Embedded Power Modules in PCB: SiC Packaging Moves On-Board

Since the start of 2026, Silan Microelectronics has been building a patent portfolio around power module miniaturization, low stray inductance and embedded PCB structures. Records from the China National Intellectual Property Administration show that a design patent covering power devices for PCB-embedded power modules was granted in May 2026, alongside several other patents addressing power module and packaging structures. Some of those filings describe shortened power paths and reduced stray inductance as the route to higher power density and better efficiency. Interest in the approach intensified again in August, and the underlying direction is becoming clear: power devices are moving from packages mounted on the board to structures integrated inside it.

Why the Traditional Three-Level Structure Is Under Pressure

Conventional power electronics follow a simple hierarchy. A semiconductor die is mounted in a package, the package is mounted on a substrate, and the substrate is soldered to the main board. Electrical connections between the die and the board pass through bond wires, solder joints or terminals.

That arrangement worked well while switching frequencies stayed moderate. It becomes a problem as silicon carbide and gallium nitride devices push switching speeds higher. Every millimeter of conductor in the power loop carries parasitic inductance, and parasitics that were tolerable at lower frequencies now produce voltage overshoot, ringing and additional switching loss.

Shortening the loop is the most direct remedy. Embedding power devices, ceramic structures and even portions of the metal heat spreading element inside the PCB reduces the distance between the die and the copper it connects to. The logic parallels what has already happened elsewhere in electronics. HDI, any-layer interconnect and advanced packaging all represent the same movement from two-dimensional routing toward three-dimensional integration. In this case, the PCB stops being a platform that merely carries packaged components and starts participating directly in device packaging and thermal management.

Where PCB Manufacturing Meets Advanced Packaging

The significance for the PCB industry goes well beyond the introduction of another specialty board type.

Historically, a PCB manufacturer solved problems in conductor formation, drilling, lamination and surface finish. Once power devices are embedded, the process also has to handle precision cavity machining, device positioning, resin filling, coefficient of thermal expansion matching and interlayer bonding reliability. The manufacturing sequence starts to resemble advanced packaging more than conventional board fabrication.

There are practical consequences. Cavity depth tolerance determines how well the device sits relative to the copper features that connect to it. Resin flow during lamination must fill around the embedded component without leaving voids or shifting it. Layer-to-layer registration after lamination must still support subsequent fine line processing. Each of these steps interacts with the others, which makes design for manufacturability review far more important than it is on a standard multilayer board.

The result is two parallel upgrade paths forming inside the PCB industry. One is driven by AI servers and high speed communications, pushing toward 16 to 78 layer boards, HDI and any-layer structures, mSAP line widths at 0.075 mm and below, and differential impedance control within plus or minus five percent. The other is driven by robotics, automotive electronics and power semiconductors, pushing toward heavy copper, hybrid ceramic structures and embedded power devices. The first path pursues signal density. The second pursues power density. Both ultimately point toward a higher degree of system integration.

Where Embedded Power Boards Will Land First

New energy vehicles are among the most promising applications. On-board chargers, DC-DC converters and traction inverters are migrating toward 800 volt platforms and beyond. As silicon carbide switching speeds increase, low stray inductance and low thermal resistance become decisive rather than desirable. If the power device sits closer to the copper layers and the heat spreading path, there is room to reduce both system volume and losses.

Robot joint drives share the requirement in a different form. A joint actuator has to be small and light while still delivering high torque, which pushes the drive electronics into a confined volume with limited airflow. Embedding the power stage removes a packaging layer and shortens the thermal path to the housing.

Low altitude aircraft electric propulsion follows the same pattern, with the added constraint that every gram matters and reliability expectations are high. AI server power delivery is a fourth case. Rack power density keeps climbing, and the power conversion stages feeding high current rails face the same pressure to shrink while handling more current.

Notably, these systems do not need power integration alone. The control section of a robot joint driver or an inverter still requires high layer count HDI, flexible circuits or rigid-flex construction. High power and high density interconnect are increasingly appearing inside the same electronic assembly, which means a supplier able to handle both is more useful than one specialized in either.

Manufacturing support for these programs spans multilayer fabrication, HDI, heavy copper, flexible and rigid-flex boards, combined with specialty structure processing and the ability to review stackup, cavity design, copper distribution and overall producibility before tooling begins. Where an industrial PCBA requirement follows the board, keeping fabrication and assembly inside one process chain avoids the situation where a cavity or embedded component problem is discovered after the design is frozen.

