PCB Short Circuit: Finding and Preventing the Ones That Escape

A short circuit on a finished board is expensive because it usually appears after the material and the assembly time have been spent. The causes are few and well understood, which means most of them can be designed out and the rest can be detected before the product leaves the factory.

Where Shorts Come From

The first source is the assembly process. A solder bridge between two adjacent pads occurs when there is too much paste for the aperture, when the stencil is not clean, when the paste is misplaced, or when the component shifts during reflow and squeezes the deposit sideways. Fine pitch packages are the usual location, because the distance between neighbouring pads is smallest there.

The second source is the board itself. Copper slivers left by an etching problem, insufficient spacing in a dense area, mask defects that expose a hidden connection, or a plated barrel that has not been fully cleared during drilling can all produce a short that is present before any component is placed. A board made to the minimum spacing a supplier can achieve is more exposed to this class of defect, because the process window is narrower, and the same applies to the etching and plating tolerances behind it.

The third source is the design. Two pads that overlap, a plane split that is not closed, a thermal relief boundary with a gap in it, or a copper pour that touches a net it was not intended to connect, all produce a short that is present by construction and will be found at the first electrical test.

Detecting Them

The bare board test, which checks insulation resistance between nets as well as continuity within them, is the first line. It checks continuity and isolation for every net on the board, using either a dedicated fixture or a flying probe, and it catches the manufacturing defects before any component is added. It cannot catch a design error, because the test programme is derived from the same netlist the board was built from; a short that exists in the netlist is considered correct.

That is why the netlist comparison matters. Comparing the netlist generated from the schematic with the connectivity extracted from the layout catches the errors that the electrical test will faithfully reproduce: a net that has been accidentally merged, a component connected to the wrong supply, a misdrawn symbol. It is a cheap check that runs before the data is released, and it is the only one that can find a design level short.

After assembly, the same isolation test is repeated on the populated board, usually as part of in circuit test. Where a short is found during in circuit test, the useful information is where it is and which net pairs are involved, and that information is best obtained by comparing the failing pin with the expected net rather than by probing randomly.

Finding the Location

A short between two nets can be narrowed down by measuring the resistance between them and by observing whether components in the path affect the reading. In many cases the reading is close to zero, which indicates a direct connection, and that suggests either a solder bridge, a copper defect, or an unintended connection in the artwork.

Visual inspection under magnification resolves the most common cases: a bridge between two pads, solder splashed on the mask, a whisker of copper across a gap. Where the short is under a component, the diagnosis is usually made by removing the part or by measuring between accessible test points on each side. The design can make this easier by providing test points on the nets that are most likely to short, and by keeping the routing of those nets separated where the layout allows. Test access is therefore a design decision with a direct effect on how quickly a fault can be found, and it is often the difference between a board that is diagnosed in an hour and one that is set aside.

Preventing Them in the Design

The first prevention is spacing. The gap between pads should be wide enough for the process that will print the paste, and the aperture design, part of the footprint review, should deliver a volume that does not bridge after reflow. Where a package forces a fine pitch, the stencil design and the paste selection are part of the solution, and the assembly partner should be consulted rather than the geometry being chosen from a standard footprint without thought.

The second is the mask. Solder mask between adjacent pads acts as a dam, and a footprint without mask between the pads relies entirely on the paste volume and the process. Where the option exists, mask defined openings that leave a web between pads reduce bridging, and the mask clearance around the pad has to be specified rather than left to a default.

The third is the plane and pour geometry. Closed contours, verified isolation between supplies, and a rule that prevents a pour from touching an unrelated net remove a whole class of errors. The pour should be checked with the design rule check and, where the tool allows, with a connectivity comparison against the schematic.

Prevention in Manufacturing

On the fabrication side, the shop controls the spacing and the etching, so the relevant requirement is that the design stays inside the capability it claims. Where a board is designed at the minimum spacing, the incoming inspection should include a check of that spacing, and the supplier should be able to demonstrate it with a coupon.

On the assembly side, the controls are the stencil condition, the paste volume, the placement accuracy and the reflow profile. Each of these is measurable, and the acceptance criteria should be stated. An assembly process that reports a bridging rate without also reporting the paste volume and the stencil condition is reporting a symptom without its cause, and the correction will be guessed rather than made. Where the assembly partner also sources the components, the same records cover the incoming material as well.

FAQ

Does the bare board test find every short? It finds shorts in the artwork and in the fabrication. It cannot find a short that exists in the netlist, because the test is generated from it.

What is the most common cause on a fine pitch package? Excess paste volume combined with insufficient mask between the pads.

Can a short appear only when the board is warm? Yes. A marginal gap can close as materials expand, which is why thermal testing is part of the reliability programme.

How are shorts under a BGA diagnosed? By measuring between accessible points on the nets, by comparison with a known good board, and where necessary by removing the device.

Summary

PCB short circuits come from three places: the assembly process, the fabrication process and the design. Detection follows the same order, with the bare board test, the populated board test, and the netlist comparison that only the design stage can perform. Prevention is mostly geometry, paste and mask discipline, and it is cheapest when the assembly partner is consulted while the footprint is being chosen rather than after the boards are built.