Most military PCB copy requests we see are sustainment jobs, not new-product jobs. A board on a radar cabinet, a power supply in a ground support unit, or a control card in a legacy vehicle system has failed, the OEM is gone or the program office no longer has manufacturing data, and someone needs to build ten more. That means recovering the layout, the netlist, and the bill of materials from a physical sample, then sorting out which parts are still buyable.
This page covers what that work involves in practice, what makes defense boards harder than commercial ones, and what we need from you to quote it.
Why defense boards get copied in the first place
The reasons cluster tightly, and they’re almost always about keeping fielded hardware alive:
- Diminishing manufacturing sources. A single obsolete logic family or a discontinued analog part kills a whole build. You need the design data before you can approve a substitute.
- Lost data packages. A subcontractor closed, the Gerbers were on a drive nobody imaged, or the delivered package is a set of PDFs with no netlist behind them.
- Long-life platforms. Airframes and vehicles that stay in service for decades outlive the electronics inside them by two or three generations of component technology.
- Repair depot support. A depot needs schematics for troubleshooting, not necessarily new boards. Netlist and annotated schematic only, no fabrication.
- Failure investigation. Boards that fail in the field get analyzed against the as-built layout, which nobody has anymore.
Note what isn’t on that list: cloning somebody else’s fielded product. We work on boards where the customer holds the design rights or has documented authority from the rights holder, projects run under NDA, and export-controlled or classified hardware is outside what we handle — that has to stay with a cleared domestic facility.
What makes a military board harder than a commercial one
Conformal coating and potting
This is the single biggest difference. Nearly every board that has seen a real environment is coated — acrylic, urethane, silicone, or in the worst case a hard epoxy pot. Coating hides silkscreen, buries part markings, and prevents any probing until it’s off.
Acrylic coatings come off with solvent and patience. Urethane is slower. Silicone peels but tears traces if you rush. Epoxy potting is the hard case: it often has to be machined and thermally removed layer by layer, and you accept that some components will not survive. If you only have one sample of a potted assembly, say so up front, because the sequencing changes completely.
Thicker stackups and mixed materials
Defense boards run heavy. Six to fourteen layers is normal, often with thick copper for power distribution, high-Tg laminate, and sometimes a PTFE or hydrocarbon RF section on the same panel. Multiple dielectric materials mean the delamination and imaging pass takes longer and each inner layer has to be scanned separately and reconciled against the drilled via pattern.
Mil-spec and screened parts
You’ll see 883-screened ICs, ceramic packages, gold-plated leads, hermetic cans, and MIL-PRF-39014 or -55365 passives. Two consequences. First, part identification is slower, because a JAN slash-sheet marking or a house part number tells you less at a glance than a commercial marking does. Second, procurement drives the schedule more than engineering does — a substitution study for one obsolete hybrid can take longer than the entire layout recovery.
Documentation expectations
Commercial customers usually want Gerbers and a BOM. Defense customers frequently want a package they can hand to a quality organization: fabrication drawing with stackup and impedance callouts, drill chart, class of workmanship called out against IPC-A-600 and IPC-6012 (Class 3 or 3/A is common for this hardware), assembly drawing, and a netlist that can be loaded into a bare-board tester. State which of those you need at quote time, because it materially changes the effort.
How the work actually proceeds
Order matters, and this is roughly what the bench sequence looks like on a coated multilayer defense board.
- Intake and documentation. High-resolution imaging of both sides as received, dimensional measurement, mechanical outline, mounting hole and connector positions. Nothing is touched before it’s photographed.
- X-ray. Before any coating removal, we shoot the assembly to map internal via structures, spot buried and blind vias, and image BGA ball arrays. On a board with plated-through and buried vias mixed, this is what tells us how many separate imaging passes the inner layers need.
- Coating removal. Solvent, thermal, or mechanical depending on the chemistry, working in small areas and re-imaging as markings become legible.
