Instrumentation PCB copy is the reverse engineering of measurement-grade boards — signal conditioners, ADC front ends, weighing indicators, gas analyzers, flow computers, lab power supplies — into a manufacturable file set. The difference from ordinary board duplication is that the copy has to preserve measurement performance, not just connectivity. Guarding, shielding, ground partitioning and specific component tolerances are part of the design, and a copy that ignores them will boot up fine and read wrong.
If you’re the maintenance or sustaining engineer holding a dead indicator board from a machine that has to run Monday, that’s the part to keep in mind while you read the rest.
Why instrumentation boards need more care than a generic copy
On a motor drive, a 5 mil error in trace placement usually costs you nothing. On a 24-bit sigma-delta front end, it can cost you counts of noise. Precision boards carry design intent that isn’t visible in a netlist:
- Guard traces and guard rings around high-impedance nodes — electrometer inputs, pH amplifiers, photodiode transimpedance stages. A guard driven by a buffered voltage looks like just another net until you trace where it goes.
- Split and stitched grounds. Analog and digital returns joined at exactly one point, often under the ADC. Copy that as a single unbroken plane and you inject digital return current into the reference path.
- Kelvin connections. Four-wire sense on shunts and load cells. The force and sense pads may be adjacent copper that looks redundant.
- Reference placement and thermal symmetry. A voltage reference sitting deliberately away from a regulator, or matched resistors placed side by side in the same thermal gradient.
- Cutouts and slots under isolation barriers, and creepage spacing that follows a safety standard rather than a routing convenience.
- Shield cans, ferrite beads and pi filters on every cable entry, sized for that instrument’s EMC test result.
So the deliverable for an instrumentation board is a copy plus an explanation. When we hand over files, the notes call out which nets are guards, where the ground join is, which resistors are matched pairs, and which parts are tolerance-critical. Without that, the next person who edits the layout will undo the design.
What actually happens to your board in the lab
Before anything is destroyed
The board gets photographed both sides at high resolution, dimensioned, and the assembly is recorded component by component. Anything with a scratched-off or laser-blurred marking gets flagged early rather than at the end. If the board still powers up, we take reference measurements first — supply rails, reference output, oscillator frequency, and where possible a working reading against a known input. That baseline is what the finished copy gets compared to later.
Getting the copper
Outer layers are scanned optically. Inner layers on a four-, six- or eight-layer instrument board come from controlled removal of the outer copper and resin, layer by layer, with each plane imaged before the next one goes. X-ray helps locate vias and buried features and is how we check a BGA or QFN with hidden pads. Copper thickness, dielectric thickness and finished board thickness are measured, not assumed, because on a controlled-impedance sensor bus the stackup is the design.
Turning copper into intent
The netlist gets extracted, then drawn as a schematic organized by function — input protection, front end, reference, ADC, isolation barrier, MCU, power. That functional grouping is the step most low-cost vendors skip, and it’s the step that makes the file usable for repair and redesign later. We annotate the analog signal chain with gains, filter corner frequencies and reference voltages taken from the actual part values on the board.
Verification
New bare boards are fabricated, built, and tested against the original baseline. For instrumentation that means more than continuity: noise floor, gain error, linearity across the input span, drift over a warm-up period, and behavior at the isolation barrier. Bare-board workmanship is judged against IPC-A-600 and fabrication against IPC-6012 class 2 or class 3, depending on what the end application needs. If you’re building to class 3, say so up front, because it changes the fab you go to.
What to send us for a quote on an instrumentation PCB copy
You don’t need to ship anything to get a feasibility answer and a price range. Send this:
- Clear photos of both sides, whole board in frame, on a dark non-reflective background, camera square to the board, no flash glare on the solder mask.
- Close-ups of the main ICs so the markings are readable — MCU, ADC, reference, op-amps, any isolator, any potted or coated module.
- Board dimensions and, if you can measure it, total thickness. If you know the layer count, tell us; if you don’t, we’ll estimate it from edge-plating and via patterns.
- One line on what you actually need: files only, working prototypes, or a small build. And whether firmware is part of the job.
