Quick answer
Formula: cycle time ≈ Σ (pick + move + place + EOAT open / close or vacuum build / vent + waits / settle)
Gate: stable cycle (with safety factor) must be ≤ line takt (seconds per piece)
Safety factor: about 1.1–1.2 for simple cells; often up to 1.3 with vision / multi-sensor coordination
Definition: catalog max TCP speed ≠ your stable cobot cycle; lock on the worst path and slowest loop
Pair with sizing: do not chase takt before payload and reach clear → Payload guide · Reach guide
The pendant stopwatch shows a “best loop” of 5.1 s against a 6 s takt—looks fine. By week two of production, vacuum retries once, vision takes an extra shot, and place waits for the tote. The slowest loop hits 6.8 s and the downstream queue grows. The fight is not “the arm is too slow.” It is treating the demo’s fastest loop as everyday cycle time. Below: cycle vs takt, a full vacuum segment table, a CNC door / chuck overhead example, safety factor, PPH, dual grip, and the cycle–EOAT–payload cross-check.
What cobot cycle time and line takt mean
Cycle time is the real seconds for a collaborative robotic arm to finish pick → place → return to a repeatable start, including arm motion and EOAT or equipment waits. MFG Calcs frames the sum as motion segments plus dwell / wait segments; the FIRGELLI cycle-time estimator multiplies the raw sum by a safety factor before talking throughput. Line takt is the average outbound interval the order or process requires—your target, not a number the arm invents.
| Term | Meaning | How to use it |
|---|---|---|
| Cycle time | Full-loop seconds | Compare to takt |
| Line takt | Target s / piece | Acceptance floor |
| Motion segments | Approach, transfer, retract | Budget worst path |
| Process / dwell | Grip, vacuum, inspect, wait equipment | Often skipped |
| Safety factor SF | ~1.1–1.3 margin | Absorbs variation and handshakes |
| PPH | 3600 / stable cycle seconds | Throughput only after SF |
Why the “fastest loop” lies
Demo paths are usually shortest, lightest, one-shot vacuum, and free of line handshakes. Production paths include farthest points, heaviest SKUs, leak retries, vision NG reshoots, and conveyor interlocks. The failure mode is not a lying controller—it is a commercial promise that used the demo denominator. Acceptance language should be: stable cycle (with SF) ≤ takt, locked on both average and slowest loop.
Step-by-step calculation
List every beat: pick, grip / seal, transfer, place, release / vent, retract, plus inspection / cure / equipment waits
Split arm-moving vs waiting—gripper, vacuum, vision, conveyor, and machine doors all count
Budget farthest point, heaviest part, awkward pose—not the pretty demo path
Add settling (often 0.2–0.5 s) and handshake delays
Raw sum × safety factor (1.1–1.3), then compare to takt; quote PPH = 3600 / stable cycle seconds
Run 20–30 real-part loops; lock average and slowest loop. Dual-grip math: Robotiq dual gripper
Worked example 1: vacuum pick-and-place (full segment table)
Target takt 6.0 s/piece. Single-cup vacuum rough sum:
| Segment | Content | Seconds |
|---|---|---|
| Approach | Above pick | 0.5 |
| Vacuum build | Seal + pressure OK | 0.8 |
| Transfer | To place zone | 1.2 |
| Place / locate | Descend + locate | 0.6 |
| Vent | Release | 0.4 |
| Retract | Return to start | 0.7 |
| Settle | Anti-bounce | 0.3 |
| Arm + EOAT subtotal | — | 4.5 |
| Vision OK wait | Shot / judge | 1.0 |
| Raw total | — | 5.5 |
| × SF 1.15 | — | ≈ 6.3 |
6.3 s > 6.0 s takt—already over. Cut waits or path, evaluate dual grip, or change EOAT path—do not pretend the 5.1 s demo ships. EOAT open-close still open → End-effector guide; path length and base → Workcell layout guide. Soft shortlist while you edit the segment sheet: Side-by-Side Comparison.
