PCB on PCB: 5 Methods That Actually Work

What “PCB on PCB” actually means

When an engineer says “PCB on PCB,” they mean a fully-fabricated, populated board sitting on top of another fully-fabricated, populated board, with electrical connections between them. The top board is usually a daughterboard, a module, or a variant that needs to be swappable. The bottom board is the carrier or base. It is not a multilayer PCB. In a multilayer board, the layers are laminated together at the factory and cannot be separated. In PCB-on-PCB, the top and bottom remain independent, mechanically detachable, and often designed by different teams.

The reason the phrase shows up so often is that it solves a real product problem: you have a stable base design and you need to vary something on top (a radio module, a sensor variant, a customer-specific I/O block) without respinning the whole assembly. The global bare-PCB market was estimated at $80.33 billion in 2024, forecast to reach $96.57 billion by 2029 at a 4.87% CAGR, and a meaningful slice of that volume is consumed by stacked designs in mobile, IoT, and modular industrial products.

How to mount a PCB on top of another PCB

The five production-realistic methods, ordered roughly by stacking height (lowest first):

1. LGA / QFN-style reflow (≤ 1 mm gap)

The top PCB has pads on its bottom surface only. The bottom PCB has matching pads. You reflow-solder the top PCB onto the bottom pads. The result is functionally identical to a surface-mount IC: a board-to-board LGA joint.

The reference design most engineers cite is the Telit HE910 cellular module, which reflows directly onto a carrier PCB. The gap between the module and the carrier is less than 1 mm, and the module is reflowed as a single part during the carrier’s own SMT process. This is the lowest-profile option in production, and it is also the least forgiving. The top PCB has to be flat enough that all pads make contact in the same instant during reflow. That flatness is bounded by IPC class 2 bow/twist tolerance on the as-shipped board, and the boards will flex and twist again during the second reflow. Coplanarity is the real failure mode, not solderability.

A useful engineering rule: if the top PCB is more than ~25% of the bottom PCB’s footprint, plan for additional mechanical fastening. Solder pads alone do not survive vibration or drop tests at that size.

2. Castellated edge plating (0 mm edge gap)

A castellation is a plated half-hole on the edge of a board. Solder paste on the bottom board, reflow, and the top board drops into the paste like a surfboard into a slot. No connector, no mating part, no extra cost beyond the PCB itself.

Two practical traps. First, castellations made by routing the board outline through a via tear the plating at the cut edge and are unreliable in thermal cycling. Most board houses can do proper edge plating that leaves the via intact, and you should insist on that. Second, castellations concentrate all mechanical stress at the solder joint along one edge. Reinforce the copper around each castellation with two small rivet vias, and add a flexible glue bead along the unconnected edge. This converts most of the load from the solder into the glue.

Castellations are also how most off-the-shelf modules (Bluetooth, Wi-Fi, GSM) are designed to mount. The 18k-view Stack Exchange thread on PCB-on-PCB stacking is dominated by engineers using or troubleshooting castellations.

3. Board-to-board stacking connectors (3–8 mm gap, swappable)

For designs where the top board must be removable (field service, configuration variants, test fixtures), a connector is the right tool. Common options:

• Hirose DF17 / FX8 series — 0.5 mm and 0.8 mm pitch, stacking heights from 4 mm to 12 mm. Used in the PiCrust stackable add-on for the Raspberry Pi.
• Molex SlimStack and similar — sub-1.5 mm stacking height, micro-pitch.
• TE Connectivity Micro-Match — IDC-style wire-to-board, useful when the boards sit side-by-side or at an angle rather than stacked.
• Standard 2.54 mm DIL SMT socket + through-hole pin header — the cheapest possible stacking connector, hand-solderable, available in single and double row. Minimum practical gap: about 6 mm including pin length.

The connector approach adds cost (the connector itself plus the placement step) but buys repairability, mechanical compliance, and a defined current rating per pin. If your design tolerates a 3 mm gap, the connector is almost always the safer answer than the LGA reflow.

4. Z-axis conductive tape / anisotropic conductive film

3M makes a Z-axis tape that conducts vertically between two boards but is insulating horizontally. You place a cut-to-size piece between the boards, apply heat and pressure, and you have a flex-tolerant, low-profile stack. Telit and a few other module vendors have used this for the module-to-carrier interface in products where the gap must be very small but the joint must survive some flex.

The honest tradeoff: Z-axis tape’s conductivity is limited. It is fine for low-current signals (under 100 mA per bond) but not for power rails. It is also a specialty material that adds procurement complexity.

5. SMT receptacle + through-hole pin header (hybrid, hand-fixable)

A combination the Stack Exchange thread mentions and is worth its own slot: a 2.54 mm DIL SMT socket on the top board, a through-hole pin header that passes through both boards, and a hand-solder joint on the bottom side of the bottom board. The pin header provides mechanical alignment and most of the current-carrying capacity; the SMT socket provides the electrical contact on the top board.

