Placement speed lies.
That is the uncomfortable truth behind many pick-and-place machine brochures, especially when a glossy spec sheet screams 80,000 CPH or 120,000 CPH while the actual SMT line, under a real job file with odd feeders, fiducial checks, nozzle changes, reject handling, and board transfer time, crawls at half of that number. Why does this still happen?
Because CPH is easy to sell and hard to audit.
Placement Speed, or CPH, means components per hour: the number of electronic components a pick-and-place machine can place on a PCB in one hour. In theory, it is a clean productivity metric. In practice, it can be a trap if you do not separate theoretical machine CPH from real line CPH.
I would not buy a placement machine from a headline speed number alone. Not anymore. A better move is to compare the machine’s rated speed against the actual board mix, component count, feeder layout, nozzle strategy, changeover pattern, and upstream/downstream bottlenecks. That is where the money is.
And the money is getting tighter. In October 2024, IPC-reported North American EMS shipments were up 14.7% year over year, with a book-to-bill ratio of 1.25, suggesting demand was running ahead of billed supply in that sample. If demand rises and your line speed estimate is fantasy, your quote, labor plan, and delivery promise all break at the same time. IPC EMS October 2024 data via Electronics Sourcing (electronics-sourcing.com)
What Placement Speed CPH Actually Measures
Placement speed CPH measures how many components an SMT pick-and-place machine can mount per hour, usually expressed as “components/hour” or “cph.” The basic CPH formula is simple:
CPH = Total Components Placed ÷ Production Time In Hours
So if a machine places 48,000 components in 1.5 hours:
48,000 ÷ 1.5 = 32,000 CPH
Clean math. Dirty reality.
The formula only becomes useful when you define what “production time” includes. Does it include PCB loading and unloading? Vision alignment? Feeder replenishment? Bad pickup recovery? Panel transfer? Operator intervention? Program optimization? Maintenance stops?
Many vendors quote ideal placement speed. Factory managers need actual placement speed.
Theoretical CPH vs Real Component Per Hour
Theoretical CPH is measured under ideal test conditions: short travel paths, simple chip parts, optimized feeders, no meaningful changeover loss, no awkward component geometry, and no line-level disruption. Real CPH includes the uglier stuff: nozzle swaps, pickup errors, board handling, fiducial recognition, component mix, feeder position, and process pauses.
That gap matters.
A practical industry estimate is that real placement speed often lands well below theoretical speed; one SMT technical explainer places actual speed around 65% to 70% of theoretical speed when auxiliary production factors are included. SunzonTech’s SMT placement speed calculation guide (sunzontech.com)
Here is the hard rule: the placement machine is rarely the whole line.
A fast mounter behind a slow printer is not a fast line. A high-speed chip shooter starved by bad feeder loading is not high-speed production. A beautiful machine with weak spare parts support is a future downtime report wearing a showroom badge.
For buyers comparing machines, start with the broader equipment category first, then narrow down by board type, product mix, and expected output. A good starting point is reviewing available pick and place machines for SMT production before pretending one CPH number answers everything.
The CPH Formula That Actually Helps
For real production planning, I prefer this version:
Actual CPH = Total Accepted Placements ÷ Total Elapsed Production Hours
“Accepted placements” means good placements, not attempted placements. Scrap, mis-picks, rejected boards, and post-AOI placement faults should not be treated as productive output. That may annoy sales teams. Good.
Use this model:
| Metric | Formula | What It Reveals | Common Mistake |
|---|---|---|---|
| Theoretical CPH | Rated placements ÷ 1 hour | Machine marketing speed | Treating it as factory output |
| Actual CPH | Accepted placements ÷ elapsed run hours | Real production speed | Ignoring stops and rejects |
| Board throughput | Finished boards ÷ elapsed run hours | Customer-facing output | Forgetting component count per board |
| Line utilization | Run time ÷ available time | Asset usage | Counting waiting time as production |
| First-pass yield impact | Good boards ÷ total boards | Quality-adjusted speed | Separating speed from defects |
Here is a more honest example.
