For an automated pass/fail fixture accepting parts from 0.0 to 0.8 mm and rejecting gaps of 1.3 mm, the correct MSA entry for an ANOVA crossed Gauge R&R is LSL = 0 and USL = 0.8, with the 1.3 mm parts included as out-of-spec samples to challenge the system's discrimination [S1]. The part spread must deliberately bracket the spec limit so the study captures behaviour on both sides of the cutoff, not just inside the acceptable window.
Choosing the right study type matters as much as the limits. For a non-destructive automated gauge that every part can be measured on multiple times, the crossed design is correct; the nested design only applies when one operator destroys the part per trial [S1]. For deeper coverage of inspection hardware, see the force gauge reference page, and for the broader dimensional context, the gauge block entry.
Anova Crossed Study: Operator, Part, Trial Counts
AIAG MSA prescribes a minimum of three operators, ten parts, and two trials per operator for a Type 1 crossed study, with measurement order randomised to suppress drift bias [S1]. The resulting variance components are equipment variation (EV), appraiser variation (AV), and total GRR, expressed as a percentage of process tolerance or process variation.
Variation budget acceptance is commonly interpreted as: under 10% GRR is acceptable, 10-30% is marginal and may be acceptable depending on cost-of-use, and over 30% is rejected [S1]. In the gap-fixture case, EV typically dominates AV because the automated cell removes the operator from the measurement loop, so the study essentially validates fixture repeatability against the 0.8 mm cutoff.
Spec-Limit Entry for an Automated Pass/Fail Fixture
The Minitab dialog asks for LSL and USL based on the engineering specification, not the fixture's hardware travel; the 0.0-0.8 mm window is the spec, and the 1.3 mm samples are deliberately out-of-spec stimuli [S1]. Running the study with a wider 0.0-1.3 mm entry would mask the discrimination behaviour near the cutoff, where the fixture must decide accept/reject.
Including two to three out-of-spec parts (for example 1.0, 1.1, and 1.3 mm) in the ten-part panel is standard practice for a Type 1 study; it is also a hidden way to verify that the fixture trips its reject signal on every over-limit sample, which is a capability test as much as a repeatability test [S1]. For related tolerances on linear position, see the draw-wire sensor spring tension and fatigue analysis.
Connected Gauges: Digital Outputs, Statistics, and Protocols
Modern digital force gauges such as the AMETEK DFS II expose RS-232 at 9600 to 115,200 baud, USB, a Mitutoyo-compatible output, and a ±2 V analog channel, and they stream results directly to SPC packages with the operator blinded to the live value [S3]. On-gauge statistics include average, Cv, standard deviation, and pass/fail against high/low limits, which lets the gauge itself enforce the 0.8 mm threshold without a host PC.
AMETEK specifies accuracy better than 0.1% full scale on integral load-cell DFS II models, with mechanical overload protection at 150% of full scale, and a resolution and repeatability envelope that fits comfortably inside a sub-millimetre force-displacement loop [S3]. When the gauge is networked to multiple test stands, the same RS-232 stream feeds the Gauge R&R spreadsheet, which is why repeatability variance between identical units is the real reproducibility question, not operator variance [S3]. The force gauge page covers the broader measurement-system categories.
Fixture Certification and MSA Documentation Discipline
Supplier gauge standards such as ABC Technologies' 80-ENG-D-413 define a 60-section deliverable that ends with Measurement Systems Analysis, Gauge Instructions, and Preventive Maintenance, in that order [S2]. That ordering is not cosmetic: it forces the build, the GRR study, the operator card, and the PM schedule to land in one controlled document, so an OEM cannot ship a fixture without a defensible MSA package.
The same standard requires that every checking gauge carry its instructions physically attached and that an electronic copy travel with the gauge, which closes the loop when the gauge is later used on a connected line elsewhere in the plant [S2]. GRR acceptance criteria, build tolerances on tooling balls and locating pins, and the SPC indicator block all live in the same controlled file, so a Gauge R&R study is auditable against the gauge's own design intent rather than a generic MSA template [S2].
Comparison of Gauge R&R Study Methods
Three methods compete on a connected line: the Range method (fast, no ANOVA), the Average and Range method (Xbar/R, intermediate), and the ANOVA method (full variance components, supports interaction terms) [S1]. For a connected, automated fixture with no operator, the ANOVA method is preferred because it isolates equipment variation cleanly and supports the Type 1 crossed design that AIAG MSA endorses [S1].
For destructive tests (such as tensile break or weld pull), the nested design replaces the crossed design, since each part is measured only once and the operator factor collapses into the part factor [S1]. The Expanded GRR (up to eight factors) is reserved for cells where fixture, operator, lot, and machine all interact, which is rare on a single automated gap station but common on a multi-fixture robotic cell.
Limitations and Failure Modes of Connected GRR
Three failure modes dominate connected-gauge studies. First, mechanical drift between calibration cycles inflates EV, which a single calibration-day study will miss; second, USB-to-SPC middleware can quantise the last digit and bias the standard deviation; third, blinding the operator changes reproducibility variance, so comparing blinded to non-blinded studies is not valid [S3]. The DFS II hides results behind a password-protected setup menu specifically to mimic a blinded production environment [S3].
Overload history is a hidden GRR threat: load cells taken past 150% of full scale often return to spec but with elevated hysteresis, so any study that ignores the gauge's own overload log understates EV [S3]. AMETEK embeds overload-event tracking and a last-calibration timestamp inside the gauge so a quality engineer can disqualify a study that ran on a damaged load cell [S3].
Standard References and Sourcing
The two anchor references for this workflow are the AIAG Measurement Systems Analysis reference book (which supplies the 10/30% rule of thumb and the 3-operator / 10-part / 2-trial design) and ISO 22514-7 for measurement-system capability [S1]. ABC Technologies' supplier standard 80-ENG-D-413 supplies the documentation discipline that any Tier-1 automotive gauge must satisfy to be accepted at a customer's plant [S2].
OEM instrumentation references, such as the AMETEK DFS II data sheet, supply the accuracy, output, and overload numbers that are quoted in the GRR study's equipment file [S3]. For sourcing lamps and light fittings used in gauge vision systems, and for the wider lighting equipment and electric lamps category, the reference entries track the photometric standards that apply. Two trackable signals for the next planning cycle: revision updates to AIAG MSA 4th-edition errata, and any tightening of the 10% GRR threshold in IATF 16949 audits.