Who Has the Final Say: Go/No-Go Gauges or CMMs? — The Full Process for Acceptance, Wear Limits, and Attribute MSA of Go/No-Go Gauges and Special Inspection Tools

Last week, a quality manager told me about an incident over the phone: a batch of bearing holes was fully inspected by IQC using go/no-go gauges, and all 200 pieces were deemed合格. However, when the goods arrived at the customer's side, the customer's sampling inspection using a CMM found that 3 out of 10 pieces exceeded the upper limit, leading to the entire batch being returned. The customer's quality engineer wrote a harsh comment in the email: "Do you never verify your inspection tools?"

What made it even more difficult for him was that he didn't know how to respond. The gauges were purchased from a reputable manufacturer and were stored in the warehouse upon receiving the certificate of conformity; they were used daily in the workshop without anyone ever saying "it doesn't pass." But when asked about the current diameter of the go gauge and the wear limit for replacement, no one in the factory could answer.

Such disputes occur almost every week. On the surface, they argue over "which is more accurate, go/no-go gauges or CMMs," but in reality, they expose three things that haven't been done: the inspection tools have not been accepted, there are no wear limits, and no attribute MSA has been performed. This article will break down these three issues and provide actionable checklists, formulas, and criteria for each.

1. Go/No-Go Gauges Measure Functional Boundaries, Not True Dimensions

First, let's clarify a premise that is often overlooked: go/no-go gauges and CMMs answer different questions.

• A CMM answers "what is the diameter of this hole?" and outputs a numerical value.
• A go/no-go gauge answers "can this hole fit and will it wobble?" and outputs a binary result:合格 or 不合格. The former is a measurement system (计量型测量), while the latter is a counting system (attribute inspection) (计数型检验).

The design of go/no-go gauges follows two principles:

1. The go gauge is designed to the maximum material condition (MMC) and must be full form. When inspecting a hole, the basic size of the go gauge equals the minimum limit size of the hole (MMC), and it is a complete cylinder. This means it checks both size and shape—no matter how large the hole is, if the roundness or straightness is poor enough to prevent the gauge from passing, the go gauge will not pass.
2. The no-go gauge is designed to the least material condition (LMC) and only checks single points. The basic size of the no-go gauge equals the maximum limit size of the hole, and it is made with a ball end or short cylinder. It requires insertion at each point and in different orientations, and it must not pass through the entire hole in one go. It only judges whether any part exceeds the upper limit, not the shape.

This is the Taylor principle of gauge design: the go gauge controls the maximum material condition, the no-go gauge controls the least material condition, and together they ensure that the hole is within the tolerance band everywhere. Understanding this, many disputes can be avoided—CMM measures 10.017 (exceeding the upper limit by 0.002), while the go gauge passes and the no-go gauge fails in two orientations. Both conclusions can be correct: the hole is within the allowed orientation but exceeds the limit in others, or there is a shape issue. Conversely, if the CMM measures 10.005, which is completely合格, but the go gauge fails, it is likely that the hole's straightness or roundness is不合格, or there are burrs or chamfer interferences at the hole's entrance.

The most critical point, as seen in the fourth and fifth rows, is that the go gauge and the no-go gauge wear in the same direction but have opposite consequences. The go gauge wears down and may pass holes that should not pass, leading to defective products reaching the customer. The no-go gauge wears down and may block holes that should pass, leading to production line stoppages. Therefore, the management actions differ—the go gauge must have a wear limit and be regularly measured, while the no-go gauge should be replaced when it wears down to the manufacturing lower limit.

2. Acceptance: Six Steps for New Inspection Tools

Most factories start their inspection tool management from "buying them," not from "acceptance." A certificate of conformity only proves that the gauge manufacturer made the tool according to the drawing, not that the tool you received matches your drawing, has not been damaged during transportation, or has no peeling plating.

The acceptance process is recommended to follow the checklist below, and any item that fails should not be marked as "合格" and stored in the warehouse:

Two details are most often overlooked:

• Resolution rule. The measuring instrument for the gauge's working surface size should have a resolution no greater than 1/10 of the measured size's tolerance band. For a Φ10H7 hole with a tolerance of 0.015, a 0.001-grade micrometer must be used to measure the gauge, not a caliper (0.02), as the latter's measurements are meaningless.
• Calibration pieces are more useful than certificates. Inspection tools with calibration pieces can be quickly verified in 30 seconds: if the calibration piece passes, the tool is likely fine; without a calibration piece, each verification must be sent to the metrology lab, which often means "never verified."

