8.4 Containment playbook: suspect lots, line stop triggers & escape handling

When a critical defect is first detected on the production floor, the primary engineering priority is Containment. Before initiating a Root Cause Analysis, it is essential to isolate the risk, define the suspect population, and establish a verified Clean Point so that production can safely resume.

Defining the suspect lot (the bracketing logic)

A “Suspect Lot” represents the volume of production that might be affected by the failure. This volume is bounded by the last known good state and the current detection point.

The “Bookend” Principle:

To execute an effective quarantine, you must first identify the temporal and physical boundaries of the risk window.

The Start Point (Bookend A): This is the last verified “Good” event. Examples include a passing hourly Quality audit, a documented machine setup verification, or a verified Golden Sample run.
The End Point (Bookend B): This is the moment the defect was detected, plus all current Work In Progress (WIP) that is still on the line.

The Execution Logic:

If the defect is continuous—such as a broken SMT nozzle—then all production units must be quarantined backwards to Bookend A. If the defect is random or intermittent and its boundaries cannot be accurately determined, the entire shift’s production volume must be quarantined for 100% visual or electrical sorting. In environments with limited traceability, such as batch-level tracking instead of individual serial numbers, it is necessary to quarantine the entire underlying Work Order.

Line stop triggers (the andon protocol)

Operators must have the authority, backed by explicit management support, to stop production when issues arise. Ambiguity in this standard can lead to increased scrap.

Triggers for an Immediate Auto-Stop:

Severity 1 Defect: Any potential safety or regulatory violation, such as exposed high-voltage traces or missing dielectric isolation.
Consecutive Failures: Three consecutive units failing the exact same test for the same defect code.
Yield Drop: Station First Pass Yield (FPY) dropping below 90% in any rolling hour.
Process Drift: A Critical to Quality (CTQ) parameter—like a reflow soldering temperature profile—drifting outside its engineered control limits (Cₚₖ < 1.0).

The Action Command:

When any trigger condition is met, the conveyor or machine must be stopped. Once the line is stopped, the Red Andon Light must be illuminated and the Lead Quality Engineer notified. Production should not resume until the Quality Engineer reviews the process and approves the “First Good Piece” check following any adjustment.

Escape handling (customer protection)

An “Escape” occurs when a defect bypasses all factory controls and is shipped to a customer or distribution hub. Taking prompt action is the most important factor in mitigating commercial impact.

Step 1: The Alert (Notification)

For any Safety or Regulatory Risk, the customer must be notified within 24 hours. For Functional or Cosmetic Risks, the customer must be notified promptly with a drafted containment plan; you should not wait until the root cause is discovered.

Step 2: The Clean Point

“Safe” audited product must be visually distinguished from “Suspect” product to restore customer confidence.

The Action: All certified safe units—those produced after an engineered fix or manually 100% screened—must be marked with a specific visual identifier. This could be a Neon Green Dot or a date-coded stamp, applied directly to both the outer packaging and the unit label.
The Communication: A formal notice detailing this specific marking convention must be sent to the customer.

Step 3: The Inventory Sweep

When an Escape is formally confirmed, these three locations must be swept immediately:

Finished Goods Warehouse: All outgoing shipments matching the part number must be halted.
In-Transit: Shipments must be recalled or diverted if possible.
Distributor Hubs: An immediate stock hold must be requested.

100% Sorting (the human firewall)

If a process is proven unstable, a temporary physical Firewall—a 100% Human or automated Machine sort—must be installed to protect the customer while engineering investigates the root cause.

Rules of Engagement:

The Instruction: Create a rapid “One Point Lesson” (OPL / visual aid) that zooms in specifically on the defect being hunted. This is more effective than relying solely on general IPC inspection criteria.
The Efficiency: 100% manual visual sorting is visually demanding. Rotate human inspectors periodically, for example every 2 hours, to maintain high focus.
The Exit Criteria: The Firewall must remain in place until the true Root Cause is demonstrably eliminated and at least 3 consecutive production lots pass the firewall with zero defects.

Recap: Line Stop Triggers, Quarantine Protocols, and Escape Containment Actions

Parameter	Requirement / Trigger	Action	Exit / Verification
Line Stop	3 consecutive failures (same defect code)	Stop line, activate Red Andon, notify Lead Quality Engineer	Quality Engineer approves First Good Piece check
Line Stop	Station First Pass Yield (FPY) < 90% (rolling hour)	Stop line, activate Red Andon, notify Lead Quality Engineer	Quality Engineer approves First Good Piece check
Line Stop	CTQ parameter out of control (Cₚₖ < 1.0)	Stop line, activate Red Andon, notify Lead Quality Engineer	Quality Engineer approves First Good Piece check
Suspect Lot Quarantine	Bounded by last verified Good event (Bookend A) to detection + all WIP (Bookend B)	Quarantine all units between Bookends A & B	N/A
100% Sorting (Firewall)	Process proven unstable	Implement 100% sort with visual aid (OPL), rotate inspectors every 2 hours	Root Cause eliminated AND 3 consecutive lots pass firewall with zero defects
Escape Containment	Defect shipped to customer/distributor	1. Notify customer per risk (24h for Safety/Regulatory). 2. Mark all safe units post-fix (e.g., Green Dot). 3. Sweep Finished Goods, In-Transit, Distributor stock.	Formal notice of marking convention sent to customer