A PCB has 420 SMT components. The line completes 75 good boards in one hour.
75 × 420 = 31,500 accepted placements per hour
So the real component per hour is 31,500 CPH.
If the machine brochure says 55,000 CPH, the real efficiency is:
31,500 ÷ 55,000 = 57.3%
Not terrible. But not brochure speed.
Why CPH Gets Inflated
There are five usual suspects.
First, test boards are too friendly. A machine placing repeat 0402 capacitors from nearby feeders will look heroic. Add BGAs, connectors, shields, odd rotations, tray parts, and long travel paths, and the number drops.
Second, feeder setup is treated like it does not exist. It exists. It eats hours.
Third, nozzle planning is weak. One nozzle change can look small. Hundreds of micro-pauses across a shift become a throughput tax.
Fourth, the line is unbalanced. The pick-and-place machine may be waiting for the printer, SPI, reflow, AOI, loader, unloader, or operator.
Fifth, sales teams quote the biggest number because buyers keep asking for it.
That last one is the scandal nobody wants to say out loud.
The better buying question is not “What is the maximum CPH?” It is: “What CPH will this line hold on my actual board family for eight hours, including changeovers?”
For companies building at scale, the answer usually pushes them toward engineered high-speed SMT mass production lines, not isolated equipment shopping. For lower volume or high-mix electronics, the smarter design may be a prototype and small-batch SMT line where changeover speed beats headline placement speed.
How To Calculate CPH Step By Step
Use this process.
- Count total SMT placements per board.
- Count only good boards produced during the measurement window.
- Multiply good boards by placements per board.
- Divide by elapsed production time in hours.
- Compare actual CPH to rated CPH.
- Investigate the gap.
Example:
| Input | Value |
|---|---|
| Components per PCB | 680 |
| Good PCBs produced | 120 |
| Production time | 2 hours |
| Accepted placements | 81,600 |
| Actual CPH | 40,800 |
Calculation:
680 × 120 = 81,600 placements
81,600 ÷ 2 = 40,800 CPH
Now compare it with the rated machine speed. If the machine is rated at 70,000 CPH, the line is running at 58.3% of rated speed.
That number is not automatically bad. Some complex boards will never reach 80% of rated speed. The point is knowing why.
What Affects SMT Placement Speed Most
Component mix matters more than many buyers admit. Tiny passives are fast. Large ICs, BGAs, QFNs, connectors, odd-form parts, and tray-fed components slow the machine because they add vision checks, nozzle constraints, pickup complexity, and placement care.
Board layout matters too. If parts are scattered across a large panel, head travel increases. If feeders are poorly assigned, pickup distance increases. If the job uses too many unique parts, the machine spends more time managing variety than placing volume.
The feeder system is the silent killer. Weak feeders produce mis-picks, pauses, and operator babysitting. Strong feeder preparation can turn the same machine into a different animal. That is why SMT feeder selection and setup should be treated as a throughput decision, not an accessory purchase.
Nozzles deserve the same respect. A worn nozzle, wrong nozzle, or dirty nozzle does not merely reduce quality. It reduces speed through pickup retries, inspection failures, and stoppages. For mixed packages, investing in the right SMT nozzle configuration is often cheaper than blaming the machine.
How To Improve Placement Speed Without Creating Defects
Speed without yield is fraud.
The fastest line in the building is useless if AOI is catching tombstoning, skew, insufficient solder joints, and BGA alignment problems. Real optimization means raising accepted CPH, not just attempted placements.
Start with feeder optimization. Put high-volume components closest to the pickup path. Group parts by nozzle compatibility where possible. Reduce head travel. Eliminate lazy feeder maps copied from old jobs.
Then tune nozzle strategy. Minimize nozzle changes. Clean nozzles on a schedule. Track pickup errors by nozzle ID, not just by job. This is basic, but basic things make money.
Next, standardize panel design where engineering allows it. Better panelization can reduce board handling time and improve placement rhythm. Bad panel design forces the SMT line to pay for design convenience.
