Aluminum PCB & Metal Core PCB (MCPCB) Manufacturing | Copper Core Options | Thermal Path Engineering

Use copper spreading planes under heat sources and dense thermal-via fields: typical 50–100 vias/cm² with Ø0.30–0.50 mm and 1.0–1.5 mm pitch. For power paths and bus bars, consider heavy copper PCB. Keep sensitive RF/analog areas isolated with thermal slots; for RF PAs, evaluate ceramic PCB modules where conductivity and CTE match are critical.

Interface quality drives TIM performance: maintain local flatness within ±25 μm and Ra ≤3 μm across pad fields. For cost/lead-time trade-offs, see our PCB assembly quote guide.

Aluminum PCB (5052/6061): ~140–160 W/m·K, CTE ~23 ppm/°C; best value for LEDs and moderate-power thermal designs.

Copper core: ~380–400 W/m·K, CTE ~17 ppm/°C; use for extreme flux or compact heat sources.

Dielectric: 1–2 W/m·K standard; 3–5 W/m·K advanced (processing adjustments required). Ultra-thin ≤50–75 μm reduces R_th but lowers isolation; we co-optimize thickness vs voltage. For module-to-system integration, see box build.

The three practical forms of a metal core PCB differ by how much current-carrying and routing capability you need above the thermal path.

Single-layer aluminum MCPCB. One copper foil, dielectric, aluminum base. This is the classic aluminum MCPCB used for LED arrays and is the lowest-cost MCPCB that still removes heat through the base. Routing is planar, so crossovers are not possible — the layout must be single-sided by design, not by accident.

2-layer and 2–4 layer metal core PCB. A 2–4 layer metal core PCB (MCPCB) adds inner routing above the thermal core, which is what allows an aluminum-core assembly to carry control electronics on the same board as the power stage. Adding layers does not change the core's job; it changes how much gate-drive, sensing and protection circuitry can sit beside the heat source. The constraint shifts from thermal resistance to dielectric breakdown voltage across the added layers.

Copper core. Copper's ~380–400 W/m·K and lower CTE suit heat fluxes above 5–10 W/cm² and compact sources where aluminum's spreading limit binds. The cost is mass and machining; the benefit is that the base itself becomes the heat spreader. See heavy copper PCB when the requirement is copper thickness in the traces rather than the base.

A related family worth understanding is the thermal substrate or insulated metal substrate (IMS): same metal-base concept, described by the dielectric system rather than the core. When a datasheet says IMS, the engineering questions are identical to MCPCB — dielectric thickness, conductivity and bond integrity.

LED lighting: street/automotive lamps target junction-to-sink <1 °C/W using aluminum MCPCB and high-reflectance masks.

Power conversion: copper-core for IGBT/MOSFET modules with heat flux >5–10 W/cm².

Automotive: cycling −40↔+125 °C with traceability and PPAP readiness. For flexible jumpers near hot zones, pair with flex PCB.

The most common request we receive is a metal core PCB for LED lighting, and the design question is almost always the same: how much copper area and how thin a dielectric are needed before junction temperature stops being the limiting factor.

Start from flux, not from board area. A metal-core LED design handling a few watts over a small emitter footprint may see local flux that overwhelms a 1 W/m·K dielectric regardless of how large the panel is, because spreading happens laterally in the copper. Across a panel of MCPCBs the same logic applies. Raising the dielectric to 3–5 W/m·K reduces the temperature drop across the bond line; increasing copper weight reduces the lateral drop. They solve different halves of the problem, which is why "upgrade the dielectric" is not always the answer.

For a high-power LED board, three parameters decide the outcome: dielectric thickness (75–150 µm standard, ≤50 µm high-performance), copper weight on the trace layer, and the density of thermal vias if the design uses them. Target junction-to-sink thermal resistance below 1 °C/W for street and automotive lamps, and verify it by measurement rather than by summing datasheet values.

Optics and solder mask interact with the thermal design. High-reflectance white mask improves light extraction but changes surface emissivity; where the board also radiates, that trade should be evaluated with the luminaire, not in isolation. For designs that must flex around a curved housing, compare flex PCB and rigid-flex PCB options before committing to a rigid metal core.

When you search for an MCPCB manufacturer, the term covers two very different operations: a plant that owns dielectric coating and vacuum lamination lines, and a trading company that passes your metal core PCB file to a third party. The difference shows up in the parameters that actually govern thermal performance, because those parameters are set by process, not by specification.

Four questions separate the two. First, who applies the dielectric? A real metal core PCB supplier controls coating thickness on its own line and can hold it within ±10%; a broker publishes a datasheet value it cannot influence. Second, what happens to a void? Void area is a lamination outcome — vacuum staging profile, peak temperature uniformity and pre-lam surface preparation decide it. Third, can the supplier alter the stackup for your flux? Moving from 100 µm to 75 µm dielectric changes thermal resistance measurably but also changes isolation margin, and only the plant can requalify that trade. Fourth, what evidence ships with the lot? ASTM D5470 thermal resistance, C-scan void maps and Hi-Pot records are produced where the boards are made.

For volume programs, confirm that incoming MCPCB materials are verified by lot (alloy grade, core thickness, dielectric conductivity) rather than by generic certificate. This is the point where FR-4-style purchasing habits cause trouble: a metal core board is only as good as the bond between core and dielectric, and that bond is not visible on a drawing.

Send the stackup you are considering — core alloy, dielectric thickness and target thermal resistance — through the MCPCB quote request and we will come back with the process window and the measured D5470 data for that build, so the thermal decision is made on numbers rather than on a datasheet claim.