Worked example 2: CNC tending (door / chuck overhead)
Target takt 12 s/piece (in-cut time not on the arm’s critical path). Arm-side rough sum: pick blank 1.0 + approach door 0.8 + wait door open 1.5 + chuck place / clamp handshake 2.0 + spindle permit 1.0 + pick finished 1.2 + exit door 0.8 + place tote 1.0 + EOAT open-close 0.8 ≈ raw ~10.1 s. With SF=1.2 → ~12.1 s—on or over the line. EVS’s machine-tending guide puts door, chuck, and spindle-ramp overhead on the same ledger—skip “wait door” for two seconds and the paper cycle always looks pretty.
| Segment | Seconds | Note |
|---|---|---|
| Arm motion total | 4.8 | Farthest path |
| Door open / close wait | 1.5 | Often skipped as “the machine’s job” |
| Chuck / clamp handshake | 2.0 | Includes permit |
| Spindle / enter permit | 1.0 | Interlock |
| EOAT open-close | 0.8 | On the checklist |
| Other settle | 0.0–0.5 | Vibration-dependent |
| Raw | ≈ 10.1 | — |
| ×1.2 | ≈ 12.1 | Compare to 12 s takt |
Cycle bands and PPH (rough)
| Stable cycle (with SF) | Approx PPH | Typical read |
|---|---|---|
| ≤ 4 s | ≥ 900 | Very tight; path and EOAT must be clean |
| 4–8 s | 450–900 | Common packaging / light handling band |
| 8–15 s | 240–450 | Common machine-tending band |
| > 15 s | < 240 | Heavy tooling / many handshakes—ask dual grip or split task |
PPH = 3600 / cycle seconds. Before quoting shift output, confirm you used stable cycle (not demo best) and already applied SF.
Dual grippers (quantify, do not guess)
Dual grip often removes empty return trips, but tool-switch time, pose, and mass all change. Robotiq publishes formulas for the uplift—use math over “it should be twice as fast.” Dual tools also raise TCP mass and CoG; rewrite the Payload guide ledger.
Cross-constraints: cycle × EOAT × payload × reach
| Cross | Question that must share one page | Revisit |
|---|---|---|
| Cycle × EOAT | Are open-close / build-vent in the segment table? How are leak retries counted? | End-effector guide |
| Cycle × payload | After full-load speed derate, is cycle still ≤ takt? | Payload guide |
| Cycle × reach / layout | Is farthest path length based on the real base? | Reach guide · Workcell layout guide |
Near full load, controllers often throttle TCP speed—published notes cite on the order of 30–50% below top rated speed—so “max speed × path length” takt estimates are optimistic.
Common mistakes
| Mistake | Floor symptom | Fix |
|---|---|---|
| Catalog max TCP speed only | 4 s on paper, 7 s on floor | Estimate with your path and accel |
| EOAT open-close skipped | Demo saves 1 s+, production slips | Put beats in the list → End-effector guide |
| Arm motion only—waits ignored | Never matches takt | Add inspect, vent, machine overhead |
| Fastest loop as daily rate | Misses when quality dips | Lock average + slowest loop |
| No safety factor in the promise | Misses after handshakes | Apply SF 1.1–1.3 before PPH |
| Takt before payload / reach | Overload to “make time” | Validate first → Payload · Reach |
| Door / chuck handshakes ignored | CNC cells miss across the board | Put EVS-style overhead on one ledger |
Published evidence bands
| Band | Claim | Source |
|---|---|---|
| Segment sum | Motion + dwell / process | MFG Calcs robotics |
| Safety factor | ~1.1–1.2; complex up to 1.3 | FIRGELLI cycle estimator |
| Machine overhead | Door / chuck / ramp must enter the sum | EVS machine tending |
| Dual-grip uplift | Quantify with formulas, not gut feel | Robotiq dual gripper |