Finally, measure the whole line. Printer cycle time, SPI inspection, conveyor transfer, reflow loading, AOI speed, and operator response time all influence PCB assembly throughput. A real optimization project should look like a line audit, not a machine-speed argument.
For facilities building from scratch or replacing multiple process islands, a turnkey SMT line solution is usually more honest than buying a fast mounter and discovering later that the printer, feeder storage, reflow oven, or AOI process cannot keep pace.
The AI Server Lesson: Why CPH Pressure Is Rising
The placement speed discussion is not academic. High-density electronics demand is increasing, especially around AI servers, power modules, networking equipment, and industrial controls.
In March 2024, Reuters reported that Foxconn expected more than 40% revenue growth from AI servers in 2024, while the AI server market was expected to grow around 30% annually from 2023 to 2025. That kind of demand does not tolerate sloppy throughput math. Reuters on Foxconn AI server demand (Reuters)
HPE showed the same pressure from another angle. In June 2024, Reuters reported that HPE’s server revenue rose 18% year over year to $3.9 billion, while AI-server revenue more than doubled sequentially to $900 million and backlog hit $3.1 billion. Reuters on HPE AI-server demand (Reuters)
Translation: more complex boards, more high-mix pressure, more schedule compression.
And more chances to lie to yourself with a clean CPH formula.
Practical CPH Benchmarks Buyers Should Ask For
Do not ask only for rated placement speed. Ask for proof under conditions close to your own production.
| Buyer Question | Why It Matters |
|---|---|
| What is the IPC-9850 or IPC-9850A-style throughput figure? | It is more meaningful than casual marketing CPH. |
| What CPH can the machine hold on mixed components? | Mixed packages reveal real capability. |
| How many feeder slots are available? | Feeder limits affect changeover and setup time. |
| What is the mis-pick rate by package type? | Pickup errors destroy actual throughput. |
| How fast is changeover between two common jobs? | High-mix factories live or die here. |
| What local spare parts and service support exist? | Downtime is negative CPH. |
A serious vendor should not be offended by these questions. If they are, walk away.
FAQs
What is placement speed CPH?
Placement speed CPH is the number of components a pick-and-place machine can mount on a PCB in one hour, usually measured as components per hour. It is used to estimate SMT machine productivity, but real CPH must include production conditions such as board transfer, feeder setup, vision checks, nozzle changes, and downtime.
How do you calculate component per hour?
Component per hour is calculated by multiplying the number of good boards produced by the number of SMT components on each board, then dividing by total production hours. For example, 100 good boards with 500 components each in two hours equals 25,000 CPH, because 100 × 500 ÷ 2 = 25,000.
Why is actual CPH lower than rated placement speed?
Actual CPH is lower than rated placement speed because factory production includes delays that ideal machine tests often exclude. Board loading, feeder replenishment, fiducial recognition, component inspection, nozzle changes, pickup errors, maintenance, operator response time, and downstream bottlenecks all reduce the number of accepted placements completed per hour.
What is a good SMT placement speed?
A good SMT placement speed depends on board complexity, component mix, panel size, feeder layout, and production volume. A high-volume passive-heavy board may justify very high CPH, while a high-mix board with BGAs, connectors, and tray-fed ICs may produce lower CPH but still be highly efficient and profitable.
How can I improve placement speed?
You can improve placement speed by optimizing feeder positions, reducing nozzle changes, cleaning nozzles, grouping similar components, improving panelization, reducing machine travel distance, balancing printer and reflow capacity, and measuring accepted placements instead of attempted placements. The goal is higher good-output CPH, not just faster machine motion.
Is CPH the same as PCB assembly throughput?
CPH is not the same as PCB assembly throughput because CPH measures component placements, while PCB assembly throughput measures finished boards or panels over time. A board with 1,000 components and a board with 100 components can have the same board output rate but radically different component per hour values.
Conclusion
Before you buy a machine based on a placement speed number, run the real math. Count accepted placements. Audit feeder strategy. Check nozzle condition. Measure the whole SMT line, not just the mounter.
And if you want a line built around real output instead of brochure CPH, start with a proper SMT production line consultation. Your future delivery schedule will thank you.