- What the instrument does and what “correct” looks like — full-scale input, output type, accuracy class. This tells us what the verification has to prove.
You’ll get back a feasibility read on the hard parts, a lead-time estimate, and a price range. To go further we need one physical board, ideally two — one for layer separation, one kept intact as the reference. On instrumentation work, having a known-good unit alongside the failed one is worth a lot, because it gives us something to measure against.
Which deliverable set fits which situation
| What you need | Deliverables | Fits |
|---|---|---|
| Rebuild the same board now | Gerbers, drill, stackup, BOM, pick-and-place | Spares for a machine that’s down; original files lost |
| Understand and maintain it | Above plus functional schematic and net annotations | Long-term sustaining, troubleshooting, training |
| Replace obsolete parts | Above plus editable layout and substitution analysis | ADC, reference or MCU end-of-life |
| Board plus behavior | Above plus firmware recovery from the MCU | Source code and hex both gone |
Most instrumentation customers land in row two or three. The pure Gerber-only tier is fine for a power board but thin for a precision analog board, because you can’t calibrate what you don’t understand.
When firmware is part of the problem
Instruments hold their personality in firmware — calibration coefficients, linearization tables, filter constants, unit conversions. Duplicating the hardware without the code gives you a board that measures nothing. If your source and hex are both gone, firmware recovery from the installed MCU is a separate workstream from the board copy, and feasibility depends entirely on the part.
Older 8051-derived and PIC parts with simple lock bits are usually straightforward. STM32 at RDP Level 1 is a different problem than RDP Level 2, which sets an irreversible state that closes the debug interface for good. MSP430 with a blown JTAG fuse depends on whether the BSL is reachable and whether a password is set. AVR security fuses, C8051 flash lock bytes, TI C2000 password locations — each family behaves differently, and the honest answer for some devices is no. We’ll tell you which case you’re in before you commit.
Note too that calibration constants often live in a separate EEPROM rather than in the MCU. On more than a few instruments, reading a 24Cxx or 93Cxx off the board recovers exactly the data that matters and no MCU work is needed at all.
Work like this is for people who own the design or the equipment: recovering firmware you paid for, keeping an out-of-production analyzer alive, migrating off a discontinued ADC, or doing failure analysis. You’re responsible for holding the rights to the design and for complying with applicable IP law, and every project runs under NDA.
Where instrumentation PCB copy gets hard, and where it stops
Straight answers, because the alternative is a surprise halfway through:
- Conformal coating and potting. Common on field instruments. Coating adds cleaning time. Fully potted modules sometimes come apart cleanly and sometimes take the fine-pitch parts with them. We evaluate before quoting.
- Sanded, relabeled or house-numbered parts. Frequent on analog front ends, precisely because the front end is the value. Sometimes we identify a part from pinout, package, decoupling and circuit behavior. Sometimes we can only characterize the function and propose an equivalent — which means a redesign of that block, not a copy.
- Custom ASICs and programmed logic. A proprietary measurement ASIC generally can’t be reproduced. A locked CPLD or FPGA is its own feasibility question, separate from the board.
- Laser-trimmed thin-film resistor networks. Matched networks with tempco specs you can’t reach with discrete parts. The copy needs the original network available or a genuine substitute.
- Long-obsolete converters. If the ADC has been dead for fifteen years, a straight copy just recreates a board you can’t populate. That’s an obsolescence migration, and it’s better to know at the quote stage.
- Damaged evidence. A burned or delaminated area destroys traces. We can often reconstruct from a second unit or from circuit logic, but a single badly burned board with no reference limits how far we can go.
What moves the price: layer count, board size, component count, whether there are BGAs, whether firmware is involved, and which deliverable tier you pick. A two-layer weighing indicator is a modest job. An eight-layer isolated analyzer board with a locked MCU and two house-numbered amplifiers is a project. The same pattern shows up in adjacent fields — the constraints on a medical device board copy or a robotics control board rhyme with these, though the acceptance criteria differ.
FAQ
Can you copy an instrumentation board if it
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
Related reading
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