- Component identification. Every part read, cross-referenced, and where the marking is a house number or fully obscured, characterized electrically — capacitance, ESR, forward voltage, transistor gain, logic behavior on a fixture.
- Desoldering and bare-board scan. Parts come off with the pad locations recorded. Outer layers get scanned at high resolution.
- Layer separation and inner-layer imaging. Controlled delamination, each inner layer scanned, plane cutouts and thermal relief patterns captured — not just signal routing.
- Netlist reconstruction and cross-check. The recovered artwork is used to build a netlist, which is then verified point to point against a second untouched sample where one exists. This catch step matters: a missed inner-layer connection under a BGA is invisible in the artwork but obvious in a continuity comparison.
- Schematic capture. The netlist is organized into readable functional blocks — power, clocking, analog front end, digital core, I/O — rather than dumped as a flat net list. This is what depot technicians actually use.
- Output generation. Gerber or ODB++, drill files, stackup and impedance targets, BOM with reference designators, pick-and-place if assembly is in scope.
If you want the general mechanics without the defense-specific parts, the walkthrough of how a physical board becomes manufacturing data covers the same pipeline on simpler hardware.
Which deliverable set you actually need
People often over-order or under-order. Here’s how the tiers map to typical defense use cases.
| What you need to do | Deliverables that cover it | Usually skip |
|---|---|---|
| Depot troubleshooting and repair | Annotated schematic, BOM, netlist | Gerbers, fab drawing |
| Rebuild identical boards | Gerbers, drill, stackup, BOM, fab drawing, pick-and-place | Full hierarchical schematic |
| Obsolescence substitution study | Schematic, BOM with alternates, netlist | Layer-by-layer artwork |
| Form-fit-function redesign on current parts | Schematic, netlist, mechanical outline, connector map | Original artwork reproduction |
| Bare-board test program | Netlist, drill, outer-layer artwork | Schematic, BOM |
Worth being precise about terminology in your RFQ too, since the distinction between copying a board and cloning it changes what shows up at handover — one reproduces manufacturing data, the other reproduces working hardware.
What to send us to get a real answer
You can get a feasibility read and a quote from photos alone. Send:
- Clear photos of both sides, whole board in frame, on a dark non-reflective background, shot straight down with even light. Add close-ups of any area with dense fine-pitch parts.
- Board dimensions and thickness, plus layer count if you know it.
- The markings on the main ICs, transcribed as text — coated boards make markings unreadable in photos, so type out what you can read under a loupe.
- Coating type if known, and whether any area is potted.
- How many samples you have, and whether we may destroy one.
- One line on deliverables: schematic only, full fab package, or copy plus assembled boards.
- Quantity you eventually intend to build, if any.
Back from us: a feasibility assessment naming any specific risk areas we can see, a lead time, and a price. If the photos leave a real question open — an unreadable custom hybrid, a suspected buried-via structure — we say so rather than quoting around it.
Honest limits on military PCB copy work
Some cases are harder, and a few sometimes aren’t feasible at all:
- Single sample of a fully potted assembly. Getting the pot off without losing information you can’t recover is genuinely uncertain. Two samples change the picture substantially.
- Custom hybrids and thick-film modules. A sealed hybrid with unmarked die inside is not a board copy problem, it’s a chip-level problem, and the answer is often to redesign that function with discrete current parts.
- Screened parts you can no longer buy. We can identify the part exactly and still be unable to help you source it. The recovered schematic then feeds a substitution study, not a like-for-like rebuild.
- RF sections. Recovering trace geometry on a controlled-impedance stripline is doable; guaranteeing the original electrical performance depends on matching the exact laminate, and some legacy materials are discontinued. Expect a tuning iteration.
- Programmed logic. If the board relies on a programmed CPLD, FPGA, or microcontroller, copying the PCB gets you an empty board. Firmware and configuration recovery is a separate scope with its own feasibility question, entirely dependent on the specific part and its protection state.
- Export-controlled and
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 quoteRelated reading