Medical device PCB copy work almost never means “make me a cheaper clone.” In practice it means one of three things: a service board for an infusion pump or patient monitor that nobody manufactures anymore, a design whose Gerbers and schematics left with an engineer or a defunct contract manufacturer, or an obsolete part that forces a redesign of a board still in the field. Those three jobs need different deliverables, and the isolation and creepage rules on medical hardware change how the copy has to be handled.
Why medical boards behave differently under reverse engineering
The thing that makes a patient-connected board distinctive is the isolation barrier. On a two-channel ECG front end or a dialysis pump controller, there is a physical gap between the patient-applied circuitry and everything downstream — no copper, no inner-layer plane, no silkscreen crossing it. Get that gap wrong by a millimeter in the recovered layout and you have destroyed the property the board existed to guarantee.
So the first thing our engineers measure on a medical board is not a trace. It’s the clearance across every barrier: primary to secondary on the supply, patient side to signal side on the front end, and the keep-outs on every inner layer under the isolators and optocouplers. Those get recorded as dimensions in the recovered stackup, not inferred from a scan. IPC-2221 gives the general spacing framework, but the actual number on your board came from the OEM’s own risk assessment, so we copy the measured geometry rather than substituting a table value.
The second difference is the parts. Medical boards are full of components you cannot identify from a top marking: laser-trimmed precision references, matched resistor networks, medical-grade isolation transformers, custom-wound sensor coils, and low-leakage op amps chosen because they leak picoamps rather than nanoamps. A general-purpose part with the same package and the same nominal value will make the board boot and will quietly wreck its accuracy.
Where the extra time goes
- Isolation geometry. Measured, not traced. Every layer checked for a plane that sneaks across the barrier.
- Analog front ends. Guard rings, driven shields, and Kelvin connections have to be reproduced as topology, not just as netlist connectivity.
- Conformal coating and potting. Coating has to come off chemically without lifting pads. Potted subassemblies are their own problem — see the limits section.
- Precision passives. 0.05% resistors and low-drift references need electrical characterization or datasheet cross-referencing, because the marking rarely tells you the grade.
- Multiple board revisions. Field units often carry mixed revs. If you send two boards from two build dates, we’ll tell you they differ rather than blending them.
What to send us, and what comes back
You don’t need to prepare a package. Photos are enough to get a real answer on feasibility and price.
- Clear photos of both sides of the board, whole board in frame, on a dark background. Straight down, not at an angle. Diffuse light, no flash glare on the solder mask.
- Board dimensions — length, width, and thickness if you can measure it. If you know the layer count, say so; if not, we’ll determine it.
- A readable shot of the main chip marking — the MCU, DSP, or ASIC, plus any part where the marking is sanded off, because that tells us early whether firmware is involved.
- One line on what you need: Gerbers only for a rebuild, a schematic for maintenance and training, a full BOM for sourcing, or a working assembled board.
- Any mention of potting, conformal coating, or a shielded can, and roughly how many units you eventually need.
What comes back from us is short: a feasibility statement that names the specific risks on your board, a lead time, and a price. If something on the board is likely to be a problem — an unidentifiable custom transformer, a fully potted power module — we say so in that first reply rather than after you’ve paid a deposit.
Which deliverable set actually fits your situation
| Your situation | What you actually need | What it does not include |
|---|---|---|
| Service department needs replacement boards for legacy equipment | Gerbers, drill file, BOM, assembly data, one verified sample | Schematic-level understanding of how the circuit works |
| Original design files are lost, product still sells | Full schematic plus editable layout in your CAD tool, so your team can maintain it | Firmware, unless the MCU is handled as a separate scope |
| One component went end-of-life | Schematic capture plus a targeted re-layout around the replacement part | A full duplicate of the original board |
| Repair engineering and fault diagnosis | Schematic and netlist; test-point mapping matters more than layout fidelity | Manufacturing-ready fabrication data |
| Controller board is dead and no programmed spares exist | Board data plus firmware readout from the original MCU | Any guarantee before the specific part is evaluated |
When the firmware is the real problem
On most medical boards the copper is the easy half. A patient monitor or a syringe driver runs code, and a bare cloned board with an empty MCU is a paperweight. Two situations come up constantly.