It is ugly. It is hand-solderable. It is field-repairable. For low-volume industrial products, prototypes, and anything that has to be assembled by a contract manufacturer without a pick-and-place fixture for the second board, it is the most reliable option in practice. The pin length has to be specified (GCT and similar suppliers sell custom lengths, but non-standard pins carry an MOQ of about 1,000 pieces).

How low can the stacking height go?

In production: about 0.5 mm with LGA reflow and a well-supported carrier. With castellations, the effective gap is the solder fillet plus board thickness — typically 1.6 mm standard FR-4 plus a thin solder joint. With a low-profile connector (Hirose FX8, Molex SlimStack), expect 3–4 mm. With a standard 2.54 mm header, plan on 6–10 mm.

The board thickness itself is one of the variables you can tune. Most production boards are 1.6 mm FR-4. The thinnest standard is 0.8 mm (used on LilyPad and Arduino Pro Micro). Anything below 0.8 mm is a flex or rigid-flex design and stops behaving like a normal FR-4 panel in the pick-and-place machine.

Is stacking actually the right answer?

Not always. A 4-layer HDI microvia PCB is equivalent in routing capacity to an 8-layer through-hole PCB, so the argument “we need to stack to get the density” usually fails on inspection. Stack only when the top board genuinely needs to be removable, swappable, or sourced from a different vendor. If it does not, consolidating into a single panel saves the cost of a second pick-and-place cycle, a second reflow, and a connector or mounting hardware.

The other failure mode worth naming: when the top and bottom boards are designed by different teams or sourced from different fabs, the mechanical alignment between them is the design constraint, not the electrical one. The connector pitch, the pad size, the keepout around the mounting area — these are the variables that decide whether the assembly works the first time. If you cannot lock those down early, every method on this list will fail in production.

Frequently Asked Questions

What does PCB on PCB mean?

PCB on PCB describes a fully-fabricated, populated board mounted on top of another, with electrical connections between them. It is distinct from a multilayer PCB, whose layers are laminated at the factory and cannot be separated.

What are castellations on a PCB?

Castellations are plated half-holes on the edge of a board, used to solder one board directly onto another. They are the most common off-the-shelf PCB-on-PCB mounting method and the standard interface for most wireless modules.

Can you solder one PCB directly onto another?

Yes, using either castellations or LGA-style pad-on-pad reflow. Castellations are lower precision and lower cost; LGA reflow gives the lowest stacking height but requires tight coplanarity control.

What is the minimum stacking height for PCB on PCB?

About 0.5 mm in production with LGA reflow, about 1.6 mm with castellations (limited by the standard 1.6 mm FR-4 board thickness), and 3–4 mm with a low-profile board-to-board connector.

Is PCB-on-PCB stacking reliable for production?

Yes, with caveats. Castellations are reliable when the edge plating is done by the board house, not by routing through a via. LGA reflow is reliable when the top board’s bow/twist stays within IPC class 2 and a second reflow is acceptable for the components on the top board.

PCB on PCB vs. a single multilayer PCB — which is cheaper?

Single multilayer is cheaper for high volume, because it eliminates the connector, the second pick-and-place step, and the second reflow. Stacking wins when the top board must be swappable, sourced separately, or upgraded in the field.

Can you stack more than two PCBs?

Yes, but each additional layer compounds the coplanarity and mechanical-alignment problem. Three-board stacks are common in sensor-fusion modules and PoE-powered industrial gateways. Beyond three, consolidate into a single panel or use a rigid-flex design.

Final Thoughts

PCB-on-PCB mounting is one of those problems that has five working answers, and the wrong one is whichever you pick without checking the constraint that actually matters for your design. The most common mistake is choosing LGA reflow for a swappable module — you will spend the next three weeks debugging a coplanarity issue that a $0.30 connector would have solved. The second most common mistake is choosing a connector for a permanent joint — you will pay for a part you do not need and add a placement step that did not have to exist.

If you are early in the design, the cheapest first move is to draw the mechanical envelope of the top board at full scale on paper and place connector pads, mounting holes, or castellation pads before the schematic is finished. The electrical design will follow the mechanical envelope; the reverse is rarely true. If you are already in the middle of a design and the top board is committed, work backward from the connector pitch and the reflow profile to figure out which of the five methods is still on the table. The constraint that eliminates options first is almost always a mechanical one — height, keepout, or coplanarity — not an electrical one.

Related Guides

• Castellated PCB Design: When Edge Plating Works and When It Doesn’t: a deeper look at the most common PCB-on-PCB method, including the manufacturer specifications that separate reliable edge plating from plated-via-butchered-by-router.
• HDI PCB Stack-Up: 4-Layer Microvia vs 8-Layer Through-Hole: the engineering case for consolidating stacked boards into a single high-density-interconnect panel when the top board does not need to be swappable.
• SMT Reflow Profile for Stacked Assemblies: the thermal budget for face-to-face reflow, including how many times each component can survive a second pass.