Small Batch PCB Assembly: Cost, Process and Trade-offs

Small batch assembly is a different business from volume assembly, and the difference is not just the quantity. At a few hundred boards the price is dominated by setup, the components are the hardest part of the job, and the process that produces the best result is often not the process used in volume. Understanding that structure is the fastest way to reduce the cost of a low volume build.

Where Small Batch Ends and Volume Begins

The boundary is not a single number, because it depends on the board and on the supplier. As a rough guide, the first ten boards are a prototype build, where the purpose is to learn. From ten to a few hundred is small batch, where the boards are used for field trials, qualification, pilot production or a product that simply sells in modest quantities. From a few hundred upwards, the economics of stencils, programming and fixtures start to work in the customer favour, and above a few thousand the process is a standard volume flow.

The practical consequence is that a design which will eventually be produced in volume should still be assembled differently for the first small batch, and it is worth planning the transition rather than treating the first run as a miniature version of the last one.

Why the Price Is Setup, Not Assembly

In volume, the price of an assembly is roughly the cost of the components plus the machine time. In small batch, most of the price is work that happens once: stencil fabrication, paste and profile setup, machine programming, feeder loading, first article inspection and documentation. The placement itself may take minutes.

That is why the unit price falls so sharply with quantity and why it is almost impossible to compare two suppliers on a per board basis at low volume. A quote that looks expensive may simply be charging the setup honestly, while a cheap quote may be planning to hand place parts, or may have underestimated the work and will return with a change order later.

The most useful thing a customer can do is reduce the setup. Using the same board panel for several variants, keeping component packages consistent between designs, providing accurate and complete BOM data, and avoiding parts that need special handling all reduce the one time work. On a pilot build, the time spent on a clean BOM and a correct assembly drawing is usually paid back before the boards are finished.

Components Are the Hard Part

At low volume the components, not the placement, decide whether a build goes smoothly.

Most parts are sold on reels with a minimum order quantity that is far above what a small batch needs. A reel of a specialised connector may hold a thousand pieces, and a pilot build needs thirty. The choices are to buy the full reel and hold stock, to buy cut tape or a partial reel from a distributor that offers it, or to accept a substitute part. Each has a consequence: stock ties up cash, cut tape may not be machine feedable, and a substitute changes the design.

Long lead time parts are the second problem. A single component on a forty week lead time can hold up a build that would otherwise take two weeks, and the usual mitigation is to check availability before the design is frozen rather than after the BOM is released. Where a part is scarce, designing in a second source, or making the footprint compatible with an alternative package, is worth the extra layout effort.

Counterfeit and out of specification parts are a real risk in the spot market, which is exactly where a small batch buyer ends up when a part is short. Buying from an authorised distributor costs more and takes longer, but it is the only reliable way to know what is in the bag. On a build that will support a qualification test or a customer evaluation, the saving is not worth the risk.

The sourcing model also matters. In a turnkey build the assembler buys the parts, which puts the availability problem on them and usually gets a better price than a small customer can. In a kitted build the customer supplies the components, which gives control and sometimes saves money, but it also means the customer carries the risk of a short shipment or a wrong part. Consignment, where the customer owns the stock but the assembler manages it, is a middle option that suits companies with an established component inventory.

Which Process to Use

The process for a small batch is chosen for flexibility rather than for cycle time.

The most common arrangement is a stencil, a semi automatic or manual printer, hand placement of the components, and a conveyor or batch reflow oven. Hand placement is viable because a few hundred boards of modest complexity can be populated by one or two people, and because setting up a machine for a single build is often slower than placing the parts by hand.

Where the board is dense or the quantity is at the upper end of small batch, a pick and place machine is used with the program developed for the build. That program is then reusable when the product moves to volume, which is one argument for using a machine even when hand placement would be cheaper for the first run.

Through hole parts are a separate decision. Hand soldering is normal at low volume, selective soldering is used where the quantity justifies the programming, and wave soldering rarely makes sense below a few hundred boards because of the setup and the fixture.

Two process details are worth specifying explicitly at this scale. The first is the reflow profile: a small batch is often run in a batch oven with less control than a conveyor, so the profile should be recorded and the same profile reused for repeat builds. The second is cleaning. Where the assembly needs to be clean, it has to be cleaned as part of the process, and a small batch operation may not have an inline cleaner, so this is a question to ask rather than assume.

Inspection and Test at Low Volume

Automated optical inspection is often not economical for a small batch, because the programming effort is comparable to the volume flow while the number of boards is small. The practical alternative is visual inspection under magnification for the visible joints, X-ray sampling of any hidden joints, and electrical test either by functional test or by flying probe.

A flying probe is well matched to small batch work because it needs no fixture. It is slow per board, but at a few hundred units the absence of fixture cost matters more. Where the design has test points and a functional test fixture already exists, running the functional test on every board is usually the most valuable use of the time, because it is the only test that proves the product works.

Two habits make low volume inspection effective. The first is a first article check on the very first board of every build, signed off before the rest are populated, since a systematic error at this stage is cheap to fix. The second is recording the process parameters used, so that a repeat build months later starts from a known profile rather than from guesswork.

Moving From Small Batch to Volume

The transition is easier when it is planned, and the decisions that matter are usually made early.

Keep the panel design stable. Changing the panelisation between the pilot and the production build invalidates the stencil, the fixture and the machine program, which are three of the four setup costs. Where the panel has to change, change it once and do it before the pilot.

Keep the component packages stable as well. A pilot built with hand placement can tolerate packages that a machine cannot feed reliably, and a part that was chosen for availability at low volume may be a reel only item with a large minimum order at volume. Checking the production availability of every part, and not only of the ones that look critical, avoids a redesign at the worst moment.

Finally, decide which test method scales. A functional test developed during the pilot is an asset that carries into production, while an inspection habit that depends on a particular person does not. Where the volume build will need in-circuit coverage, the test points should be designed in from the start even if the pilot does not use them, because adding them later costs a layout revision and a new fabrication round.

FAQ

• What is the minimum order for assembly? Many assemblers accept a single board, though the setup cost makes the unit price high. The practical minimum for a meaningful price is usually a few tens of boards.
• Is hand assembly acceptable for a product? For prototypes and small batches, yes, provided the process is documented and the joints are inspected to a defined standard.
• Can I supply my own components? Yes, but then the build is exposed to the accuracy of the kit and the assembler will usually not warrant parts they did not buy.
• How long does a small batch take? Two to three weeks is typical from a complete kit, with the range set mostly by component availability rather than by assembly time.

Summary

Small batch assembly is priced by setup and complicated by components. The placement is quick; the stencil, the program, the profile and the first article inspection are what take the time, which is why reducing that one time work is the most effective way to cut cost at low volume.

Components decide the schedule. Minimum order quantities, long lead time parts and the risk of unauthorised stock make sourcing the real constraint, and the turnkey route moves much of that problem to the assembler. The process itself is chosen for flexibility, with hand placement and batch reflow being entirely appropriate where the design is not dense, and a prototype build being the stage at which the panel, the packages and the test points should be fixed for the volume process that follows. Whether the pilot runs on a bench or a production line, the same discipline applies: control the test, document the profile, and keep the design stable between builds.