The acceptance record should not just say "合格" but should include three numbers: the actual measured diameter of the go gauge, the actual measured diameter of the no-go gauge, and the acceptance date. These three numbers serve as the baseline for calculating wear and determining replacement cycles.

3. Wear Limits: One Formula and One Example

Go/no-go gauges are consumable items that wear out over time. The core of managing them is not "maintaining precision" but "replacing them before they lose precision." This involves three steps: setting the manufacturing tolerance, setting the wear limit, and setting the verification cycle.

Step 1: The gauge manufacturing tolerance is 10% of the workpiece tolerance. This is a standard gauge design approach (consistent with ISO 1938 series and domestic smooth gauge standards): for a workpiece tolerance of 0.015, the gauge manufacturing tolerance is approximately 0.0015, and in engineering practice, it is taken as 0.001. The ratio of the gauge tolerance to the workpiece tolerance essentially represents "how much of the tolerance budget is consumed by the inspection tool." A 10% ratio is a conventional balance between precision and cost; a ratio of 20% or more makes the inspection tool a significant source of process capability loss.

Step 2: The go gauge has a wear allowance, but the no-go gauge does not. The go gauge is designed to the maximum material condition (slightly larger than the hole's minimum limit) and gradually approaches the true lower limit as it wears. It should be scrapped when it reaches the lower limit, and the illusion of "using it a bit more for safety" must be avoided—wearing down the go gauge leads to misjudgment. The no-go gauge is designed to the least material condition, and wearing it down leads to misjudgment. It should be replaced when it wears down to the manufacturing lower limit.

For a hole Φ10H7 (+0.015/0):

In actual practice, two points need to be aligned with the gauge manufacturer:

1. Directly mark the wear limit size on the gauge drawing to avoid on-site calculations.
2. If the gauge is manufactured with a wear allowance (e.g., the go gauge is made to 10.0015 ～ 10.0025), the wear limit is 10.0015, as marked on the drawing. The most common dispute arises when the gauge drawing does not specify the wear limit, and the on-site personnel rely on their feel until the gauge "obviously loosens."

Step 3: The verification cycle is based on "number of uses or time, whichever comes first." For frequently used go/no-go gauges, it is recommended to verify every 5000 uses or 3 months, whichever comes first; for less frequently used gauges, every 6 months. The verification content includes only two items: passing the calibration piece and measuring the go gauge diameter with a micrometer. The ledger should at least answer five questions:

1. Number and specifications, corresponding process and characteristics.
2. Go/no-go gauge baseline size and wear limit.
3. Last verification date, actual measured diameter, and verifier.
4. Cumulative number of uses or production batches.
5. Scrap date and reason (exceeding wear limit, dents, peeling plating).

The ledger fields are not complex, but the challenge is ensuring that someone actually measures the gauges every three months. A practical approach is to include "inspection tool verification" in the changeover inspection form and tie it to the first article inspection—verifying the inspection tool during changeover is more effective than a separate "inspection tool verification plan."

4. Attribute MSA: Go/No-Go Gauges Need It, But the Method Is Different

When a customer auditor asks, "Have your inspection tools undergone MSA?" many people's first reaction is, "How do you do GR&R on go/no-go gauges, which only give合格 or 不合格?" The answer is: counting type measurement systems use attribute consistency analysis, not the GR&R for range and mean.

The process can follow these steps:

Sampling. 50 pieces, with the key being the distribution—30% to 50% must be boundary pieces (within ±10% of the tolerance band limits). A sample of entirely obvious合格 pieces will show 100% consistency for anyone measuring, making the study meaningless. Boundary pieces can be selected from the retained pieces during the worst process capability period or specifically manufactured.

Measurement. 3 inspectors × 3 times each × 50 pieces, with piece numbers and measurement order randomized, and no communication between inspectors. The condition of the inspection tool must be verified before the study (refer to Section 3).

Calculate four indicators.

Focus on the misjudgment rate. An overall consistency of 90% may seem good, but if all 10% inconsistencies are concentrated on "judging out-of-tolerance pieces as合格," the measurement system is unusable. Therefore, the statistics should be separated: the misjudgment rate (不合格 judged as合格) should approach 0, while the false rejection rate (合格 judged as不合格) can be tolerated but has a direct impact on cost. The judgment criteria should be clearly defined in the procedure document: misjudgment rate > 2% or Kappa < 0.75, the inspection tool must be taken out of service and rectified.

