Cell: High-mix assemble / dispense (anonymized)
Claimed cycle: 3.6 s
Measured median: 4.2 s
Format: Waterfall of time. Not a speed-override pep talk.
Cycle 4.20 s
├─ Wait for part present .......... 0.85 s
├─ Approach / travel .............. 0.70 s
├─ Process (dispense) ............. 1.10 s
├─ Retreat / clear ................ 0.55 s
├─ Handshake / MES bit ............ 0.45 s
└─ Micro-recoveries (avg) ......... 0.55 s ← the lie in “3.6”
Bar 1 — Wait for part present
Sensor chatter + operator pre-stage late.
Robot is fast. The line is not.
Fix: presentation fixture, not jerk limits.
Bar 2 — Travel
Path looks short in teach. In production it doglegs around a temporary cable tray “for a week” (third month).
Fix: restore clearance; retime.
Bar 3 — Process
Dispense volume increased last month without cycle budget change.
Physics taxed time. Marketing kept 3.6 s on the slide.
If process volume grew, cycle time grew—unless you lie on the slide.
Bar 4 — Handshake
MES complete bit waits on a slow SQL tag.
Robot idle in “almost done.”
Fix: local handshake then async MES — with genealogy honesty elsewhere.
Bar 5 — Micro-recoveries
Following-error recoveries and gripper retries average 0.55 s into the median.
Not rare. Not “outliers.”
Hidden when engineers time one clean cycle after lunch.
Median includes the retries you pretend are exceptions.
Actions ranked by seconds returned
| Action | Est. s back | Owner | | --- | --- | --- | | Kill temporary tray dogleg | 0.25 | Facilities + robotics | | Fixture presentation | 0.40 | Manufacturing eng | | Async MES handshake | 0.30 | Controls | | Gripper retry root cause | 0.35 | EOAT + process | | Accept new dispense budget | truth | Leadership |
Adjacent fences
Payload/inertia logs own mass lies. Punch lists own commissioning RED items. Muting protocols own scanner proofs. Dress-pack guides own cable life. This waterfall owns where the seconds went. Do not raise speed overrides to hide MES latency.
Close
3.6 s was a clean-cycle fantasy. 4.2 s is the plant. Publish the waterfall—or keep arguing about servo brands.