Solder Ball Placement in Small-Batch SMT Assembly
Small-batch SMT assembly rarely stops at one reflow. Boards that combine a fine-pitch package with an odd-form module, or that mix a consumer sensor with an industrial interface, often need a second assembly operation after the main build. Solder ball placement gives that operation a flexible, accurate, and relatively inexpensive route, and it is now common well outside the large EMS houses that originally developed it. The process works best when the four steps around it are specified in advance: flux printing, ball placement, inspection, and reflow.
Where the process fits
The technique applies to lands that need individual spheres rather than a pre-formed package. Reballing a BGA after rework, attaching a package to a board that cannot survive a full second reflow, building a package-on-package stack, and populating a cavity or a stiffener area are all typical uses. What they share is that the solder volume and the sphere position must be controlled locally, not by a paste stencil covering the whole board.
Because the equipment is essentially a printer plus a placement head, the process can be inserted into an existing line. Both machines can be converted back to normal printing and placement when needed, which matters for a small-batch shop that cannot dedicate a line to one product.
Flux printing
The first step is flux printing. A paste flux is printed through a stencil whose openings are sized from the land geometry and the sphere diameter. Two squeegees are used in the same pass: a rigid blade spreads a thin, uniform layer of flux across the stencil, then a rubber squeegee pushes it through the openings onto the lands.
Flux printing is what holds each sphere in place before reflow and what allows it to form a proper fillet. Coverage on the land, not stencil thickness, is the controlling variable, and it is normally set with a design-of-experiments study. After printing, the board is examined under a microscope to measure how much of each land is covered and to confirm that the result repeats across the panel.
Stencil care and process control
Stencils for this process are fragile and expensive. Solid debris or stray material can block an opening, and the only safe cleaning method is a filtered air gun. Solvents such as isopropyl alcohol dissolve the polymer that bonds the stencil, so cleaning is done with a lint-free wipe dampened with deionized water, followed by an air dry.
The printing result is also hard to inspect by eye, because paste flux is transparent. A missing or displaced deposit is invisible in normal light. That makes the microscope check a process requirement rather than an optional extra, and it makes the print parameters worth recording so that drift can be recognized from measurement rather than from a defect.
Placement accuracy matters less than flux condition. A sphere dropped onto a land with a clean, tacky flux deposit will self-center during reflow; a sphere placed on a smeared or oxidized deposit will not.
Ball attach stencil design
The ball attach stencil is a separate tool with its own design rules. Its openings are also derived from the sphere diameter and the land size, but for a different reason: the openings must pass whole spheres without smearing them through wet flux.
A two-layer construction solves this. The main body is an electroformed stencil whose aperture walls are smoother than laser-cut or chemically etched walls, so a ball can travel through without catching. The bonded layer builds the tool to a thickness close to the ball diameter. That combination keeps flux off the stencil surface and lets each sphere drop cleanly to its land, where the printed flux holds it until reflow.
Placement and inspection
Once the tool is aligned, spheres are placed across the array. Automated optical inspection follows immediately. The two dominant defects are missing balls and misplaced balls, and both are visible to AOI because a sphere is optically distinct from a bare land.
Alarm limits should be set from the actual array, not from a generic template. A 0.4 mm pitch array with 400 spheres has a different tolerance for a single missing ball than a 1.0 mm array with 40 spheres, because the electrical consequence differs. The inspection program should also record the position of each defect, so that a systematic offset in one region can be traced back to stencil alignment rather than treated as random.
Rework: missing balls and alignment defects
Rework depends on the defect. Boards with missing balls are repaired offline on a semi-automatic reballing station. One head dispenses paste flux onto the bare land and a second places a single sphere; an imaging system with adequate magnification is needed to position both accurately.
Misplaced spheres are a different problem. In most cases the only reliable fix is to clean the residual flux from the board and repeat the print-and-place sequence, because nudging or re-flowing a displaced ball leaves flux residue under the package. Attempting to salvage the board without cleaning produces joints that pass a first inspection and fail later.
Alloy selection and the second reflow
Alloy choice is driven by the thermal sequence. When a second reflow follows, the spheres must not remelt the joints formed earlier in the build. For lead-free products, a common choice is a narrow-range alloy such as SAC105, whose melting point sits slightly above the lead-free paste used elsewhere on the board.
The reflow profile then has to be split. It must be hot enough to wet and collapse the spheres into the lands, and cool enough that the first-level joints remain solid. That window is narrow, so profiling with a bare board and thermocouples at several locations is worth the setup time. After reflow, AOI confirms the final condition of the array, and the results feed back into the print parameters for the next batch.
Because the window between the two melting points is small, the profile should be verified on the actual assembly rather than extrapolated from a similar product.
When the process pays off
Solder ball placement is worth tooling when the batch is too small for a custom package, when a package must be attached to a board that cannot take another full reflow, or when a repair is needed on a board that already carries significant value. For very small quantities, hand placement under a microscope remains viable, but the stencil-based route wins as soon as the array count rises above a few dozen spheres or first-pass yield starts to matter.
Whatever route is chosen, the process rewards discipline in three places: a flat, clean stencil; a measured flux deposit; and a reflow profile verified on the real assembly. Those three controls turn a delicate operation into a repeatable one. More detail on the assembly sequence itself is available in our guide to the PCBA development process, and the sequence rules that apply before this step are covered in placement order and pad positioning. Floor practices that keep yield stable during small-batch runs are discussed under PCB design quality characteristics.
FAQ
Can reballing be done on an already assembled board? Yes, and that is one of the main reasons the process exists. Local heating removes the old spheres, the lands are cleaned and re-fluxed, new spheres are placed, and the joint is reflowed without exposing the rest of the assembly to a full profile.
How many missing balls are acceptable? The limit comes from the electrical function of the array, not from a general rule. Power and ground spheres rarely tolerate any loss, while spare or mechanical spheres may allow one or two. Acceptance criteria should be written down before production starts.
Do the two stencils have to be purchased together? No, but they should be designed together. The flux stencil controls how much material reaches the land, and the ball attach stencil controls how the sphere arrives. Mismatched designs make the process sensitive to small changes in either one.