First: you own the product, the source has been lost, and the only surviving copy of the binary is inside the microcontrollers on units in the field. Second: the MCU itself is obsolete and you need the binary in hand before you can plan a port to a current part. Both are firmware recovery jobs, and both start with identifying the exact device and how it was locked.
Feasibility depends entirely on the part. An older 8-bit device holding code behind lock bits is a very different job from an STM32 sitting at RDP Level 2, where the debug interface is permanently disabled by design. An MSP430 with a blown JTAG fuse, a password-protected BSL, a C8051 with its flash lock byte set — each has its own realistic outcome, and we won’t quote a chip we haven’t identified. We describe what’s achievable for your specific device rather than publishing a success rate, and we never claim a part is guaranteed readable. Work like this runs under NDA, and you need to hold the rights to the design or have the owner’s authorization — that’s a condition of taking the job, not fine print.
The limits worth knowing before you ask
Some medical boards are harder than others, and a few cases are genuinely marginal. Here’s where we push back:
Fully potted assemblies. High-voltage sections in defibrillators and some HV supply modules are potted in hard epoxy. Removing it without shearing component leads is slow, and sometimes the module comes out unrecoverable. We’ll try if you can supply a sacrificial unit, but we won’t promise a result from a single sample.
Custom magnetics and sensors. A medical isolation transformer or a custom pressure sensor with a house part number can be characterized electrically, but matching it exactly may require a wound-to-spec build with a magnetics vendor. That’s a sourcing project attached to the copy, not part of it.
Sanded or relabeled parts. Common on boards the OEM wanted to protect. We work back from pinout, package, and surrounding circuit function, and decapsulation is an option, but the answer sometimes lands as “compatible family” rather than a definite part number.
HDI with stacked microvias. Small wearable and implant-adjacent boards use blind and buried vias in dense stackups. X-ray and controlled layer separation handle it, but the timeline grows and it usually takes more than one sample.
Regulatory reality. We reproduce electrical and mechanical fidelity. We do not grant your board an FDA clearance or a CE mark. If a copied board goes back into a regulated product, your quality system decides whether that’s a like-for-like service part or a design change needing new verification. That’s your call, and it’s worth making before the project starts, not after.
How medical work compares to neighboring industries
If you’ve had industrial boards copied before, expect a slower, more measurement-heavy process here. An industrial control board tolerates a resistor swapped for a similar tolerance; a medical front end may not. The closest neighbors in our experience are boards from defense and aerospace programs, where documentation discipline and traceability matter as much as the copper, and precision test and measurement instrumentation, where analog accuracy drives every decision. Motion-heavy medical hardware — surgical positioners, lab automation — overlaps with robotics control board work on the servo and encoder side.
FAQ
Can you copy a medical device PCB if I only have one board and it’s dead?
Usually yes. A dead board still carries its layout, and layer separation doesn’t require a working unit. What one dead board can’t give you is functional verification of the copy, or a firmware readout if the MCU itself is damaged. If you can find a second unit — even a scrapped one — send both.
Will the copied board pass the same electrical safety tests as the original?
It should, if the isolation geometry, materials, and part grades are reproduced faithfully — and that’s exactly what we measure and document. But we don’t test to safety standards, and formal compliance evidence belongs to whoever owns the device. Plan to re-verify through your own quality process.
Do I need the firmware, or just the board?
Depends on whether working spares exist. If you can still program blank MCUs from a binary you hold, you only need board data. If the only copies of the code live inside deployed units, firmware extraction has to be scoped alongside the copy — and feasibility depends on the specific microcontroller and its protection level.
How long does a medical device PCB copy take?
A two-layer sensor interface board is typically a matter of days. A dense multilayer controller with fine-pitch BGAs, conformal coating, and precision analog runs meaningfully longer, because the inner layers are imaged one at a time
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