SMT PCB assembly: what happens on the line, and what you control
Solder paste printing, paste inspection, pick and place, reflow, AOI and X-ray: what each station on an SMT line does, which design and file choices change the result, what drives the quote, and what to ask your assembler.
Filed under Engineering and services
Short answer: SMT PCB assembly is the process of soldering surface-mount components onto pads on the surface of a printed circuit board. The line runs in a fixed order: solder paste is printed through a stencil, a solder paste inspection (SPI) machine measures every deposit, a pick-and-place machine sets each part into the wet paste, a reflow oven melts the paste into joints, then automated optical inspection (AOI) and X-ray check the result before electrical test. Most defects start at the paste print, and most of the cost of a small batch is setup, not machine time.
If you are buying a board rather than building one, you do not need to run an SMT line. You do need to know what happens on it, because the files you send, the parts you choose and the questions you ask decide how the line behaves on your job. This note walks through SMT PCB assembly station by station, explains which design and file choices change the result, and lists what drives the quote. It pairs with turnkey PCB assembly, which covers who buys the parts, and PCBA: what to ask before you send Gerbers, which covers the file package in detail.
SMT vs through-hole: what changes
Surface-mount technology (SMT) is the assembly method; a surface-mount device (SMD) is the component it places. SMD parts sit on flat copper pads and are soldered all at once in an oven. Through-hole parts push their leads through drilled holes and are soldered afterwards, by wave, selective soldering or hand. Almost every real board mixes both: small resistors, capacitors and chips go on by SMT, while heavy connectors, transformers and parts that take mechanical abuse often stay through-hole.
That split matters for cost. Every through-hole part adds a separate soldering step after reflow. Queen EMS, an assembler, describes the standard flow and the exceptions: "Double-sided boards run steps 2–7 twice, once per side, and boards with through-hole parts add a wave, selective or hand-soldering step after reflow." If a through-hole part has a surface-mount version that does the job, ask your designer whether it can switch.
The SMT line, station by station
1. Solder paste printing. A thin stainless steel stencil, cut with an opening over every pad, is aligned to the board and a squeegee pushes solder paste through it. The paste is tiny solder spheres suspended in flux. Jon Cusack, founder of Cusack Electronics, is blunt about where the trouble starts: "Solder paste printing produces most of the defects the rest of the line then spends its time catching, and the inspection you pay for downstream is largely cleanup after a bad print." Stencil thickness and opening shape set how much paste lands on each pad; Cusack notes that "Most stencil designers work to an area ratio above 0.66 for that reason."
2. Solder paste inspection (SPI). A 3D scanner measures the height, area, volume and position of every deposit before any part is placed. A bad print found here is wiped and reprinted. Queen EMS puts the economics in one line: "Fixing a print costs a wipe; fixing it after reflow means rework on soldered parts." Not every shop runs SPI on simple boards, but on fine-pitch parts or packages with hidden joints it is worth asking for. Cusack's view: "SPI is the only station on the line that inspects a cause rather than an effect."
3. Pick and place. The placement machine picks each part from a reel, tray or tube with a vacuum nozzle, checks it with a camera and sets it on its paste deposit at the coordinates in the centroid (XY) file. Nothing is soldered yet; the wet paste holds the parts. The machine finds the board's exact position from fiducials, small copper marks on the board.
4. Reflow. The populated board rides a conveyor through an oven with several heating zones: preheat, soak, reflow above the solder's melting point, then controlled cooling. Every joint forms at the same time. The right profile depends on the paste and on the least heat-tolerant part on the board, and it should be checked with a thermocouple-instrumented board. Cusack's warning applies to any number you read, including ones here: "Never profile from a number found in an article, including this one."
5. AOI and X-ray. Cameras compare every part and visible joint against a reference and flag missing, shifted or rotated parts, bridges and thin solder. Cameras have a limit, and Cusack names it: "What AOI cannot do is see underneath a part." Ball grid arrays (BGAs) and no-lead packages hide their joints under the body, so Queen EMS states that "X-ray is needed for BGA balls, QFN center pads and other hidden joints." If your board has a BGA or a QFN with a center pad, ask whether X-ray is in the quote and what sample rate it covers.
6. Test, then cleaning and coating. Flying-probe or in-circuit test checks the assembly electrically, and functional test checks that the board does its job. Boards that will be conformal-coated are usually cleaned first, so no flux residue is sealed under the coating. On a first build, this is also where a first article inspection signs off one board before the rest of the lot continues.
Double-sided and mixed-technology boards
A board with parts on both sides runs the print, place and reflow sequence twice. The lighter side usually goes first; on the second pass the board is upside down and surface tension holds those parts in the molten solder. Cusack: "Heavy parts on the bottom get glued or moved to the top instead." Through-hole parts then go on after both reflow passes. Each extra pass is more setup, more handling and more heat on the parts that already went through the oven once, so a single-sided layout, where the design allows it, is the cheaper and simpler build.
