Blog / OEE
OEE · Guide

OEE for SMT Lines: Boards, Feeders, Placement Rate

OL By OEE Lab Editorial |Updated 2 November 2026 |7 min read
Short answer

Measure OEE across the whole SMT line: count panels at the end of the line, take the ideal cycle from the slowest machine for that product, and treat only first-pass boards as good. Changeovers and long stops reduce availability, while feeder stops under your threshold and slow cycles reduce performance.

Line first, machine second

An SMT line is a chain: printer, inspection, placement, reflow, optical inspection. Boards travel through in sequence, so the line can go no faster than its slowest step for that product.

That is why line OEE is the number to report. A placement machine with a high utilization can still be held up by a stalled oven, and line OEE vs machine OEE shows the difference.

Machine-level data still matters when you look for a cause. The line number tells you how much you lost, and the machine logs tell you where.

Set the ideal cycle from the program, not the brochure

SMT numbers often go wrong here. The headline rate on a placement machine, in components per hour (CPH), comes from a test under set conditions, and your boards are different.

Different parts, nozzle changes, large components and board size all change the real rate. Using the brochure number can make performance look terrible on every product.

Use the ideal cycle for each product: the time per panel at the bottleneck machine, from its optimized program with no waiting.

Placements per panel: 480
Program rate at the bottleneck: 43,200 CPH = 12 per second
Ideal cycle = 480 / 12 = 40 s per panel
Hypothetical product. Take the rate from a demonstrated best run of this program, and include any fixed board transfer time.

If two machines share the placements, the slower one sets the pace. Check whether the printer or oven holds the line back on some products, as the bottleneck can move with the mix.

See ideal cycle time for the general method.

Worked example: one shift

Example: a hypothetical line is scheduled for 480 minutes, and the product above has a 40-second panel cycle.

  • Changeovers: 3 x 20 min = 60 min
  • Long stops, such as feeder faults and a printer issue: 45 min
  • Run time: 480 - 60 - 45 = 375 min
  • Panels at line end: 480
  • Panels passing first time: 468

Availability = 375 / 480 = 78.1%, performance = (480 x 40 s) / (375 x 60 s) = 85.3%, and quality = 468 / 480 = 97.5%.

OEE = 78.1% x 85.3% x 97.5% = 65.0%. As a check, 468 x 40 s = 312 minutes of full-rate good output, and 312 / 480 = 65.0%.

Scheduled time480 min
Changeovers60 min
Long stops45 min
Speed and short stops55 min
Quality8 min
Full-rate good output312 min = 65.0%
Where the 480 minutes went in the hypothetical shift. The method follows Hansen's standard OEE factors, and the four losses plus 312 minutes add to 480.

The speed bar hides many small feeder events. That is the next thing to dig into.

Feeder stops: the small loss that adds up

A placement machine consumes parts from many reels. Each empty reel, feeder error or pick failure can halt the machine for a short time.

Those stops are often brief, so they fall below the stop threshold and appear as a speed loss with no label. What are micro-stops explains why this happens.

You can estimate the size of the reel problem before you measure it.

  • Placements in the shift: 480 panels x 480 placements = 230,400
  • Reel ends, with an average of 4,000 parts per reel: 230,400 / 4,000 = about 58
  • Time if each stops the machine for 30 seconds: 58 x 30 s = 29 minutes

Splicing a new reel onto the old one lets the machine keep running through most of these changes. Reel-low warnings and pre-loaded feeders can also help, and they turn this estimate into a target.

Look at pick errors as well. The machine logs usually show pick and placement errors per feeder and nozzle, so sort them, and take the worst few to the pick and place troubleshooting guide.

What goes where on an SMT line

EventOEE factorData source
Product change, feeder setup, first article checkAvailabilityLine state, operator tag
Long machine fault or oven alarmAvailabilityMachine log or signal
Waiting for boards or for partsAvailabilityTransfer signals, operator tag
Short feeder and pick stopsPerformanceMachine log
Slow transfer, or program below demonstrated ratePerformancePanel timing
True defect found at inspection, reworked or scrappedQualityOptical inspection, after review
Inspection false call, board passes after reviewNot a quality lossReview station
A consistent rule set. Reworked boards do not count as good in OEE, even when they ship.