When only one inspection tool is involved and there is no personnel difference, the process can be simplified to "the same tool measuring 50 pieces 3 times," focusing on self-consistency without the need to involve 3 people.

5. Dispute Resolution: Follow a Fixed Sequence, Don't Argue in the Meeting Room

Returning to the opening scenario. When the go/no-go gauge judges合格 and the CMM judges out-of-tolerance, the correct sequence of actions is fixed. Completing these five steps will resolve 90% of disputes:

Step 4 is the easiest to skip and the most likely to lead to a disadvantage. If the CMM measures 10.0165 and the upper limit is 10.015, the difference is 0.0015, and the device's expanded uncertainty is 0.002—this "out-of-tolerance" is not valid in metrology. Conversely, if a piece judged合格 by you is only 0.001 away from the upper limit, it should not be used as evidence that "we have no problem."

More important than arbitration is turning disputes into prevention: retain three boundary pieces from each dispute, label them, and store them in the limit sample cabinet. The next time the same issue arises, there's no need to argue; just compare. This is the lowest-cost way to convert a dispute into a capability improvement.

6. A Case Study: 8.2% Rework Rate Traced to a Single Plug Gauge

A certain automotive parts company, with a steering knuckle bearing hole Φ40H7, experienced excessive axial clearance after bearing press-fitting. The assembly line's rework rate was consistently around 8%, and there were three customer complaints within three months, all suspecting "unstable pressure from the press-fitting equipment."

A complete traceability investigation later revealed the issue was with the inspection tool:

• The hole diameter was fully inspected using a go gauge, which did not exist in the ledger. It was purchased two years ago with the tooling and had never been verified.
• The metrology lab measured the go gauge diameter at 39.988, which was 0.012 smaller than the wear limit of 40.000, far exceeding the allowed wear.
• The result was that a batch of holes with diameters between 39.988 and 40.000 were judged合格, leading to insufficient interference fit and excessive axial clearance after press-fitting.
• An attribute consistency study (50 pieces, 18 of which were boundary pieces) showed that the inspection tool's consistency with the benchmark was only 78%, with a Kappa value of 0.62 and a misjudgment rate of 8%.

The corrective actions involved three simple steps: establish a gauge ledger and mark the wear limits; set the verification cycle to 5000 uses or 3 months, whichever comes first, and add a line to the changeover inspection form; re-conduct the attribute MSA and keep 20 boundary pieces on hand. Three months later, the assembly line's rework rate dropped to 1.1%, and similar customer complaints were eliminated.

In this case, there was no "high-tech" action. All losses stemmed from one simple issue: no one was responsible for replacing the gauge before it failed.

7. Five Common Pitfalls

1. New gauges are used without acceptance records. Whether the size is correct and whether there are any dents are entirely based on the supplier's certificate of conformity. The three numbers on the acceptance form (go, no-go, date) are the foundation for all subsequent wear management.
2. Using a caliper to measure go/no-go gauges. The resolution is insufficient, and the differences measured are entirely due to instrument noise, leading to the incorrect conclusion that "there is no significant wear."
3. Continuing to use a no-go gauge that has worn down. "It can still fit a little," but in reality, it is misjudging good pieces, causing daily arguments on the production line and eventually leading to the gauge being discarded, rendering inspection ineffective.
4. No markings, no ledger, no cycle for gauges. When an auditor asks, "When was this plug gauge last verified?" and there is no answer, this is almost always a nonconformity.
5. Treating the "合格" from go/no-go gauges as a dimensional report to the customer. Go/no-go gauges only prove that the hole is within the functional boundary and do not provide any dimensional values. When the customer requests Cpk or actual measurements, a measurement system must be used, and the two sets of data cannot be substituted for each other.

One Sentence Summary

Go/no-go gauges manage functional boundaries, not true dimensions—proper acceptance, wear limits, and attribute MSA will naturally resolve disputes.

Gauges manage functional boundaries, not true dimensions

Knowledge code: 6.2.1

Version: v20261019

Author: QTank QTank is dedicated to providing systematic professional knowledge, methodologies, and practical tools for quality management practitioners, helping companies continuously improve their quality capabilities.