BGA Soldering: Voids, Reflow Profile and Inspection

BGA soldering is the point at which the package stops being a part you can inspect with your eyes and becomes a set of joints you have to trust. The balls are underneath the body, the pitch is fine, and the process window is narrower than for any leaded package. Getting the solder right is therefore a matter of controlling the paste, the profile and the placement, and then using inspection to confirm what the process did.

Why BGA Joints Are Harder to Control

Three things make a ball grid array different. The joints are hidden, so a defect cannot be seen and often cannot be found electrically until the assembly is in the field. The pitch is fine, so the paste deposit for each ball is small and the aperture that forms it sits at the limit of what a stencil can hold. And the joint is formed largely by the surface tension of molten solder pulling the ball into contact with the pad, which means the process has to give the solder enough heat and enough time to collapse rather than merely to melt.

The consequence is that a BGA process is judged by what happens to the solder before reflow as much as during it. Paste volume at the pad, the flatness of the package and of the board, and the uniformity of the reflow profile across the whole device all contribute, and a weakness in any one of them shows up as an open joint or a void.

Paste Volume and Stencil Design

Paste volume is set by the stencil. The aperture has to deliver enough solder to form the joint and not so much that the ball floats. For area array packages the usual target is the same volume per unit of pad area that a well printed leaded pad receives, which is achieved by keeping the aperture close to the pad size and the stencil thickness adequate to the pitch.

Where the aperture cannot be as large as the pad, a reduction of around ten percent per side is common, and the resulting loss of volume is recovered by choosing a slightly thicker stencil rather than by distorting the aperture. Paste release is the other half of the problem: small apertures hold paste in the walls, so a stencil with a smooth, adequately polished wall and a good area ratio releases more reliably than one that is merely the right size. Where the design places a via inside a pad, paste can wick into the via and leave the joint short, which is why via in pad or plated through is a decision to make before the stencil is cut.

Reflow Profile for a Ball Grid Array

The reflow profile is what makes the balls collapse. The assembly has to reach a temperature above the liquidus of the paste for long enough that every ball in the array is fully molten, which means the profile has to be measured at the coldest point of the device rather than at the hottest. On a large package the centre of the body heats more slowly than the edge, and a profile that looks correct on a thermocouple taped to the corner can leave the centre short of temperature.

The practical approach is to instrument a representative board with thermocouples under the package, at the thermal mass that heats slowest, and to build the profile from those measurements. Ramp rate, soak time and peak temperature are then read where the solder actually is. Because the joints are hidden, this is the only direct evidence that the process reached the required temperature everywhere.

Voiding: Causes and Acceptance

Voids form when flux volatiles are trapped in the molten solder and cannot escape before the joint solidifies. Their size depends on the paste chemistry, the profile and the pad geometry. A paste whose flux volatilises late, a profile that ramps too quickly through the preheat, and a large thermal pad that lets the centre of the joint freeze before the volatiles have left all increase voiding.

Acceptance is where judgement matters. Some voiding under a large thermal pad is normal and harmless. What is not harmless is a single large void concentrated in one region, because the heat path through the joint is reduced exactly where the remaining solder is thinnest, and a power device that relies on that pad will run hotter than the design assumed. A specification written as a single percentage over the whole pad can therefore pass a joint that will fail in service. Where the joint carries heat rather than current, the rule should be written in terms of the largest contiguous void and where it sits, not the total.

<img src="https://www.gopcba.com/wp-content/uploads/2026/08/SMT-PCB-Assembly.jpg.webp" alt="X-ray image of BGA solder joints” />

Inspection and Acceptance

Optical inspection cannot see the joints, so the inspection plan for a ball grid array has to be built around what can be measured. Transmission X-ray imaging shows the solder under the package as a density map, which reveals a missing ball, a bridged pair or a ball that has not collapsed. Two dimensional imaging is fast and adequate for gross defects. Computed tomography, reconstructed from a set of angled views, separates the layers of the package and makes void measurement meaningful, because a void percentage read from a single projection is the sum of every void along the path.

An acceptance specification should say which defects are absolute and which are graded. A missing ball or an open joint is a reject. A void is graded, and the grading should be written in terms that reflect the function of the joint. Where the pad is a thermal path, the largest contiguous void and its position matter more than the total, and the limit should be set so that the remaining solder still spreads heat across the whole pad.

Rework and Repair

Rework of a ball grid array is a controlled process, not a repair by hand. The package is removed with a profile that matches the original reflow, the site is dressed and re-printed with solder paste through a stencil made for that site, and the replacement is placed and reflowed with a profile that accounts for the thermal mass of the site and of any adjacent parts. Where the design has a via in the pad, the site has to be treated so that solder paste does not drain into the via during rework, which is another reason the pad and via arrangement is worth deciding at the design stage.

Two layout decisions make rework possible at all. The first is leaving enough clearance around the package for the nozzle and for the site tooling. The second is keeping the fanout accessible, so that the joint being reworked is not surrounded by parts that will be disturbed by the heat. The routing that leads away from the array is part of the same question, and the options are described in escape routing and fanout rules. The pad geometry that keeps every joint solderable is covered in PCB pad design standards.

FAQ

Why can a BGA joint not be inspected optically? Because the balls sit underneath the package body. Transmission X-ray shows the solder as a density map, and computed tomography separates the layers of the package so that voids can be measured.

Is any void under a ball a defect? No. Some voiding is normal. What matters is a large contiguous void, particularly under a thermal pad, because it reduces the heat path exactly where the solder is thinnest.

What controls paste volume on a fine pitch array? The stencil. The area ratio of the aperture governs paste release, and where the aperture must be smaller than the pad the lost volume is recovered by using a slightly thicker stencil.

Can a ball grid array be reworked by hand? Not reliably. Rework uses a controlled profile to remove the package, a site stencil to re-print solder paste, and a second profile to attach the replacement, with clearance around the package designed in from the start.