How Many Operators Does an Automated SMT Line Actually Need? A Staffing Method You Can Recompute

A planning method, not a magic number — count the operator-minutes, staff positions instead of tasks, and let automation remove tasks one by one.

Ask five factories how many operators their SMT line needs and you will get five different answers — and all five can be right. The machines may look the same, but the staffing is decided by what the people still have to do: who loads boards, who splices reels, who answers the SPI screen, who runs the changeover, and who fixes the boards the inspection stations reject. Automation does not remove "operators" in general. It removes specific tasks, one at a time. Count the tasks and the number follows.

An SMT production floor — printer, placement and inspection machines in line. No operators are visible in this frame; the staffing question is how many people this floor needs per shift, not how many machines it holds. Photo: JHIMS YSM20R product page (Changsha case image).

1. First, fix what you are counting: positions, people, or roster

Most staffing arguments start with three different numbers confused as one. Separate them before you calculate anything.

• A position is a job that must be covered for the whole shift — for example, "line operator". A position exists even if its work is intermittent.
• People per shift is how many individuals are on the floor at any moment. Two part-time positions can be one person only if the tasks never collide — and on a running line, they usually do.
• The roster (headcount) is how many people you employ to keep those positions filled across holidays, sickness, training and days off. A line that needs 2 people per shift, running two shifts, does not employ 4 people. It employs more, because no one is present every working day.

This article answers the first two: positions per shift, then people per shift. The roster follows from your own absence rate in Section 6. If a supplier quotes you "one operator per line", ask which of the three numbers they mean — the answer is usually the position count, and the roster is larger.

2. The task inventory: everything a person still does on an SMT line

Walk the line for one shift and write down every job a person does, including the ones that happen "between" machines. The table below is the inventory to start from. Cross out the rows your line has automated, and add the rows your line still does by hand.

Two rows deserve attention because they are the ones most staffing plans forget. The first is first-article check: after every changeover someone must confirm the first boards before the line runs at full speed, and that person cannot be the same person who is still loading feeders. The second is false-call review: SPI and AOI do not only find defects, they also flag good boards. Someone has to look at each flag and decide. On a line with a high false-call rate, review work alone can fill half a position. Our guide to deciding which inspection step to buy first, SPI or AOI first, explains why the two stations create different review loads.

3. The method: work content divided by takt, then the coverage rules

The calculation has three steps, and the order matters. It follows the standard lean staffing method: operators equal total work content divided by takt time, rounded up — and the roster then allows for absence.

Step 2 alone almost always under-counts, because it assumes one person can teleport between stations and do two things at once. The coverage rules in Step 3 are what turn a minutes total into a staffing plan. If your Step 2 result is 1 person and your Step 3 result is 2 positions, staff 2 — the second person is not waste, they are the reason the line does not stop every time a reel runs out during a changeover.

4. Worked example: one line, one shift, stated assumptions

All numbers in this section are illustrative planning numbers. They show the method working; they are not measurements from any real line and not a benchmark. Replace every input with your own timed observations before you use the result.

Assumptions: an 8-hour shift is 480 minutes; breaks and a start-of-shift meeting take 30 minutes, leaving 450 available minutes. The target is 600 good boards per shift, so takt is 450 × 60 ÷ 600 = 45 seconds per board. One product family runs all shift, with one changeover.

The arithmetic: work content per board is 400 × 60 ÷ 600 = 40 seconds. Operators on pure minutes = 40 ÷ 45 = 0.89, rounded up to 1 person. That is the floor, not the answer.

The coverage rules change it. The line needs one person present for the full shift to load boards, watch the line and respond to stops — that is one dedicated position. Feeder work, kitting and the changeover happen in parallel with production and cannot wait for that person to be free, and the changeover needs two people at the same moment. So the plan is 2 positions per shift: a line operator and a second person covering material, feeders, changeover support and rework between kitting runs. As a cross-check, one published electronics-manufacturing manpower plan budgets about three fixed SMT staff per line per shift, with kitting, rework and line leadership shared across lines — the same shape as this result, with shared roles counted separately.

5. Three line types: what changes, and which roles disappear

The table below applies the same method to three line configurations. The position counts are outputs of the worked example, not benchmarks — change the inputs and the outputs move. What does not move is which roles each configuration keeps.

Note what automation never removes: the first-article check after a changeover and the rework bench. A line can run with one operator only if someone else — a shared technician, the next shift, or a supervisor — owns those two jobs when they arise. A staffing plan that counts one operator and assigns nobody to rework has not saved a person; it has hidden one.

6. From positions per shift to the roster: cover absence honestly

Positions per shift are not headcount. People take holidays, fall sick and go on training. The roster formula is simple: people per crew = positions per shift ÷ (1 − absence rate), rounded up.

Using the worked example with an illustrative absence rate of 8%: 2 ÷ (1 − 0.08) = 2.17, rounded up to 3 people per crew. Running two shifts therefore employs 6 people to keep 2 positions filled around the clock — 4 filled slots plus cover. Put your own absence rate into the formula; a line with 15% absence needs a different roster from a line with 5%, even if the machines are identical. And cross-train at least two people per critical position. A position that only one person knows how to run is a single point of failure, and it will fail on the day that person is away — so build the training plan before you need the cover, not after.

