BGA and QFN Assembly: Controlling Joints You Cannot See

Most of a board is inspectable. A resistor or a fine-pitch lead can be examined from the side, and a joint that has not formed correctly is usually visible to an operator or to an inspection system. Packages with terminations underneath the body remove that possibility, which is why they are treated as a separate problem rather than a variation on the same one.

Assembling a board that carries BGA and QFN devices is therefore less about the placement itself than about controlling the conditions around it: the volume of paste that is printed, the accuracy with which the device is placed, the thermal profile that forms the joints, and the method used to confirm the result.

What Makes These Packages Different

A ball grid array carries its connections as an array of spheres under the body. There is nothing to see after reflow, and the joints cannot be inspected visually; the device’s own weight and the collapse of the balls are part of how the connection is formed. A quad flat no-lead package carries its connections as lands at the edge and underneath, with an exposed thermal pad in the centre, and the solder volume that suits the fine peripheral lands does not necessarily suit the large central pad.

Both types concentrate a great deal of connection into a small area, which is precisely why they are used: they shorten the electrical path, save board space and improve the performance of high-speed and power-hungry designs. The same density is what makes the process window narrow. A small error in paste volume or in the profile that would be harmless on a device with generous leads can produce voids, bridges or incomplete joints here.

Stencil Aperture and the Volume of Paste

Printing determines most of what follows. The quantity of paste is set by the aperture in the stencil, the thickness of the stencil and the way the paste behaves as the stencil is released.

The objective is not to print as much solder as possible but to print the amount that the joint requires. For the fine pitch lands of a QFN, an aperture that is too large invites bridging between adjacent pads, while an aperture that is too small produces a joint with insufficient solder. For the exposed thermal pad, a single large opening tends to trap flux and air and produce voids, so the aperture is normally divided into a pattern of smaller areas and, where the design allows, connected by narrow channels so that the paste can escape during reflow.

These decisions are made from the land pattern and the stencil rather than from the assembly line, which is why the device positions and the package variants should be identified before the stencil is ordered. A stencil that was generated for a board without these devices is not a stencil that can simply be used again.

Placement Accuracy and the First Article

Placement has to be good enough that every sphere or land sits within its own pad. A device placed with a small offset can still form joints when the surface tension pulls it into position, but the margin has been consumed and the peripheral joints become uneven.

The fiducial arrangement matters more here than on a board with coarse devices, because the machine corrects its position from those marks. Where the fiducials are few, or placed so that rotation error cannot be measured, an offset that appears at one corner of the panel becomes a misalignment at the other.

A first article check on a fine-pitch board should confirm more than the device identity: the paste deposit before placement, the position of the device after placement, and the appearance of the joints around the visible periphery after reflow. Those three observations use the information that is available and leave only the hidden joints to the following stage.

Reflow Profile and Thermal Mass

The profile has to be established for the actual assembly. Board thickness, the amount of copper, the presence of a heavy connector or a metal core and the distribution of the devices all change the way heat arrives at the joints, and a profile that suits a thin board may leave the inner rows of a package on a thick one insufficiently heated.

Two properties of the profile matter most for these devices. The ramp has to be controlled, because heating too quickly can drive flux away from the joint before it has done its work and can cause the paste to spatter. The time above liquidus has to be long enough to let the spheres collapse and form a joint but not so long that the flux is consumed and the surfaces oxidise.

Measuring the profile on the real panel, with thermocouples attached to representative positions, is the only way to know where the process window actually is. Particularly for products that will operate in industrial or power environments, this is worth doing during the trial rather than after a field problem.

<img src="https://www.gopcba.com/wp-content/uploads/2026/09/5-2.jpg" alt="X-ray inspection image of BGA solder joints” />

Inspection and What Each Method Proves

Optical inspection confirms what it can see: that the devices are present, correctly oriented and that the visible peripheral joints have formed. It cannot report on joints underneath a package, and it should not be presented as though it could.

X-ray inspection answers the questions that optics cannot. From the image, the presence of bridging between adjacent connections, the existence and approximate size of voids, the alignment of the sphere array and the condition of the thermal pad can be assessed. It does not have to be applied to every board to be useful; the first article, a sample from a batch and any position that has proved difficult are enough to establish whether the process is behaving.

The practical arrangement on a small batch is usually a full optical inspection plus a defined X-ray check on the first article and on the critical positions. That combination produces a record of the process rather than a claim about it.

Information the Assembly House Needs First

Fabrication data, the bill of materials, the coordinates and the process notes are the starting point for any order, and for a board with these packages they should be accompanied by an explicit list of the devices whose process requirements are unusual, together with the intended function of the product. Where an exposed pad is to be connected to a thermal or ground plane, that intention is worth stating, along with the final use of the board if it will operate in a demanding environment.

The operations that follow are the familiar ones: SMT assembly for the placement work, verification through PCBA testing, and the controls that keep a batch repeatable under quality management.

FAQ

Why is a visual inspection not enough for a BGA? Because the connections are underneath the package. Optics can confirm the device and its orientation, but the joints themselves require an X-ray image.

What causes voids under a thermal pad? Usually a paste volume or stencil pattern that traps flux and air, combined with a profile that does not give them time to escape.

Should every board be X-rayed? Not necessarily. A first article and a defined sample, together with any position with a history of difficulty, are usually enough to show that the process is under control.