Updated July 2026 · 8 min read · Adente Vision Engineering Team
What is inspection uptime, and how is it different from line uptime and OEE?
Inspection uptime is the fraction of scheduled production time during which the inspection station is available and inspecting. It is not the same as line uptime, which measures whether the production line is running, and it is not OEE, which is a composite of three separate factors. A line can be running while its inspector is down, and that gap is exactly what this metric exposes.
OEE, overall equipment effectiveness, is the product of three terms: Availability multiplied by Performance multiplied by Quality. Availability is the time term, how much of the scheduled time the equipment was actually running. Performance is the speed term. Quality is the good-parts term. Inspection uptime feeds the Availability factor of your process, and when the inspection station stops, it can drag the whole OEE number down even though the line itself never faulted.
The reason to separate inspection uptime out is accountability. If you only track line uptime, an inspector that drops out for ten minutes an hour hides inside a healthy-looking line number, and the escaped parts that slip through during those minutes are invoiced later as a quality problem. Naming inspection availability as a KPI puts the stop where it belongs.
Why should inspection availability be its own KPI?
Inspection availability deserves its own line on the report because the cost of an inspector being down is not lost production, it is uninspected parts shipped as if they were checked. When the station is offline, the line often keeps moving, so the failure is silent until a customer finds a defect that should have been caught.
That silent-failure property is what makes a blended line-uptime figure misleading. A station that is up 99% of scheduled time and one that is up 90% look almost identical on a line that never stops, yet the second one ships ten times as many unchecked parts. Commonly cited benchmarks put top-performing OEE near 85% and its availability term near 90%, but treat those as industry reference points, not a promise for any specific line, because the honest number depends on your equipment and your maintenance regime. What you can control is measuring your own inspection availability and driving the stop causes out of it.
What actually stops an inspection station?
Three failure classes account for most inspection downtime: network, thermal and calibration. A station that depends on a server, a switch or a WAN link inherits every hop as a single point of failure, so an IT outage that never touches the machine can still halt inspection. Thermal stops happen when a fan clogs or an ambient spike pushes compute past its limit. Calibration stops happen when lighting drifts or a fixture moves and the station has to be re-taught.
Each class has a design answer. Network dependence is removed by deciding on the device rather than in the cloud. Thermal risk is reduced by a fanless enclosure rated for the ambient range on your floor. Calibration drift is contained by controlled lighting and on-device preview so a re-aim is a short operator task, not a service call. The point of listing them is that inspection availability is a design outcome, not luck.
How do MTBF and MTTR describe a vision cell's reliability?
MTBF and MTTR are the two numbers that decompose availability for any vision cell. MTBF, mean time between failures, is how often the station stops: higher is better. MTTR, mean time to restore, is how quickly it comes back: lower is better. Availability rises when you push MTBF up or pull MTTR down, and the dependability vocabulary that defines these terms is standardised in IEC 60050-192.
For a vision station, MTBF is driven by how many moving or failure-prone parts it carries, a fan, a separate PC, a network path, and MTTR is driven by how fast a technician can restore or swap the unit. A station light enough for one person to remount in minutes has a structurally lower MTTR than one that needs two people and a service visit. These are the mechanics behind the availability percentage.
How do the inspection uptime metrics compare?
Each metric answers a different question, and confusing them is how downtime hides. The table below separates them so a report reader knows exactly which number is being quoted.
| Metric | What it measures | How to read it |
|---|---|---|
| Line uptime | Share of scheduled time the production line runs | Can look healthy while the inspector is down |
| Inspection availability | Share of scheduled time the station is inspecting | The KPI this post argues for; inspecting time / scheduled time |
| OEE | Availability x Performance x Quality for the process | A composite; inspection availability feeds its Availability term |
| MTBF | Mean operating time between failures | Higher is better; how often the station stops |
| MTTR | Mean time to restore after a stop | Lower is better; how fast you recover |
How does a fanless, self-contained edge design protect uptime?
A self-contained edge unit protects inspection availability by removing the failure classes above at the source: it decides on the device, so no network hop can stop it, and it has no fan to clog. Adente Vision is an edge-AI visual inspection unit built by ADENTE Advanced Engineering Technologies, part of the Aden Group, sold through automation system integrators, and its standard AV-S100 model is a fanless, IP54 enclosure rated for a 0-45 C ambient with 10-90% humidity.
Deciding on-device means an IT outage, a switch reboot or a WAN failure does not touch inspection, because capture, decision and logging all happen inside the enclosure. Model updates arrive by USB stick, so the station can sit on an air-gapped line and still be maintained, which removes a whole category of network dependency from the availability equation. The unit weighs under 9 kg, so one person can remount or swap it in minutes, which is a direct reduction in MTTR. None of this raises inspection accuracy on its own; what it protects is the number of hours per year the accuracy is actually available on the line.
How do you make the inspection uptime number auditable?
Inspection uptime is only useful if every stop is logged with a timestamp and a cause, so the KPI can be checked rather than asserted. An auditable record turns "the inspector felt flaky this week" into a Pareto of stop causes you can act on, and it lets you attribute an escaped defect to a specific window when the station was down.
Log each stop with its start time, duration and category, network, thermal, calibration or planned, and reconcile that log against the line schedule. Over a month, the split tells you whether to attack MTBF, harden a network path or fix a fixture. For the network side of this argument in more depth, see the sibling post on why edge keeps inspecting when the network goes down, and for the pillar view of AI visual inspection, see the guide.