PCB to PCB Soldering Methods

In short: Boards are joined either by soldering them directly (castellated half-holes, pin soldering) or through a mating connector (edge connector, board-to-board header), and the choice trades cost and height against rework, alignment accuracy and mechanical strength.

Almost every real product has more than one board: a radio module on a carrier, a display on a mainboard, a daughter card on a backplane. The way two boards are joined decides assembly cost, how much vibration the product survives, and whether a failed module can be replaced. This article compares the four most common methods and the physics behind their trade-offs. If you need a refresher on boards themselves, start with what a PCB is.

The four joining methods at a glance

All four methods do the same job, which is to carry signals and power across a gap while holding two rigid boards in a fixed relationship. They differ in whether the joint is permanent (solder) or separable (connector), and in which part of the assembly carries the mechanical load.

Key fact: A soldered joint is the most reliable electrically, but it is the only one of the four that cannot be undone without a hot-air station or an iron and wick.

Castellated holes: the module-on-a-board approach

A castellated board has plated through-holes placed exactly on the board outline, then routed in half. What remains are semicircular copper-lined notches along the edge, called half-holes or castellations. The module is then reflowed onto matching pads on the carrier, and solder wicks up the half-barrel to form a visible fillet. This is how most Wi-Fi, BLE and cellular modules are built, and the same layers and vias covered in via types in PCB apply to the half-hole barrel.

Because the module has no connector body, the stack height is only the board thickness plus a paste layer, often under 2 mm. The cost is that routing the holes in half leaves burrs and copper that tears if the router is dull, so fabricators need a controlled-depth or two-pass profile. A common design rule is a hole diameter of 0.6 to 1.0 mm, a pad extension of at least 0.8 mm outward and a pitch no tighter than 1.27 mm for reliable paste printing.

Edge connectors: gold fingers and card guides

An edge connector uses exposed copper pads, the fingers, on the board edge that slide into a spring-contact socket. Fingers are finished with hard gold, typically 0.76 to 1.27 micrometres over nickel, because soft immersion gold wears off after a few dozen insertions. The board edge is also chamfered at 20 to 45 degrees so it guides into the socket without catching.

The weakness is mechanical tolerance. Board thickness varies, 1.6 mm plus or minus 10 percent is normal, and the socket contact force must still hold in the worst case. A thin or warped board loses contact first. Designers keep a copper-free keep-out behind the fingers, add a keying notch so the card cannot be inserted backwards, and bevel the leading edge. Because only the spring force holds the card, products that vibrate (automotive, industrial) add a retention latch or a screw.

Board-to-board headers: pins and sockets

Board-to-board connectors come as 2.54 mm through-hole pin headers or as fine-pitch surface-mount pairs down to 0.4 mm. The male and female housings carry guide features that give the coarse alignment, and the contacts float slightly so a few tenths of a millimetre of misalignment do not stress the solder joints. Stack height is set by the housing, so a 5 mm, 8 mm or 12 mm part is chosen to clear tall components.

Alignment is the main trade-off. Two connectors placed on two different boards each carry their own placement tolerance, and the sum of them must stay within the float range. A pair of floating connectors with 0.5 mm float tolerates a stack of two typical placement errors of 0.1 mm plus board-edge variation. A rigid connector pair does not, and one pin ends up pre-loaded in the solder, which cracks during thermal cycling. Mounting two or more connectors on the same board pair is where designers most often get caught, so use one connector plus standoffs, or choose floating parts.

Key fact: If two or more connectors must mate at once, tolerance stack-up, not electrical performance, usually decides which part you can use.

Pin soldering: the oldest and strongest joint

Pin soldering means a row of straight pins is inserted into plated through-holes on both boards and soldered by wave, selective solder, or hand. The plated barrel grips the pin, and a joint with good barrel fill, at least 75 percent under IPC-A-610 Class 2, resists both pull-out and shear far better than a surface pad. The pins also act as a spacer.

The cost is process complexity. A double-sided reflow board cannot go through a wave without masking, so the pins are usually soldered in a second operation, and the heat can reflow nearby SMT parts. Hand soldering a 40-pin row takes about 90 seconds per board and introduces operator variation. For anything over a few thousand units, selective soldering or pin-in-paste is cheaper, and solder wetting is checked with X-ray. A thermally heavy ground pin needs a thermal relief, otherwise the iron cannot wet it, which links to PCB thermal management.

Reflow in motion: how paste self-aligns a castellated module

One reason castellated modules work is surface tension. When paste melts at about 217 to 245 degrees Celsius for SAC305 solder, the liquid pulls the module toward the position that minimises solder surface area. A placement error of up to roughly a third of the pad width is pulled back into line, so the module centres itself on the carrier. Too much paste makes the module float and rotate, and too little leaves open joints, which is why the stencil aperture for edge pads is often reduced slightly.

Choosing a method: signal, mechanics and rework

• High-speed signals. Connectors add a discontinuity of a few tenths of a nanosecond; a castellated or soldered joint has the smallest stub. For multi-gigabit lanes, check signal integrity and pin assignment, since adjacent pins create crosstalk.
• High current. A 2.54 mm pin handles about 3 A. Castellated pads are limited by the half-hole copper plating (about 25 micrometres), so power rails use several parallel pads.
• Vibration and shock. Soldered methods win; add underfill or adhesive on the module corners for extra security.
• Serviceability. Connectors win. If a module fails in the field, swap it in seconds.
• Manufacturability. Fine-pitch connectors need tight coplanarity and good paste printing; see DFM for PCB for the rules.

Key fact: In design reviews, ask three questions in order: must it be replaceable, how much height do we have, and how many mating points need to align at once.

Cadence Allegro, Siemens Xpedition and Altium all let you define a mechanical keep-out and 3D connector models, so you can check stack height and collision before fabrication. Teams hiring for hardware roles, such as those placed through ChipXpert training, often ask exactly these trade-offs in interviews.

Want to design reliable multi-board systems? Explore the hands-on PCB tracks in the CourseTron courses.

FAQ

Can I hand-solder a castellated module?

Yes, with a fine tip and flux, but heat the pad and half-hole together. Hot air with a little paste gives better and more consistent fillets.

Why do edge-connector fingers use hard gold?

Hard gold, alloyed with cobalt or nickel, resists wear from repeated insertion, while soft immersion gold is too thin and rubs off.

Which joint is best for vibration?

Soldered through-hole pins or castellated joints with adhesive. Spring-contact connectors can fret unless they are latched or screwed.

What is tolerance stack-up in board-to-board connectors?

It is the sum of placement and board-edge errors on both boards. If it exceeds the connector float, the contacts are stressed.