Metal Core PCB Reverse Engineering: Cost & Lead Time

Metal core PCB reverse engineering costs more than an equivalent FR-4 board with the same copper geometry, and the reason is physical: you can’t backlight an aluminum-backed board, X-ray contrast through the base metal is poor, and the dielectric has to be measured rather than assumed. A single-layer aluminum LED board is usually a few days. A hybrid multilayer aluminum-backed driver board with a fine-pitch controller is meaningfully longer, and price follows the same split.

What actually moves the price on an aluminum or copper-backed board

Set layer count aside for a moment. On a metal core board, these are the real cost drivers, roughly in order of impact:

• Single-layer IMS versus hybrid stackup. A classic insulated metal substrate is one copper layer, a thin thermally conductive dielectric, and an aluminum base. Nothing is hidden. A hybrid board — a two- or four-layer FR-4 sub-laminate bonded to a metal base — has inner layers you can’t see, and the metal base kills the two cheapest imaging tricks in the business.
• Copper weight. Metal core boards routinely run 2 oz, 3 oz, or heavier. Thick copper etches with sloped walls and rounded edges, so the copper you photograph is wider at the base than the artwork the fab actually drew. Recovering the intended geometry takes measurement and judgment, not just tracing an outline.
• Component count and repetition. A 1.2 m LED bar with 144 identical emitters and 24 identical resistors is cheap per component. A compact motor driver with 60 distinct parts, three package families, and a shunt network is not.
• Mechanical features in the base metal. Countersunk mounting holes, milled slots, chamfered edges, tapped holes, thickness steps, and thermal pad footprints all have to be dimensioned. Some LED modules are essentially a machined part that happens to carry copper.
• Whether the dielectric needs characterizing. Cross-sectioning a coupon to measure dielectric and copper thickness adds lab time. For a 3 W-per-emitter LED array it matters. For a low-power sensor board on aluminum, sometimes it doesn’t.
• Deliverable set. Gerbers plus drill plus a mechanical drawing is the baseline. Add a BOM, add a schematic, add firmware recovery from the onboard MCU, and each is separate engineering effort with its own risk profile.

We quote per board rather than from a rate card, because two boards with the same outline can differ by a factor of several in effort. If you want the general logic behind how reverse engineering is priced before you ship anything, start there and then send photos.

How metal core PCB reverse engineering recovers the copper geometry

Single-layer insulated metal substrate

These are the friendly ones. The board arrives, we photograph it as received, log component references and markings, then desolder everything with hot air — using low-temperature alloy where the parts are worth preserving for functional testing later. Aluminum sinks heat aggressively, so the board sits on a preheater at temperature before anything touches it. Skip that step and you’ll lift pads chasing solder that refuses to melt.

With the board bare, solder mask comes off chemically and the copper is scanned at high resolution against a calibrated reference scale placed in the same frame. Then it’s vectorization: pad shapes, trace centerlines, clearances, thermal reliefs, and the exact geometry of any large copper pour used as a heat spreader.

The engineer is watching for two specific things. First, etch compensation on heavy copper — a 3 oz trace can measure 40 to 60 µm wider at the base than the design intent, and if you copy the measured width straight into Gerber you slowly grow every trace and shrink every gap. Second, deliberate creepage and clearance gaps, particularly between LED strings, between a string and the mounting holes, and between primary-side copper and the aluminum base on boards that see mains voltage. Those gaps are usually a safety decision, not a routing accident, and they get preserved even when they look wasteful.

Then we measure. Total stack thickness with a micrometer at several points across the panel, base metal thickness after copper and dielectric are removed from a corner coupon, copper thickness from a microsection, and dielectric thickness from the same section. Those four numbers are what a fab actually needs to build the board again.

Hybrid multilayer bonded to a metal base

Here the process changes completely. You can’t transmit light through an aluminum or copper base, and X-ray through 1.0–1.6 mm of aluminum plus an internal copper plane gives weak, low-contrast images of the inner layers. So the base metal has to be separated from the laminate stack first — normally by machining the base down on a mill in controlled passes until the bond line is reached, then releasing the last of the metal without damaging the copper layer directly above it.

That step is slow and it is unforgiving. It is also the single biggest reason hybrid metal core boards take longer than plain FR-4 of the same layer count. Getting a second sample board is genuinely valuable here, because the first one absorbs the learning.

After separation, it becomes conventional work: image the outer layer, remove it, image the next, keeping every layer registered to the same drill pattern so vias line up across the stack. The method matches how we handle layer-by-layer separation and imaging on any multilayer board. The metal base just adds a stage at the front and takes away your ability to sanity-check registration by shining a light through the finished stack.

The dielectric layer is where cheap quotes go wrong

On an FR-4 board, “FR-4, Tg 150, 1.6 mm” is usually enough of a material call. On a metal core board the dielectric is the technology, and it is not generic. Thermally conductive prepregs range from roughly 1 W/m·K up through 3 W/m·K and beyond, with different filler loading, different dielectric withstand voltage, and different thickness. Thermal resistance from junction to base scales with thickness and inversely with conductivity, so a nominally identical rebuild on a slightly thicker, slightly less conductive dielectric can run measurably hotter at the same drive current.

A quote that never mentions the dielectric is quoting a two-dimensional artwork copy, not a thermally equivalent board. We call out what we measured, what we inferred, and what the customer needs to confirm — usually the isolation voltage requirement, because that’s the one number the original designer knew and the board itself won’t tell you.

How long a metal core board takes, by class

Lead times below are working days once the sample is physically in the lab, assuming deliverables are Gerber, drill, and a mechanical drawing. Firmware work, schematic capture, and BOM sourcing run partly in parallel but add days.