![\"PCB](\"https:\/\/pickandplacemachine.com\/wp-content\/uploads\/2026\/01\/PCB-Handling-Machines3.jpg\")
<\/a><\/figure>\n\n\n\nComplete pick and place machine workflow: the operating cycle you can actually audit<\/h2>\n\n\n\n
I\u2019ll lay out the \u201ccomplete pick and place machine workflow\u201d the way a line engineer audits it\u2014step-by-step, with the ugly parts included.<\/p>\n\n\n\n
1) Program import and library mapping (where failures are born)<\/h3>\n\n\n\n
Centroid file in. Gerbers in. Package library applied.<\/p>\n\n\n\n
If your package definition is wrong (body size, lead span, pickup point, rotation), the machine will still place parts. It\u2019ll just place them wrong at scale. That\u2019s how you get \u201cmystery defects\u201d that show up as AOI noise or reflow defects later.<\/p>\n\n\n\n
And yes, this is still part of the operating cycle, because the machine will keep paying for this mistake every single cycle.<\/p>\n\n\n\n
2) Feeder indexing and component presentation<\/h3>\n\n\n\n
Feeders advance the tape. The pocket presents the component at a pickup location.<\/p>\n\n\n\n
This looks simple until you measure it. Tape drag, cover tape peel angle, pocket tolerances, and splice quality all change how consistently a component sits when the nozzle comes down. You want stable presentation. You rarely get perfect presentation.<\/p>\n\n\n\n
This is why feeder quality and maintenance aren\u2019t \u201cnice to have.\u201d They\u2019re cycle time and yield.<\/p>\n\n\n\n
3) Pick: nozzle down, vacuum on, pickup verification<\/h3>\n\n\n\n
The head moves to the pickup point. Z-axis down. Vacuum engages. Z up.<\/p>\n\n\n\n
Then comes the part most people forget: pick verification. The controller watches vacuum pressure (and sometimes flow) to confirm \u201cpart attached.\u201d A weak seal or cracked nozzle can pass sometimes and fail other times, which is the worst kind of failure because it destroys trust in your alarms.<\/p>\n\n\n\n
Short sentence: mispicks happen.<\/p>\n\n\n\n
4) Vision: pre-align, center-of-gravity correction, rotation<\/h3>\n\n\n\n
Now the machine checks what it actually picked.<\/p>\n\n\n\n
For many parts, the camera captures an image and computes offsets: X\/Y shift, theta rotation, sometimes even skew. If you run \u201con-the-fly\u201d vision, this happens while the head is moving, which saves time but increases sensitivity to lighting drift and camera calibration.<\/p>\n\n\n\n
Want an uncomfortable question? How often do you re-validate your vision thresholds after you change reels or suppliers?<\/p>\n\n\n\n
5) Board entry, clamping, and fiducial acquisition (the real coordinate reset)<\/h3>\n\n\n\n
The PCB enters on a conveyor. It clamps or supports. The machine finds fiducials.<\/p>\n\n\n\n
Fiducials are not \u201cnice.\u201d They are the machine\u2019s anchor points. The controller uses them to calculate board offset and rotation so the CAD placement coordinates match the physical PCB sitting in the machine right now.<\/p>\n\n\n\n
This is where your \u201chow does a pick and place machine work\u201d question becomes pure geometry. Without fiducials (or with bad fiducials), you\u2019re placing blind.<\/p>\n\n\n\n
6) Place: motion profile, Z control, placement force, and settle time<\/h3>\n\n\n\n
Head moves over the placement coordinate. Z down. Part placed. Z up.<\/p>\n\n\n\n
