SMT Processing Cost: What Drives the Price of an Assembly Run

SMT processing cost is quoted as a single number, but it is assembled from a handful of measurable items: setup time, tooling, the number of placement points, the number of unique components, inspection requirements and the volume of the run. Understanding how those items combine explains why two boards of similar size can be quoted very differently.

It also shows which design decisions reduce cost, and which ones merely move cost from fabrication to assembly.

Setup Time Is Charged Once, Then Spread

Every assembly run begins with assembly setup: loading feeders, verifying part orientation, programming or loading the placement program, and proving the first article. This cost is nearly fixed regardless of quantity, so it dominates the price of a small batch and disappears into the noise on a large one.

The number of unique components drives setup time more than the number of placements. Ten thousand placements of forty parts load faster than two thousand placements of four hundred parts, because feeder positions, nozzle changes and verification steps scale with variety.

Tooling: Stencil, Fixtures and Programming

A stencil is required whether the run is large or small, and its price depends on size, thickness and process. Laser-cut stainless steel with electro-polished apertures is standard; step stencils, which have two thicknesses for parts that need different paste volumes, cost more but can remove a whole process step.

Fixtures and carriers add to the tooling list for thin boards, flex circuits and panels that cannot be handled flat. Programming and first-article inspection are also part of tooling in practice, because they consume engineering time that is not recovered per unit.

Placement Points and Machine Time

Placement points are the hourly machine time consumed by the program. Each point is one pick, one alignment step and one placement, and the machine time per point depends on package type. Small chip components are placed at very high rates; fine-pitch ICs, connectors and odd-form parts take longer because they need vision alignment and slower acceleration.

A rough way to compare designs is to count points by class. Two boards with the same total placement points can differ substantially in machine time if one is mostly 0402 passives and the other is mostly fine-pitch devices with connectors.

Component Count and Unique Part Numbers

Component count affects two cost lines: placement time and material handling. The number of unique part numbers affects purchasing, incoming inspection, reel setup and the risk of a wrong part on a feeder.

Consolidating values is the most direct way to reduce this. Using one 100 nano-farad capacitor value across a design instead of four similar values removes three feeder positions and three purchase lines, and it reduces the chance of a misload that stops the line.

Inspection and Test Requirements

Inspection adds both time and equipment cost. Automated optical inspection is charged per board or per panel and is fast. X-ray inspection is slower and is billed per side or per board, so it is normally applied to assemblies that actually contain hidden joints.

Functional test and in-circuit test add the cost of fixtures and test time. Where a design can be verified with a simple power-on test, that should be reflected in the drawing, because unspecified test requirements are quoted conservatively.

Volume, Panelization and Changeover

Volume changes the economics of every other line item. A run of fifty boards pays setup and tooling in full; a run of five thousand amortizes them across the lot. Panelization matters here, because more boards per panel means more units per machine cycle and lower handling cost per unit.

Repeated orders of the same design benefit from stored programs and repeat tooling. A design that changes on every revision forces the setup cost to be paid again, which is a strong argument for consolidating changes into one revision rather than issuing several.

Where Design Choices Move the Number

Design decisions that reduce SMT processing cost are mostly unglamorous: fewer unique part numbers, standard footprints, adequate pad spacing, fiducials on every panel, consistent orientation of polarized parts, and test points large enough to probe.

Decisions that look free during design and cost money at assembly include zero-ohm jumpers instead of routed copper, tight component-to-component spacing that forces slower placement, and packages that require an additional reflow or a selective soldering step.

Material Cost Versus Process Cost

Material is quoted separately from process in most quotations, and the two behave differently. Material cost scales almost linearly with quantity and with component count; process cost drops steeply with quantity because setup is amortized.

That difference explains a common surprise. A design with an expensive laminate and few parts can be cheaper per unit at high volume than one using standard material with many unique parts, because the second design keeps paying process cost at every order.

Where both are significant, ask for the split at two quantities, for example one hundred and one thousand units. The slope between the two numbers is the per-unit cost, and the intercept is the one-time cost.

Reading a Quotation

A useful quotation separates one-time charges from per-unit charges. One-time items are stencil, programming, fixturing and first-article inspection. Per-unit items are machine time, inspection and material handling. Requesting that split makes it possible to compare suppliers and to see which item dominates at your volume.

It also reveals where a change is worth making. If tooling dominates, reducing part variety is less valuable than ordering a larger batch. If machine time dominates, reducing placement points and simplifying packages has the greater effect.

Related reading: placement order and pad positioning, design guidelines for manufacturable boards, and PCBA development process.

Cost Terms That Are Frequently Misread

Quotations often list a single board price that already includes component markup. Ask whether the quoted figure covers the bill of materials or only the processing service, because the two are not comparable and the difference is often larger than the processing cost itself.

Minimum order quantity and panel utilization change the effective price as well. If a board does not tile economically on a standard panel, the quoted unit price carries the unused panel area, and the fix is usually a small change to board dimensions or to the panel layout.

Finally, treat expedited production as a separate line rather than as a premium on the whole quote. Priority scheduling affects the queue, not the process, and the per-unit processing cost should not change because of it.

A final word on comparison shopping: quote the same design, at the same quantity, with the same inspection requirements. A lower price that excludes X-ray, functional test or component procurement is not a lower price at all.

FAQ

Why is a small batch so expensive per board? Because assembly setup, stencil and first-article inspection are one-time costs that few units must absorb. Per-unit machine time is usually a small part of the total at low volume.

Does adding more boards per panel reduce cost? It reduces handling and machine cycles per unit, which lowers per-unit cost, but only when the panel still fits the line and the placement program remains efficient. Extremely dense panels can slow the line instead.

What reduces cost most for a new design? Fewer unique part numbers and standard footprints. Both attack setup time, which is the largest controllable item in a low-volume quotation.