What Is Gauge Stability (Drift) and How Do You Evaluate It in MSA?

In any measurement system, we often check for accuracy and repeatability. But there is another critical property that can quietly ruin your process data over weeks or months: stability (also called drift). If your gauge readings slowly change even though the part being measured does not, your control charts and capability indices will eventually lie to you.

This article explains gauge stability as defined in the AIAG Measurement Systems Analysis (MSA) Manual, 4th Edition, and shows you how to run a simple stability study using a control chart.

---

What It Is

Gauge stability (or drift) is the total variation in the bias of a measurement system over time when measuring the same master or reference part under the same conditions (same operator, same method, same environment).

In plain terms: you measure a known reference part today, next week, and next month. If the average of your readings moves away from the reference value as time passes, your gauge is not stable.

AIAG MSA 4th Ed. treats stability as a long-term property. It is different from:

• Bias – the difference between the observed average and the reference value at one point in time.
• Linearity – how bias changes over the range of measurement.
• Repeatability & Reproducibility (GRR) – short-term variation among repeated measurements.

Stability is about time.

---

How It Works: The AIAG Stability Study Method

According to the AIAG MSA Manual (4th Ed.), the recommended procedure is:

1. Obtain a reference standard – a master part or value whose true measurement is known (e.g., a certified gauge block or a designated production part measured on a high-level instrument).
2. Select one operator and one gauge (the one you want to evaluate).
3. Measure the reference part repeatedly at periodic intervals (e.g., daily, weekly) over a long period (weeks or months). At each time point, take 3 to 5 readings.
4. Compute the average of the readings at each time point.
5. Plot the averages on an I-MR chart (Individuals and Moving Range) or an Xbar-R chart over time.
6. Analyze the chart for:

- Points outside the control limits
- Runs, trends, or cycles
- A shift in the process average

Key formula – Control limits for the average chart (Xbar chart):

\[
UCL = \bar{X} + A_2 \cdot \bar{R}
\]
\[
LCL = \bar{X} - A_2 \cdot \bar{R}
\]

Where:

• \(\bar{X}\) = grand average of all subgroup averages
• \(\bar{R}\) = average of the subgroup ranges
• \(A_2\) = constant from the control chart table (e.g., for subgroup size \(n=5\), \(A_2 = 0.577\))

If the plotted averages stay within the control limits with no obvious pattern, the gauge is considered stable. If they drift outside or show a trend, the gauge has a stability problem (drift) that must be corrected (calibration, maintenance, replacement).

Note: AIAG MSA 4th Ed. emphasizes that stability should be evaluated before conducting a GRR study, because unstable measurements make any short-term study meaningless.

---

A Worked Illustrative Example

Example data (illustrative only) – not from any real factory.

Suppose you have a reference block with a known value of 10.000 mm. You measure it once per week for 10 weeks, taking 5 readings each time. The subgroup averages (\(\bar{X}\)) and ranges (\(R\)) are:

Week | Avg (mm) | Range (mm)

• 1 | 10.002 | 0.004
• 2 | 10.001 | 0.005
• 3 | 10.003 | 0.003
• 4 | 10.002 | 0.004
• 5 | 10.004 | 0.006
• 6 | 10.003 | 0.004
• 7 | 10.005 | 0.005
• 8 | 10.004 | 0.004
• 9 | 10.006 | 0.005
• 10 | 10.005 | 0.004

Calculations:

• Grand average: \(\bar{X} = 10.0035\) mm
• Average range: \(\bar{R} = 0.0044\) mm
• For subgroup size \(n=5\), \(A_2 = 0.577\)

Control limits:

\[
UCL = 10.0035 + 0.577 \times 0.0044 = 10.0060
\]
\[
LCL = 10.0035 - 0.577 \times 0.0044 = 10.0010
\]

All weekly averages fall between 10.0010 and 10.0060, with no trend upward or downward beyond random variation. Conclusion: The gauge is stable over this 10-week period (illustrative result only).

Notice that the averages hover slightly above the reference value of 10.000 – that is a bias issue, not a stability issue. Stability only asks: does the bias stay constant over time?

---

Common Pitfalls

1. Using a production part instead of a reference standard – if the part itself changes (wear, temperature), you will mistake part drift for gauge drift.
2. Too few time points – a stability study needs enough time span (AIAG suggests weeks or months) to catch slow drift; a single day of data is not a stability study.
3. Ignoring the range chart – even if the averages look stable, an unstable range chart indicates the gauge's precision is changing over time.
4. Confusing bias with stability – a gauge can be biased but perfectly stable (constant offset). Stability is about the consistency of that offset over time.
5. Skipping stability before GRR – per AIAG MSA 4th Ed., an unstable gauge invalidates any GRR or capability conclusion.

---

Closing

Gauge stability is the "long-term truthfulness" of your measurement system. A quick weekly check against a reference part, plotted on a control chart, will alert you to drift before it contaminates your process decisions. To run this analysis quickly and correctly, use the free Gauge Stability tool at 6sq.com – it follows the AIAG MSA 4th Ed. method and gives you the control chart and drift assessment in seconds.