But there\u2019s a lot hiding inside \u201cplace.\u201d The machine must control placement height, contact, and release timing so the part stays on the paste and doesn\u2019t shift. If you push too hard, you smear paste or tilt parts. Too light, and parts can stick to the nozzle and \u201cdrop later,\u201d which is chaos.<\/p>\n\n\n\n
And yes\u2014settle time is real. The head needs a tiny pause to stop vibration from turning into placement error at high speed.<\/p>\n\n\n\n
7) Post-place checks and exception handling (where your throughput disappears)<\/h3>\n\n\n\n
If the machine thinks something went wrong, it doesn\u2019t just shrug. It does something:<\/p>\n\n\n\n
- \n
- Retry pick<\/li>\n\n\n\n
- Send part to reject bin<\/li>\n\n\n\n
- Pause for operator<\/li>\n\n\n\n
- Mark placement as \u201csuspect\u201d<\/li>\n\n\n\n
- Trigger a stop if error rate spikes<\/li>\n<\/ul>\n\n\n\n
Every exception adds time. And exceptions cluster. One bad reel can drag an entire shift.<\/p>\n\n\n\n
8) Repeat: optimize head path, minimize travel, balance feeders<\/h3>\n\n\n\n
This is the SMT pick and place process at scale: the controller optimizes travel paths and tries to keep the head doing useful work instead of commuting across the table.<\/p>\n\n\n\n
But feeder layout still dominates. Put high-usage parts far apart and you just forced longer travel forever. Put odd-form parts in the \u201cwrong\u201d zone and you\u2019ll pay in slower vision cycles.<\/p>\n\n\n\n
9) Board exit to downstream process (reflow, AOI, SPI)<\/h3>\n\n\n\n
The board leaves. The next board enters. The cycle restarts.<\/p>\n\n\n\n
Here\u2019s where I get opinionated: if you don\u2019t design the line<\/em> and only \u201cbuy a machine,\u201d you\u2019ll keep blaming the placer for problems caused by printer setup, stencil cleanliness, paste condition, and conveyor handoffs.<\/p>\n\n\n\n
If you\u2019re planning a full line, read how turnkey lines are usually scoped, not just sold. The \u201cturnkey SMT line solutions\u201d approach tends to expose bottlenecks earlier, especially around feeder strategy and inspection integration. (Pick and Place Machine<\/a>)<\/p>\n\n\n\n
![\"PCB](\"https:\/\/pickandplacemachine.com\/wp-content\/uploads\/2026\/01\/PCB-Handling-Machines2.jpg\")
<\/a><\/figure>\n\n\n\nPick and place machine cycle time: the math people avoid<\/h2>\n\n\n\n
Cycle time is not the same as \u201cmaximum CPH\u201d (components per hour). Specs are typically measured under ideal conditions: small components, optimized layout, minimal vision overhead, no feeder errors, no nozzle changes, no retries.<\/p>\n\n\n\n
Let\u2019s do real math.<\/p>\n\n\n\n
If a machine is rated at 50,000 CPH<\/strong>, the theoretical time per placement is:<\/p>\n\n\n\n
- \n
- 3,600 seconds\/hour \u00f7 50,000 placements\/hour = 0.072 seconds per placement<\/strong><\/li>\n<\/ul>\n\n\n\n
Now say your board has 350 placements<\/strong>. Pure theoretical placement time:<\/p>\n\n\n\n
- \n
- 350 \u00d7 0.072 = 25.2 seconds<\/strong><\/li>\n<\/ul>\n\n\n\n
But your real board time isn\u2019t 25.2 seconds, because you also pay for:<\/p>\n\n\n\n
- \n
- Fiducial find time<\/li>\n\n\n\n
- Conveyor load\/unload<\/li>\n\n\n\n
- Vision time for non-standard parts<\/li>\n\n\n\n
- Nozzle changes<\/li>\n\n\n\n
- Feeder indexing variability<\/li>\n\n\n\n
- Retries and rejects<\/li>\n<\/ul>\n\n\n\n
That\u2019s why \u201cpick and place machine cycle time\u201d needs to be measured on your<\/em> BOM, your<\/em> board, your<\/em> feeder plan.<\/p>\n\n\n\n
