A heavy copper PCB copy is the same layer-separation and Gerber-rebuild process used on any board, with one change that drives everything: the copper is 3 oz to 20 oz thick instead of 1 oz, so trace geometry, etch compensation, and the stackup itself have to be measured rather than assumed. Get the copper weight wrong and the rebuilt board looks right on screen and fails at rated current. That measurement is the whole job.
Why thick copper changes the copy, not just the fab
On a standard 1 oz board, a trace edge is nearly vertical and the etched width is close to the artwork width. At 6 oz or 10 oz, it isn’t. Thick copper etches with a pronounced trapezoidal cross-section — the top of the trace is narrower than the base, and the etch factor the original fab used got baked into their artwork.
So when we scan a 10 oz power plane and measure a 3.2 mm trace at the top surface, that is not the number that goes into the Gerber. The engineer has to work back to the artwork dimension the original manufacturer drew, using the measured base width, the measured copper thickness, and a realistic etch factor for that weight. Skip that step and every conductor in your rebuilt board comes out undersized.
Spacing has the same problem in reverse. Heavy copper needs wider gaps than fine-line work because the sidewalls slope outward and the solder mask has to climb over a tall conductor. If a board has 0.5 mm nominal spacing between 8 oz busbars, that spacing was chosen deliberately, and it has to survive into the new artwork intact.
What we actually measure on the bench
- Copper thickness per layer, taken from a cross-sectioned coupon cut from a non-critical edge of the board, read under a metallurgical microscope. This is the only reliable way to distinguish 4 oz from 6 oz on an inner layer.
- Trace base width and top width on the widest current-carrying conductors, so the etch compensation is derived from that specific board rather than from a table.
- Dielectric thickness between each layer pair, which on mixed-weight boards varies from layer to layer because the resin has to fill around tall copper.
- Plated hole wall thickness and finished hole diameter, since heavy copper boards often carry higher plating specs to handle current through vias.
- Overall board thickness and warp, measured before anything is delaminated, because thick-copper stackups are frequently thicker and stiffer than the layer count alone would suggest.
How layer separation works on a heavy copper board
Separation on a normal multilayer board is a controlled sequence: strip the mask, image the outer layers, then remove copper and dielectric layer by layer, scanning each one before it’s destroyed. Thick copper resists that sequence in a specific way — the copper takes much longer to remove chemically, and the resin between layers is often unevenly distributed because it flowed around tall features during lamination.
That means the timing on each etch step has to be adjusted as we go, and it means we scan more often. If a 6 oz inner layer takes three times as long to clear as a 1 oz layer, there are three times as many chances to over-etch into the dielectric and lose a feature we haven’t recorded yet. We work from at least two sample boards where possible, so one can be sacrificed aggressively while the other stays intact as a reference for continuity checks.
Mixed-weight stackups are common in power designs — 2 oz signal layers with 6 oz or 10 oz power layers in the middle. Those are handled as separate problems within one board. The signal layers get imaged and vectorized like any other multilayer work; the power layers get the cross-section treatment. Our notes on thick copper board duplication go deeper into the etch-compensation math if you want the detail.
Where thermal features hide
Heavy copper boards usually exist because something gets hot. Copper coins, embedded slugs, plated slots used as busbar mounts, and thermal vias in dense arrays all show up. A copper coin pressed into a cavity will not appear as a normal layer feature during separation — it looks like a void until you cross-section through it. X-ray before separation catches most of these. So does asking the customer what the board drives.
What the layer count and copper weight do to the schedule
Two variables set the effort: how many layers and how thick the copper is. They multiply rather than add.
| Board type | Main difficulty | Relative effort |
|---|---|---|
| 2-layer, 3–4 oz outer only | Etch compensation on outers; minimal separation | Lowest — often a few days |
| 4-layer, 2 oz signal + 6 oz power | Mixed etch rates, uneven dielectric | Moderate |
| 6–8 layer, multiple heavy inners | Long etch cycles, more cross-sections needed | High |
| Heavy copper with coins or embedded slugs | Features invisible to normal imaging; X-ray plus destructive section | Highest |
| Heavy copper + fine-pitch BGA control section | Two incompatible process windows on one board | Highest |
That last row is worth flagging. Modern motor drives and inverters often put a fine-pitch controller on the same board as 10 oz phase copper. The BGA region needs high-resolution imaging and careful net tracing; the power region needs cross-sections and aggressive etching. Same board, two different workflows, and the schedule reflects both. It’s the same conflict that shows up in RF board reverse engineering, where controlled-impedance sections sit next to ordinary digital.