The Real Barrier Is Thermomechanical Reliability

Embedding a power device is not simply a matter of adding a cavity to an existing process. The die, the ceramic, the copper layers and the resin system each have a different coefficient of thermal expansion, and under repeated power cycling and thermal shock those differences generate interfacial stress. Over thousands of cycles, that stress accumulates at the material boundaries.

Embedding depth, cavity dimensions, resin flow behavior and interlayer registration accuracy then feed back into both subsequent circuit processing and long term device reliability. A cavity that is slightly too deep changes the thermal path. A cavity that is slightly too shallow changes the bond line. Neither defect may appear at outgoing inspection, but both may shorten service life.

This is why embedded power board industrialization depends on stronger upfront engineering review and tighter process control than conventional board work. The relevant capability is not a single specification but a combination: consistent plating across the panel, controlled lamination with predictable resin flow, registration accuracy maintained through multiple lamination cycles, and inspection that can verify internal geometry rather than infer it.

Where a program also includes assembly, connecting board fabrication to SMT and quality management processes such as incoming material inspection, solder paste inspection, automated optical inspection and X-ray gives the design team a continuous record from raw material through finished assembly. For a structure where internal defects are hard to rework, that record has more value than it does on a conventional board.

It is worth being precise about what such capability does and does not represent. A manufacturer able to build high layer count, HDI, heavy copper and rigid-flex boards, and to review embedded structures for producibility, is positioned to support prototype validation and small to medium volume introduction. That is not the same as having established volume production of embedded power modules, and program planning should reflect the difference.

Testing, Rework and Field Service Implications

Embedding a power device inside a board changes the economics of failure in ways that are easy to overlook during design.

Once the stackup has been laminated, the device is no longer accessible. It cannot be desoldered and replaced. That reality forces a different quality philosophy upstream. Die-level verification before embedding becomes mandatory rather than optional, and electrical testing has to occur at intermediate stages of the build so that a defective device is caught before it becomes permanently inaccessible. A board that fails final test for a reason buried three layers down is not a rework candidate, it is scrap.

Inspection strategy has to match. Optical inspection can only verify what is visible on the surface, so X-ray and computed tomography become the primary means of confirming cavity fill, void content, device position and bond integrity. For high value assemblies, the cost of that inspection is far lower than the cost of discovering a systematic void problem after a production run has been completed.

Rework planning belongs in the original design discussion. If a power stage is embedded, the surrounding circuitry should be arranged so that the rest of the assembly can still be serviced. Connectors, control sections and interface components should remain accessible, because a serviceable board with one non-repairable section is far more practical than a fully integrated board that must be discarded entirely.

Field service expectations follow from the same logic. When an embedded power stage fails in the field, the replacement action is the board or the assembly, not the component. For automotive, robotics and industrial customers, that shifts the emphasis toward accelerated life testing before release. Thermal cycling, power cycling, high temperature reverse bias and humidity testing all exist to establish that the embedded structure will survive its intended service life, because there will be no opportunity to correct it afterward.

Suppliers serving this market therefore need electrical test capability at intermediate build stages, not only at final test. That requirement, more than any single process specification, is what separates a manufacturer genuinely prepared for embedded power work from one that can produce a cavity but cannot prove what is inside it.

A Shifting Boundary

The pattern of patent activity around PCB embedded power modules reflects something larger than one company’s product roadmap. Power semiconductor packaging and PCB manufacturing are converging. When high speed signals keep moving toward higher layer counts and finer mSAP lines while power devices simultaneously move inside the board, the value of a PCB supplier is no longer defined only by trace area and layer count.

It extends into device integration, thermal management and power module manufacturing. For engineers planning the next generation of power conversion hardware, the practical implication is that board design, thermal design and packaging decisions can no longer be made in sequence by different teams. They have to be considered together, and the manufacturing partner has to be involved early enough to influence the stackup while it can still be changed.

Boards that combine advanced PCB fabrication capability with heavy copper and embedded structures will not be the majority of production volume. They will, however, define the ceiling of what a power electronic system can achieve in size, weight and efficiency. The companies that build that capability early will be the ones positioned to serve PCB manufacturing programs that today still exist only as patent filings and concept studies.