What you control: the design and the files
Most SMT problems a buyer can prevent are decided in the design files.
- Fiducials. The machines align to them. JLCPCB's guide advises: "Include at least two global fiducials per board, placed diagonally." and "Place fiducials at least 5 mm from the board edge to avoid interference from panel rails or V-scores."
- The centroid file. It tells the machine where each part goes and which way it faces. Cusack: "Rotation errors in the centroid file are the single most common data problem I see." A polarized part placed backwards solders perfectly and fails test on every board, so check the first board against the drawing before the run.
- The paste layer. Send it. Without it the stencil is drawn from the pad layer, and paste volume is the variable most defects trace back to.
- The BOM. Manufacturer part numbers and approved alternates, not descriptions. An unapproved substitution can make a production board behave differently from the prototype you tested.
- Moisture-sensitive parts. Many plastic packages absorb moisture and can crack in the oven. They arrive in sealed bags, and as Cusack puts it, "Once a bag is opened, a clock starts." Ask how the assembler tracks open time and when it bakes parts.
- Panelization. Small boards are usually built several to a panel with rails and breakaway tabs. Ask the assembler for its preferred panel before the board is laid out, not after.
What drives an SMT assembly quote
An SMT quote is mostly four things: one-time setup, placements, special processes and parts. Setup covers the stencil, machine programming and loading feeders, and it is paid once per build whatever the quantity. Cusack explains why that dominates small runs: "Setting up the 400-part board takes far longer, which is the real reason small batches cost more per unit." One assembler's published guide, from FastTurn PCBs, lists setup per run and a per-placement rate, and its worked example comes to this: "Setup and placement rates dominate at low volumes"; at higher quantities the parts themselves become the biggest line. Treat any published rate as that shop's pricing, not a market price.
Things that raise the number: a second side, fine-pitch parts and BGAs (and the X-ray that comes with them), through-hole parts that need a separate soldering step, conformal coating, test fixtures, and a BOM full of parts with long lead times. On lead time, Cusack is direct: "Component availability is usually the long pole." A board can sit ready to run for weeks waiting on one part. If you are comparing quotes, ask each supplier to split one-time charges from per-board cost, as covered in what NRE cost is, so you can see which numbers repeat on a reorder.
What to ask your SMT assembler
- Is SPI run on every board, and do you keep the data?
- Is AOI run after reflow, and is X-ray included for BGAs and bottom-terminated parts? At what sample rate?
- Which IPC workmanship class will the board be built and inspected to?
- How do you control moisture-sensitive parts once a bag is opened?
- Will you send photos and measurements of the first board before the lot runs?
- Which charges are one-time, and which repeat on a reorder?
- Will any part be substituted without my written approval? The answer should be no.
If the electronics are one piece of a larger product, our note on EMS contract manufacturing covers who owns the design, the test fixtures and the leftover parts.
How JCH runs an SMT board
Jack runs board builds for US companies with partner assemblers. He reviews the file package and BOM before anything is quoted, so the quotes compare like for like. He settles the workmanship class, alternates and one-time costs in writing, and he checks the first boards against your drawing before volume. The steps are on the electronics and PCBA process page, and inspection is covered in how lots are inspected. If you have a board ready to quote, send the package.
Straight answers
What is SMT PCB assembly?
SMT PCB assembly is the process of soldering surface-mount components onto pads on the surface of a printed circuit board. Solder paste is printed through a stencil, parts are placed into the paste by a pick-and-place machine, and a reflow oven melts the paste to form every joint at once. The board is then inspected by AOI, X-ray where needed, and electrical test.
What are the steps of the SMT assembly process?
Solder paste printing, solder paste inspection (SPI), pick and place, reflow soldering, automated optical inspection (AOI) and X-ray for hidden joints, then electrical test and, where specified, cleaning and conformal coating. A double-sided board runs print, place and reflow twice, and through-hole parts are soldered after reflow.
What is the difference between SMT and through-hole assembly?
SMT parts sit on flat pads on the board surface and are soldered together in a reflow oven. Through-hole parts have leads pushed through drilled holes and are soldered afterwards by wave, selective or hand soldering. Most boards mix both, and each through-hole part adds a separate step.
Why do small SMT batches cost more per board?
Setup is paid once per build whatever the quantity: the stencil, machine programming and loading and checking feeders. On a small run that fixed cost is spread over few boards. Placement time per board is short once the line is running.
Does every SMT board need X-ray inspection?
No. Cameras (AOI) can check visible joints. X-ray is needed when joints are hidden under a part, such as BGA balls and the center pads of QFN packages. If your board has those, ask whether X-ray is included and at what sample rate.
How this note was made: researched and drafted with AI assistance, checked against the sources listed below and edited at JCH Design & Manufacturing.