What you get back, and what verification means here
Standard handover is a full Gerber set with drill files, a stackup drawing that states copper weight per layer and dielectric thicknesses, a BOM with the actual markings we read off the components, and — if you asked for it — a schematic and netlist. For heavy copper work we add the cross-section data and the etch compensation applied, because your fab will ask for it.
Verification on a power board is not just a netlist compare. We check that every conductor in the new artwork carries the same current capacity as the original, using the measured copper weight and IPC-2152 as the reference for trace current versus cross-sectional area. If a rebuilt trace comes out even 15% narrow on a 40 A phase leg, that’s a thermal failure waiting for a hot day. Acceptability of the finished article — plating thickness, annular ring, laminate condition — is judged against IPC-A-600 and IPC-6012 class as agreed up front.
If you want prototypes built, first articles get sectioned again after fabrication to confirm the copper came out at the specified weight. Fabs sometimes substitute a lighter foil and make up the difference in plating. On a signal board nobody notices. On a 10 oz busbar layer it changes the temperature rise.
What to send us for a heavy copper PCB copy quote
- Clear photos of both sides, whole board in frame, on a dark background, with even light. Angle one shot low across the surface so thick traces cast a shadow — that alone tells us a lot about copper weight.
- Board dimensions and overall thickness, measured with calipers. Thickness plus layer count is a strong hint at the stackup.
- Copper weight if you know it, or the current rating of the equipment. “Drives a 15 kW three-phase motor” is useful even without a copper spec.
- The main chip markings — controller, gate drivers, any programmable logic. Photograph them close up.
- One line on deliverables: Gerbers only, Gerbers plus BOM, or full schematic recovery as well.
- How many boards you can send. Two is much better than one. Separation is destructive.
Back from us: a feasibility read, a lead time, and a price. Pricing on these boards tracks layer count, copper weight, board area, component count, whether there’s a BGA, and whether firmware is involved. We won’t quote from a description alone — thick copper hides too much.
Honest limits
Some heavy copper boards are harder than others, and a few aren’t worth doing as a straight copy.
Single sample only. We can still work, but the margin for error shrinks and we may recommend a non-destructive X-ray pass first, which adds cost and doesn’t fully replace cross-sectioning.
Potted or conformally coated assemblies. Power boards are often encapsulated. Removing potting without tearing pads off a heavy copper surface is doable but slow, and sometimes the coating has to come off before we can even quote confidently.
Embedded copper coins and press-fit busbars. These reproduce, but the fab window narrows and unit cost goes up. Sometimes the honest answer is a partial re-layout using a different thermal approach rather than a literal copy.
Heat damage. A board that failed thermally may have delaminated locally or lost copper adhesion. Measurements taken through a damaged region aren’t trustworthy, and we’ll say so rather than guess.
Firmware. If the controller is locked and you need the code, that’s a separate feasibility question decided by the specific part and its protection state — not by the board. We’ll assess it independently.
On the legal side: you need to hold the rights to the design, or have authorization from whoever does. Legitimate work here is overwhelmingly maintenance of out-of-production industrial equipment, obsolescence migration, and recovering files a company owns but lost. Every project runs under our standard confidentiality terms, and your boards and files stay yours.
FAQ
Can you tell the copper weight from photos alone?
Not reliably. A low-angle photo shows that copper is thick, and board thickness plus layer count narrows it, but distinguishing 4 oz from 6 oz on an in
Working on a board like this?
Send the chip marking or two photos. You get feasibility, lead time and price within 24 hours, and the check costs nothing.
Get a free quote