And if you\u2019re running mixed builds or prototypes, you\u2019ll feel it even more\u2014changeover, kitting, and setup discipline dominate. That\u2019s why prototype and small-batch lines are their own category, not a footnote. (Pick and Place Machine<\/a>)<\/p>\n\n\n\n
Components that matter: feeder, nozzle, vision (and the parts vendors oversimplify)<\/h2>\n\n\n\n
People love listing components like it\u2019s a catalog. The useful way is: what can fail, and how fast does it fail?<\/p>\n\n\n\n
- \n
- Feeder<\/strong>: Determines component presentation consistency. Bad feeders create mispicks, skew, and time-wasting retries.<\/li>\n\n\n\n
- Nozzle<\/strong>: Determines pickup reliability and placement stability. Wear, contamination, and wrong nozzle choice quietly kill yield.<\/li>\n\n\n\n
- Vision system<\/strong>: Determines whether the machine corrects reality or repeats bad assumptions. Lighting drift and dirty optics make \u201crandom\u201d errors that aren\u2019t random.<\/li>\n<\/ul>\n\n\n\n
Modern assembly research keeps landing on the same pillars\u2014visual positioning, trajectory planning, and force\/position coordination\u2014because those three decide whether automated assembly stays accurate at speed. ([PMC<\/p>\n\n\n\n
][3])<\/p>\n\n\n\n
Compliance and safety: the part procurement forgets<\/h2>\n\n\n\n
If you\u2019re in the EU (or selling into it), machine safety compliance is not optional. The legal baseline is moving from \u201cnice documentation\u201d to \u201cprove it.\u201d Regulation (EU) 2023\/1230<\/strong> updates the machinery framework and tightens expectations around safety, documentation, and modern control systems. (EUR-Lex<\/a>)<\/p>\n\n\n\n
Why does that matter for an operating cycle article? Because guarding, interlocks, and safe motion design can change how operators interact with feeders, maintenance access, and recovery steps after an error. Safety design shapes downtime.<\/p>\n\n\n\n
Real-world pressure: automation is rising, and SMT is in the blast zone<\/h2>\n\n\n\n
Here\u2019s one stat that matters because it explains the business pressure behind faster, more reliable cycles: China hit 470 industrial robots per 10,000 workers in 2023<\/strong>, overtaking Germany, according to an IFR report covered by Reuters. (Reuters<\/a>) The IFR\u2019s own release frames it as a rapid shift in factory automation density worldwide. (IFR International Federation of Robotics<\/a>)<\/p>\n\n\n\n
And in Europe, EMS companies reported \u20ac57.3B<\/strong> in 2023 PCBA revenues (11% growth) in a large survey summarized by the Global Electronics Association site, which also flagged inventory and consolidation pressures. (electronics.org<\/a>)<\/p>\n\n\n\n
Translation: more boards, more variants, tighter lead times, fewer excuses. Your pick-and-place operating cycle can\u2019t be \u201cgood enough.\u201d It has to be explainable and controllable.<\/p>\n\n\n\n
![\"PCB](\"https:\/\/pickandplacemachine.com\/wp-content\/uploads\/2026\/01\/PCB-Handling-Machines1.jpg\")
<\/a><\/figure>\n\n\n\nOperating cycle checkpoints you should measure (not argue about)<\/h2>\n\n\n\n
- Nozzle<\/strong>: Determines pickup reliability and placement stability. Wear, contamination, and wrong nozzle choice quietly kill yield.<\/li>\n\n\n\n
- Feeder<\/strong>: Determines component presentation consistency. Bad feeders create mispicks, skew, and time-wasting retries.<\/li>\n\n\n\n
- 350 \u00d7 0.072 = 25.2 seconds<\/strong><\/li>\n<\/ul>\n\n\n\n
- 3,600 seconds\/hour \u00f7 50,000 placements\/hour = 0.072 seconds per placement<\/strong><\/li>\n<\/ul>\n\n\n\n