception policy
The drawing-upload form loads as you reach this section.