Key Takeaways: Micro-stops, production interruptions lasting 5 to 30 seconds, represent 8-15% of available production time in most manufacturing operations. They're the largest single source of OEE performance rate losses, yet they're completely invisible to PLC-based monitoring because they're shorter than typical sensor sampling intervals. Fabrico's computer vision captures every micro-stop with video evidence, making the invisible visible and giving maintenance teams the root cause data to eliminate them permanently.
See our roundup of the analytics layer that helps surface these losses.
The OEE performance rate measures how efficiently a machine runs during the time it's actually available, the ratio of actual output to what it would produce if running at ideal speed without interruption. Most manufacturers focus OEE improvement efforts on availability losses (equipment failures) because they're the most visible. Breakdowns get reported, generate work orders, and receive maintenance attention.
Micro-stops receive none of this attention, because they're invisible to the monitoring systems most plants rely on.
A micro-stop is a production interruption lasting between 5 and 30 seconds. The machine stops, an operator clears a jam or repositions a part, and production resumes.
From a PLC signal perspective, this event may not register at all, many PLCs sample machine state every 1-5 seconds, and a 10-second stop may generate only one or two samples before the machine is running again. Most OEE systems aggregate these events into an "unknown" category or miss them entirely.
Yet these events are massive in aggregate. In a high-speed packaging operation running 1,200 bottles per minute, a 15-second micro-stop loses 300 bottles of production. If this happens 20 times per shift, not unusual in operations with material feed issues or worn components, the cumulative loss is 6,000 bottles per shift, or 8.3% of an 8-hour shift's theoretical output.
This 8.3% loss shows in the OEE performance rate calculation as a gap between actual and ideal output. But without knowing which of the 20 micro-stop events is causing it or why, the maintenance team has no actionable information. They see the loss in the OEE number; they can't see the cause.
The technical reason micro-stops are invisible to PLC-based OEE monitoring is sampling rate combined with signal interpretation:
Sampling intervals: Most industrial PLCs sample discrete inputs (machine run/stop signals, production counters) at intervals between 100ms and 5 seconds, depending on application.
A 10-second micro-stop generates 2-100 samples depending on the PLC scan rate, but many OEE systems filter out events shorter than a configurable threshold, typically 30-120 seconds, to avoid cluttering the downtime log with operator-cleared jams that "don't count" as downtime.
This filtering is the standard industry practice, and it systematically hides the most common source of performance rate losses.
Signal interpretation: Even when a 15-second stop is detected, the PLC signal tells the OEE system "machine was stopped for 15 seconds." It doesn't tell the OEE system why, whether it was a material jam, a misfeed, an operator adjustment, a sensor false trigger, or any of the dozens of other causes that create micro-stops.
Without root cause information, the event is logged as "unknown stop" or similar, data that quantifies the loss but provides no direction for eliminating it.
Operator-cleared jams: Many micro-stops are cleared by operators in the normal course of production, the jam is a normal occurrence that operators manage as part of their work, not a maintenance event. These events never generate an alarm, never trigger a work request, and never appear in the maintenance system. They're production overhead that nobody tracks, nobody analyzes, and nobody has a mandate to reduce.
The result: a systematic category of OEE losses representing 8-15% of production capacity that is invisible to every monitoring system except camera-based observation.
You can use production line cameras without turning them into surveillance.
Fabrico's Inefficiencies Zoom-In feature uses cameras positioned at production stations to continuously observe machine and operator activity. The system applies computer vision algorithms to detect anomalies, deviations from the normal production pattern, in real time.
When the computer vision system detects an anomaly, a micro-stop, an operator reaching into a machine to clear a jam, a part that stopped at an unexpected position, it captures a short video clip (typically 10-30 seconds) of the event. The clip includes a few seconds before the anomaly began and the complete recovery sequence.
These clips are automatically categorized by event type, tagged with the asset, time, and duration, and integrated into Fabrico's OEE calculation as performance rate losses.
The maintenance team can review video clips grouped by event type to identify recurring patterns, not just that 20 micro-stops occurred in a shift, but that 14 of them were material jams at the same feed point, and the video shows the jam occurring because a worn guide rail is allowing the material to tip 5mm past its proper alignment.
That specific root cause, the worn guide rail, generates a targeted CMMS work order in Fabrico. The maintenance action is precise: replace the feed guide rail at the specific point where the wear is occurring. The result: 14 of the 20 daily micro-stops are eliminated, and the OEE performance rate recovers the 8.3% of daily capacity that those stops were consuming.
The computer vision capability extends beyond micro-stop capture. Fabrico's Inefficiencies Zoom-In also detects:
Each of these loss categories represents production capacity that PLC-based OEE monitoring misses entirely. Fabrico's computer vision makes them visible, quantified, and actionable.
Capturing micro-stops is the first step. Eliminating them systematically is the goal. Fabrico provides the structured path from computer vision event capture to permanent loss elimination:
Step 1: Quantify and rank. Fabrico's AI Agent ranks micro-stop event types by total production impact, events per shift × duration per event × production value per hour. The ranking immediately shows which micro-stop category is the highest-value target. In most operations, the top 3 micro-stop types account for 60-75% of total micro-stop losses.
Step 2: Root cause from video evidence. For the highest-ranked micro-stop type, maintenance teams review the accumulated video clips in Fabrico. The computer vision has already categorized and tagged the clips, the team sees 14 clips of the "feed guide jam" type, can watch a representative sample, and identifies the consistent visual pattern: the material enters the feed mechanism at a consistent angle that causes the jam.
Step 3: Create targeted CMMS work order. The root cause identified from video evidence generates a CMMS corrective action work order in Fabrico. The work order includes the video evidence (attached to the work order), the frequency and timing data, and the estimated production impact of the failure. The maintenance team has complete context before arriving at the machine.
Step 4: Verify elimination through OEE monitoring. After the corrective action is completed and the work order closed, Fabrico monitors the micro-stop event frequency for 7 days. If the feed guide replacement eliminated the jams, the "feed guide jam" event count drops to zero.
If jams continue but at reduced frequency, the repair addressed only part of the root cause. The verification data closes the improvement loop, confirming success or triggering further investigation.
This systematic approach, identify from data, understand from video, act with work order, verify with OEE trend, is what distinguishes sustainable micro-stop elimination from one-time fixes that recur because the actual root cause was never understood.
Operations that deploy Fabrico's computer vision for micro-stop capture and apply the systematic elimination approach consistently report 6-12% improvement in OEE performance rate within 90 days. On a 10-line plant generating $4,000/hour, a 9% OEE performance rate improvement = $314,000/month in additional production capacity, from the same equipment, with the same maintenance team, using computer vision to see losses that previously had no visibility.
Micro-stop detection works best as one part of continuous AI production monitoring across the whole line, not as a standalone alarm.
Login
Blog
