Troubleshooting Fine Pitch Solder Bridging on an SMT Line

After reflow, the optical inspection keeps flagging the same area of a batch of display driver boards. Under magnification, the cause is visible: a row of solder bridging between adjacent leads of one fine pitch device. The immediate response is to rework the flagged boards and let the line continue, and that response leaves the next board to fail in the same way. The productive response is to stop the batch and re-read the records from the printing, placement and reflow stages in sequence.

Bound the Problem Before Touching the Component

The first step is not to remove the device. It is to establish how many boards are affected and where they sit in the production sequence. Boards are sorted into those that are clean, those that are suspected and those with a confirmed bridge, and the production order and the affected designators are recorded.

In this case the result was informative: the defect was concentrated on one side of one device rather than scattered across the panel or the batch. A random distribution suggests an intermittent cause, such as incoming material or a handling event. A localised distribution suggests something systematic, and the location of the defect points towards it.

The Printing Record

The SPI data for the affected boards was retrieved next. The total paste volume was within limits, which on its own would have exonerated the printer. Two details changed that conclusion. There was a slight offset on the pins adjacent to the bridge, and the paste edge on the device side of those pins was not clean.

That combination places the origin of the defect before reflow rather than in the oven. A deposit that is offset and ragged has less separation between adjacent pads than the stencil aperture intended, and once the solder melts, surface tension pulls the two deposits together.

The line was paused for that designator and the underside of the stencil was inspected. Residual paste was found near the fine pitch apertures. After cleaning and reprinting, the pad outlines were sharp again. The action looks routine, and it is more effective than repairing finished boards, because it removes the source of the defect class rather than its products.

Two process points follow from it. The under-stencil wipe interval is a parameter that has to suit the aperture area of the design: a board with a large number of fine pitch apertures needs a shorter interval than a coarse board. And the paste itself has a working life at room temperature, after which its viscosity and release behaviour change; the printing of a long run should be scheduled against that life.

Placement Confirmation

Restoring the paste does not prove the problem is solved, because placement can produce the same symptom. A dense display driver board carries resistor networks and small capacitors close to the device, and a placement offset of a fraction of a millimetre pushes the paste towards one side of the pads.

The machine records showed coordinates inside their tolerance, which is not the same as being correct. A microscopic check of the first article showed the device landing marginally towards the affected pins. The board fiducials, the vision recognition of the package and the nozzle condition were checked together, and the coordinate was corrected for that position.

A small verification run followed rather than a resumption of the batch. The new first article went through paste inspection, placement confirmation, reflow and microscope examination in that order. The bridges were gone and the device position was stable, so the batch was restarted.

Should the Reflow Profile Be Suspected?

Finding a printing and a placement problem does not eliminate the reflow stage. Solder on fine pitch leads re-wets during reflow, and if the soak, the peak or the conveyor speed is unsuitable, the risk of bridging can be amplified rather than created.

The profile records for this batch were compared with those of a known good batch, and no anomaly was found, so the zones were left alone. The discipline behind that decision matters more than the result. When several parameters are changed at once and the defect disappears, nobody learns which change mattered, and the same defect returns when the other parameters drift.

So the only changes kept were the stencil cleaning and the corrected placement, with the reflow conditions unchanged from the qualified state. Continuous verification boards then passed, which confirmed the direction of the investigation.

Releasing Reworked Boards

Boards that already carried a bridge need a defined route back into the batch. Microscope re-inspection confirms that the bridge is gone and that the joint has re-wetted properly, that no residue or loose material remains between the leads, and that the neighbouring joints were not disturbed by the rework.

The electrical test for that designator is then repeated rather than assumed, because a bridged joint that has been separated can leave a joint with less solder than the process intended. Boards that were reworked, boards held for review and boards that passed should be stored and counted separately, so that a later report from the field can be traced back to a batch.

What the Records Have to Contain

An investigation of this kind depends on three records being available. The inspection system must store its results per board and per position rather than as a batch average, because the average is exactly what concealed the offset here. The stencil must carry a revision identifier so that a cleaning or replacement event can be tied to a point in the production run. And the placement programme must log coordinate changes with a time stamp, so that a correction can be separated from the boards built before it. Without those three records the defect can be described but not located, and the next occurrence starts the investigation from the beginning.

The Checklist Worth Keeping

Five items come out of this case and apply to any fine pitch defect. Separate and count the affected boards before touching a component. Read the SPI data per position rather than as an average. Inspect the stencil underside and respect the cleaning interval. Confirm the placement centre on the first article under a microscope. Change one parameter at a time and run verification boards before resuming the batch.

Our SMT assembly process applies those steps in the order above, and the records that make the sequence possible, including the SPI data and the profile, are held with the batch in our quality management system. Where a defect turns out to originate in the design rather than the process, the same evidence is what supports the change, and the layout side of it is handled in our design review.

FAQ

Does a correct paste volume rule out the printer? No. Volume is an average over the aperture, and an offset edge can bridge two pads while the volume remains inside its limit.

Why not just adjust the profile as well? Because a change made without evidence cannot be evaluated. If several parameters move together, the cause and the cure are both unknown.

How many verification boards are enough? Enough to include a full cycle of printing, placement, reflow and inspection, which in practice means a small panel rather than a single board.