Escape Routing Rules: Trace Width, Angles and Pad Fanout

Layout has two halves: deciding where the components go, and getting the copper out of them. The second half is where most manufacturability problems are created, because the escape from a pad is the moment when a designer commits to a trace width, an angle, and a relationship with a neighbouring net. Every one of those choices has a process limit behind it, and the limits are stricter than most beginners expect.

Minimum Trace Width and What It Costs

Conventional fabrication processes will build a trace width of 4 mil, or 0.1016 mm, without complaint. Narrower lines are possible in some shops, down to about 3.5 mil or 0.0889 mm, but the capability comes at a price. Etching narrow copper is a chemical process with statistical variation, and the thinner the trace, the larger the proportion of the line that is affected by undercut, over-etch and copper thickness variation. A design that uses the absolute minimum width everywhere will produce a higher scrap rate at the fab, and that cost is passed back to the customer either as a price increase or as a delivery delay.

The practical approach is to use the narrowest width only where it is required, such as inside a fine-pitch BGA escape region, and to widen the trace as soon as the routing allows. A trace that starts narrow and opens up carries less resistance, causes less signal loss and gives the fabricator more margin to work with.

Angles Matter More Than They Look

Copper etching is directional at the microscopic scale, and traces drawn at arbitrary angles are harder to reproduce faithfully than traces that follow a lattice. Routing at 45 degrees and 135 degrees produces consistent results across the panel, and it makes the artwork easier to review. Random-angle routing is a common source of shorts and opens, because the etch resist behaves differently at each orientation.

Right angles and acute angles within the same net should be avoided for a second reason that has nothing to do with etching. A right-angle corner changes the effective width of the transmission line at the corner, which creates a local impedance discontinuity. The discontinuity reflects part of the signal, and the sharp point of the corner behaves as a small radiator. This is why 45 degree routing is not merely a stylistic preference: it is the standard way of avoiding both the manufacturing issue and the reflection.

The best escape routing also considers the reference plane. A trace that runs over a continuous plane at a controlled height has a predictable impedance, and that predictability is only maintained if the geometry stays consistent along the route. Widening a trace abruptly, or routing diagonally over a plane edge, breaks the assumption, and the penalty shows up as the kind of radiated noise described in this article on radiated EMI in switching regulator layout.

Pad Fanout and Actual Pad Shape

Component land patterns are designed as regular shapes. A BGA pad is round, a QFP pad is oblong, and a chip component pad is rectangular. The pads that end up on the finished board are often not regular, because the escape trace is not the only thing added to the copper. Solder mask registration introduces its own variation, and the mask window is typically 0.1 mm larger than the pad on each side, so the aperture shifts relative to the copper.

Consider a 0402 resistor. If the two pads have their escape traces running in diagonally opposed directions, the resulting copper plus the mask offset can produce an asymmetric pad, with a small rectangular extension attached to one corner of the original land. That asymmetry changes the balance of surface tension when the solder is molten, and the component rotates during reflow.

The cure is symmetry. Fan the escape trace out symmetrically about the long axis of the pad and the tendency to rotate is strongly reduced. If the fanout is also symmetric about the short axis, the tendency of the part to drift sideways is reduced as well. The rule generalises: the more the copper around the two terminations of a chip component resembles a mirror image, the more reliably the part self-centres.

Solder Mask, Bridging and Adjacent Pads

Two adjacent pads on the same net should not be joined by a short direct trace between them. The correct sequence is to escape each pad normally and then connect the two traces further away from the pads. A direct link between pads creates a narrow bridge of exposed copper that is easy to short during hand soldering or rework, and it also removes the solder mask dam that would otherwise separate them. This is particularly relevant for fine-pitch connectors and for decoupling capacitor arrays, where pads are close together and rework is likely.

The same reasoning applies to the mask itself. The mask dam between two pads has a minimum width, and the mask must flow and cure without slumping into the gap. Where the dam is marginal, the ink rheology becomes the limiting factor, and that behaviour is described in this discussion of solder mask ink thixotropy.

Connector Escapes and Mechanical Stress

A connector is a mechanical part first and an electrical part second. Traces leaving a connector pad should exit from the centre of the pad and continue outward in the same direction as the pad. Angled escapes, and particularly traces that leave at a tangent and then turn, create a stress concentration at the point where the copper meets the pad. When the mating half is inserted or removed, the board flexes, and the trace tears at exactly that weak point. The failure is invisible during assembly and appears as an intermittent open on a board that has been plugged and unplugged a few hundred times.

Where the connector carries high current, the escape should also be sized for the current and not merely for the pitch. A narrow escape on a power connector becomes a fuse, and a board that survives the initial test can fail in service.

Putting the Rules Into Practice

None of these rules are difficult to apply, but they must be applied in the right order. Set the layer stack and the reference planes first, because they define the impedance and therefore the trace geometry. Place the connectors and the large parts next, since they constrain the escape directions. Only then route the fine-pitch escapes, using the minimum width where it is genuinely needed and widening as soon as the space appears. Check every escape for angle, symmetry and pad relationship before moving on, and review the pad geometries once the copper is complete rather than assuming that the library footprint is what will be built. Designers who want the deeper dimensional rules on hole and pad geometry will find them in this article on PCB pad design standards.

FAQ

Why should traces avoid 90 degree corners? A right angle changes the effective width of the transmission line at the corner, creating an impedance discontinuity that reflects part of the signal. The sharp point also acts as a small radiator. Routing at 45 degrees avoids both effects and is also easier for the fabricator to etch consistently.

How do I stop 0402 and 0201 parts rotating during reflow? Make the copper around the two terminations as symmetric as possible. Fan the escape traces out symmetrically about the long axis of the pad to control rotation, and about the short axis to control drift. Unequal copper produces unequal surface tension, and the part moves toward the heavier side.

Can I connect two adjacent pads on the same net with a short trace between them? It is better not to. Escape each pad normally and join the traces away from the pads. A direct link removes the solder mask dam and creates a bridge of exposed copper that is prone to shorting during hand soldering and rework.