7. Which automation actually removes a task — with named examples

Automation earns its place in the staffing table only when it deletes a row from the inventory. Four examples from machines JHIMS supplies show how the test works. Each figure below is the manufacturer's published specification for that model; confirm it against the machine's own datasheet before you plan around it.

• Feeder reloads: the Yamaha YSM20R offers a nonstop feeder exchange system, so component reloads do not stop production, and a product changeover in under 30 seconds with a single operator and automatic program recall. That deletes the "stop the line to reload" minutes and shrinks the changeover row — it does not delete the person who prepares the next feeder carriage.
• Inspection review: the S630 3D SPI programs from Gerber in under 5 minutes and changes over in under 3 minutes, and its closed-loop control sends XY offset data back to the printer and triggers stencil cleaning when it detects missing prints. Fewer false calls and automatic correction shrink the review row; they do not remove the person who dispositions the boards the system still flags.
• Stencil cleaning: the L5 solder paste printer provides combined dry, wet and vacuum stencil cleaning modes, and the M8 printer is the high-precision counterpart in the same line. Automatic cleaning deletes manual cleaning stops from the shift — but only if the cleaning cycle is actually enabled in the program.
• Downstream fastening: once boards leave the SMT line, screw fastening is still manual work unless it is automated. A desktop machine such as the ZM320 sits on a workbench and fastens while the operator loads the next module — the operator's job changes from driving screws to loading and unloading, which is fewer minutes per module, not zero minutes.

The ZM320 desktop screw fastening machine: gantry beam and dual worktable. Frame from the original ZM320 video supplied by JHIMS — not a YouTube screenshot, and not an SMT placement machine. Shown here as a downstream automated station: it changes the operator's job from driving screws to loading and unloading.

The test to apply to any automation claim is short: name the task it deletes, name the minutes it deletes, and name who still does the rest. If a vendor cannot answer all three, the headcount saving is a sales claim, not a staffing plan. For the wider picture of how these stations connect into one line, see how SMT line integration changes PCB assembly — and treat its headline efficiency figures as that article's claims, not as numbers this article has verified.

8. Common mistakes that produce the wrong number

• Counting machines instead of tasks. Seven machines do not need seven people, and one "automated" line does not need zero. Count minutes, not machines.
• Forgetting simultaneity. Two 30-minute jobs that must happen at the same time need two people, even if the shift total fits one person.
• Planning for the steady state only. Changeover days, new-product introduction and the first shift after a holiday all need more hands than a steady Tuesday. Staff the peak you actually run, or schedule the peak when the extra hands exist.
• Leaving rework unstaffed. Every flagged board needs a decision and usually a repair. No rework cover means flagged boards pile up beside the line until someone senior stops their own job to clear them.
• Treating feeder count as free. A board with 140 feeder positions needs more splicing, more verification and more kitting than a board with 40 — even on the same machine. Label feeders are part of this cost, which is why feeder discipline matters; see when tape-and-reel is not enough for FUJI NXT label feeders.
• Copying another factory's number. Headcount travels badly. Different board mix, different changeover frequency and different inspection rules change the task list, and the task list is the whole calculation.

A 10-minute staffing check for your own line

1. Write down your target boards per shift and your available minutes. Compute takt.
2. Time each task in Section 2 for one real shift. Use observed minutes, not remembered ones.
3. Divide total operator-minutes by available minutes. That is your floor.
4. Mark every pair of tasks that must happen at the same time. That sets your positions.
5. Add the roster step with your real absence rate.
6. Re-run the table for your busiest day — most changeovers, most feeders, newest product. Staff that day, or reschedule it.

FAQ

Can one operator run a whole SMT line?
On pure minutes, sometimes — the worked example needs 400 operator-minutes against 450 available. In practice, no, not alone: changeover needs two people at once, breaks need relief, and a feeder running out during first-article check needs a second pair of hands. One dedicated operator plus shared support is the realistic minimum for a single line.

Does buying a faster placement machine reduce headcount?
Rarely by itself. A faster mounter shortens the machine cycle, not the operator's task list. Headcount falls when automation deletes tasks — nonstop feeder exchange, automatic stencil cleaning, closed-loop inspection — not when a machine runs faster while the same person still loads, splices and reviews.

How many operators do two lines need — double?
Not quite. Shared roles (kitting, rework, line leadership, maintenance) spread across lines, so two lines need fewer than twice the staff of one. But each line still needs its own dedicated line operator during production. Count shared roles once and dedicated roles per line.

Should the line leader be counted as an operator?
Count them separately. A line leader who also runs a position cannot lead when the position needs them. If the leader covers breaks or changeovers, list that as relief cover in the roster, not as a free operator.

What if our boards change every day?
Then changeover minutes dominate the table and the position count rises with changeover frequency, not with board volume. High-mix lines should staff the changeover pattern first and the steady-state running second — and fast-changeover features (stored programs, exchange carriages, sub-30-second changeover) pay back in headcount, not just in uptime.