Custom Pad
Why RF Layouts Depart From Standard Library Pads
A footprint library ships with parametric pads (round, rectangle, rounded-rectangle, oblong) that cover the majority of through-hole and surface-mount parts. Those primitives stop being adequate the moment electrical behavior, not just mechanical mounting, depends on the copper shape. At microwave frequencies a pad is no longer a simple connection point; it is a short length of transmission line and a lumped shunt capacitance. A 0.8 mm square SMA center-pin pad on a 0.508 mm RO4350B substrate adds roughly 0.04 pF of parallel-plate shunt capacitance, enough to drag the launch return loss toward 15 dB by 30 GHz once via and fringing parasitics are added. The custom pad lets the designer trade pad width, taper length, and ground clearance to flatten that discontinuity rather than accept whatever the library happened to provide.
The same freedom solves assembly problems. A power amplifier on a thermal pad needs a large copper land for heat spreading, but a large land tied directly to a plane wicks solder away during reflow and tombstones small passives nearby. A custom pad with a spoked thermal relief keeps the copper area for dissipation while throttling the heat path during soldering. Paste apertures can be split into a window pattern so solder volume drops to 50 to 80 percent of the pad, preventing voiding under the device. None of these adjustments are expressible with a single library primitive, which is why the padstack editor exposes per-layer geometry for copper, solder mask, and paste independently.
Because the shape is defined once and instanced everywhere, a custom pad behaves like any other reusable design object. Editing the master padstack ripples through every placement, the router treats it as a normal connection target, and the verification flow checks it against the same fabricator constraint set as the rest of the board. That discipline is what separates a deliberate custom pad from a one-off manual copper edit that the toolchain cannot track.
Tuning the RF Launch Pad
The most common RF custom pad is the connector or edge launch. The goal is to make the pad capacitance and the line inductance form an effective characteristic impedance near the system value, usually 50 Ω. Designers either narrow the pad to reduce capacitance, lengthen a taper to grade the transition, or pull back the reference plane beneath the pad to lower the parallel-plate capacitance. The relations below give the first-order numbers a layout engineer checks before committing the shape.
Cpad ≈ ε0 × εr × A / h
Discontinuity reactance at the launch:
XC = 1 / (2πf × Cpad)
Effective impedance dip from added shunt C:
Zeff ≈ √(L / (Cline + Cpad))
Annular ring (min copper around drill):
AR = (PadDia − DrillDia) / 2 ≥ 0.05 to 0.15 mm
Where ε0 = 8.854×10−12 F/m, εr = substrate dielectric constant, A = pad area, h = dielectric height, f = frequency, L and Cline = per-section line inductance and capacitance. Example: A = 0.8 mm × 0.8 mm, εr = 3.66, h = 0.508 mm → Cpad ≈ 0.041 pF, giving XC ≈ 130 Ω at 30 GHz.
Custom Pad Types and When to Use Them
| Custom Pad Type | Primary Goal | Typical Geometry | Key Constraint | Where It Appears |
|---|---|---|---|---|
| Tapered launch | Hold 50 Ω into the line | Trapezoid, 0.5 to 2 mm taper | Match Cpad to line | SMA, edge launch, GSG probe |
| Tuned antipad | Cancel via/pad shunt C | Enlarged plane clearance | Keep AR ≥ 0.05 mm | Backplanes, > 10 GHz vias |
| Teardrop | Suppress trace-to-pad reflection | Filleted trace entry | Smooth < 30° angles | High-speed and mmWave traces |
| Thermal relief | Solderable land on a pour | 2 to 4 spokes, 0.3 mm wide | Throttle reflow heat | Power devices, ground tie |
| Windowed paste | Control solder volume | Split paste apertures | 50 to 80% paste coverage | Large thermal/QFN pads |
Frequently Asked Questions
How do you create a custom pad in an EDA padstack editor?
Open the padstack or pad-shape editor and define copper geometry per layer instead of a built-in primitive. In Allegro the Padstack Designer references a flash or DRA shape; Altium offers Rounded Rectangle, Polygon, or imported regions; KiCad 7 builds a custom shape from a primitive plus added rectangles, circles, and polygons. Assign independent copper, mask, and paste geometry, set the antipad and thermal relief on internal planes, then validate against the fabricator rules. The shape is saved as a reusable padstack so all instances stay synchronized.
When should an RF designer use a custom pad instead of a standard library pad?
Use one when a stock round or rectangular land would create an impedance step, a solder defect, or a thermal bottleneck. RF examples include tapering an edge-launch pad so its capacitance matches the 50 Ω line, opening a tuned antipad to offset via and pad shunt capacitance, adding a teardrop to suppress reflections above 20 GHz, and adding a thermal-relief spoke so a power device still solders to a copper pour. If the library pad already meets impedance, manufacturability, and thermal targets, keep it.
What manufacturing rules must a custom pad still satisfy?
The custom shape still obeys fab and assembly rules: a minimum annular ring of about 0.05 to 0.15 mm around any drill, minimum copper-to-copper spacing of roughly 0.075 to 0.1 mm, and solder-mask sliver and registration limits near 0.05 mm expansion. Paste apertures usually run 80 to 90 percent of pad area to control solder volume, and acute angles below 30° should be filleted to avoid acid traps. A design rule check after editing confirms the shape passes the board-house tolerance set.