Counting boards correctly

  • Pick one unit. Count panels or single boards, and keep it the same as the unit in your ideal cycle. A 6-up panel counted as 6 boards against a per-panel cycle inflates performance six times.
  • Count at the line end. After reflow and inspection, you count what truly made it through.
  • Handle double-sided boards. Each side is usually one pass through the line, so treat each side as its own product with its own cycle.
  • Agree on false calls. If inspection stops a good board, the board is good, and the time goes to speed loss. Track the false call rate apart, as good count explains.

The neighbor metric, first pass yield, covers the quality figure in more depth.

Where the data comes from

SMT lines are often well equipped, but the data can be scattered across several vendors' tools.

  • Machine logs: placement machines keep detailed records of errors and counts, often exportable or readable through vendor software.
  • SMEMA signals: the common board-handoff signals between machines, "machine ready" and "board available", can show starved and blocked time at each handoff. A wrong tap can upset board transfer, so leave it to a qualified electrician or the machine maker, listening only.
  • IPC-CFX (IPC-2591): an open standard for machine data, found on some newer equipment. Hermes (IPC-HERMES-9852) passes board data along a line and is the successor to SMEMA, so check which of these your machines support.
  • A sensor on the line-end conveyor: a simple panel counter that suits most lines, fitted with the conveyor stopped. See part counting sensors.

Older lines without those interfaces can begin with that counter and a state signal. Routes are listed in how to connect old machines.

Changeovers on a high-mix line

Many SMT lines run lots of short jobs, so changeovers can outweigh every other loss. Here, 60 of the 168 lost minutes came from three product changes.

The fixes are familiar: pre-kit feeders offline, group products to share feeder setups, and verify programs before the stop. SMED changeover reduction gives the method, and changeover frequency shows the tradeoff with batch size.

Include the first article check in changeover time. The line is not producing sellable boards until it has passed.

Traps in SMT numbers

  • Using brochure CPH. It makes every product look slow, so people stop trusting the number.
  • Reporting only placement machine utilization. This can look healthy while the line waits for boards.
  • Mixing units. Panels and boards in one report lead to impossible results. Over 100% is a warning sign, covered in OEE over 100%.
  • Skipping the oven. It runs at a steady conveyor speed, so a slow profile can quietly set your pace. See reflow oven troubleshooting for faults.

Your first week

Set one ideal cycle per product, from a demonstrated program rate, and start a panel count at the line end. Then pull a week of feeder and pick errors, and rank them by time lost.

For more on this sector, see OEE and downtime in electronics.

See it live in Fabrico

Small stops are easy to lose in a shift total. Fabrico, a light MES for OEE, takes stops, speed and counts live from PLCs or IoT sensors, with add-on AI cameras counting output where a machine has no usable signal, shows OEE per line and shift, and lets operators tag the cause at the line. See how Fabrico tracks line stops.

What people ask next

How do you calculate OEE for an SMT line?

Multiply availability, performance and quality for the whole line. Use run time over scheduled time, panels x ideal cycle over run time, and first-pass panels over total panels. The worked example above shows each step.

Should I use the machine's rated CPH as the ideal rate?

Usually not. The headline rate comes from a test under set conditions, and real boards run differently. Use a program-specific ideal cycle at the bottleneck, as in ideal cycle time.

Do feeder stops count as downtime or performance loss?

It depends on your stop threshold. Longer stops count against availability, and short ones show up as lost speed. Set the line with stop threshold for OEE, and keep it fixed so that the trends mean something.

Are boards reworked after inspection good parts?

Not for OEE quality, which counts only boards right the first time. Track the rework separately. A board that was flagged falsely and passes review is still a good board.

Can I calculate OEE for each machine in the SMT line?

Yes, and it helps to find causes, but keep the line number for reporting. A machine that is starved or blocked looks poor through no fault of its own. Blocked and starved explains how to read that time.

See OEE per line and shift on your own machines

Fabrico is a light MES for OEE. It takes stops, speed and counts from PLCs or IoT sensors, with add-on AI cameras counting output where a machine has no usable signal, and operators tag the cause at the line.

Book a Fabrico demo

← More OEE Lab guides · Pick and place troubleshooting →

OEE essentials: OEE calculator · What is OEE · Six big losses · Availability · TEEP · Cost of downtime · OEE benchmark · Free OEE tracker · Free OEE software · How to track downtime · How to implement OEE · Light MES vs full MES