Rigid-flex PCB reverse engineering costs more and takes longer than a rigid board of the same layer count — usually noticeably so. The reason is the transition zone: adhesive, coverlay, stiffeners, and asymmetric stackups all have to be measured and documented before anyone can build the board again. Layer count still drives most of the price, but on rigid-flex the mechanical construction drives the schedule.
If you’re collecting quotes, the fastest way to get a real number is to send photos and the folded-and-flat dimensions. The rest of this page explains what we do with them.
Why rigid-flex costs more than a rigid board with the same layer count
On a four-layer rigid board, layer separation is a fairly mechanical exercise: grind, image, verify, repeat. On rigid-flex, the copper you care about is sitting on polyimide, bonded with acrylic or modified-acrylic adhesive, and covered by coverlay rather than liquid photoimageable solder mask. That changes both the imaging and the documentation.
Three things add real hours:
- Stackup asymmetry. A rigid-flex build often has two flex cores running continuously through the rigid sections, with additional rigid-only layers added outboard. Recovering which layers are continuous and which stop at the transition is the single most important thing we determine, and it takes cross-sectioning to do honestly.
- The transition zone itself. Where the rigid stops and the flex begins, there’s typically a keep-out for plated features, a coverlay overlap onto the rigid area, and sometimes a bead of adhesive or epoxy fillet. All of that is fabrication data, not electrical data, and it has to be measured with calipers and a microscope, not inferred.
- Bend-region trace geometry. Traces in the flex arm are usually routed perpendicular to the bend axis, hatched instead of solid on planes, and free of vias. If we redraw them as if they were rigid traces, the board fails after a few hundred flex cycles. Preserving that geometry is deliberate work.
The general framework for pricing and scheduling is the same one we use on any board-level reverse engineering project — layer count, board area, component count, whether BGAs are present, and which deliverables you need. Rigid-flex just adds a mechanical documentation layer on top of it.
What actually happens to your board, in order
1. Non-destructive capture first
Before anything gets touched, both sides get scanned at high resolution with the board flat, plus photographed in the folded state so we can record the bend radius and the direction of the fold. Component markings are read and cross-checked. Flex arms get measured for length, width, and thickness at several points, because polyimide plus adhesive plus coverlay stack up differently along the arm than they do in the rigid area.
X-ray comes next on any board with vias in or near the transition. On rigid-flex we’re specifically looking for whether the plated barrels stop at the rigid layers or continue into the flex core, and whether there are blind or buried structures inside the rigid stack.
2. Cross-section
This is the step that separates a usable rigid-flex data package from a guess. A coupon is cut from a rigid section, a second from the transition, and a third from the flex arm. Potted, polished, and photographed under a metallurgical scope, those three sections give dielectric thicknesses, copper weights, adhesive thickness, coverlay thickness, and the exact layer count in each region. IPC-6013 is the acceptability document for flex and rigid-flex construction, and we reference it when describing the recovered build — but the measurements come from your actual board, not from a standard’s nominal table.
3. Layer imaging
Rigid areas get imaged the way any multilayer does: controlled grinding down to each copper layer, scanned, aligned to a common datum. Flex layers are different — the polyimide is thin and the copper is often thinner, so we typically separate the flex core by delaminating rather than grinding, which preserves the coverlay openings and the hatched plane pattern.
4. Netlist and schematic
Every layer image gets vectorized, stitched, and turned into a connectivity model. That model is then verified against the physical board with a flying-probe or point-to-point continuity check — nets we derived from imagery get confirmed on the real hardware. If you’ve ordered a schematic, component identification and functional grouping happen here, and the flex arm shows up as an explicit interconnect with its own net list rather than being hidden inside a copper pour.
5. Output and verification
Gerbers or ODB++, drill files, stackup drawing with the flex construction called out, fabrication notes covering coverlay and stiffener placement, bend-area instructions, and the BOM. If you want a prototype, we build it and run continuity plus a flex cycle check before shipping.
What drives the quote up or down
| Factor | Effect on cost and schedule |
|---|---|
| Total layer count in the rigid section | Biggest single driver. Each additional inner layer adds imaging and alignment time. |
| Number of flex arms | Each arm needs its own cross-section, measurement set, and bend documentation. |
| Stiffeners (FR4, polyimide, steel) | Adds material identification and a separate mechanical drawing. Moderate. |
| Blind or buried vias in the rigid stack | Significant. Requires additional X-ray and section work to map hole spans. |
| Fine-pitch BGA or 0.4 mm pitch parts | Significant. Escape routing on a flex core is dense and unforgiving. |
| Controlled impedance on the flex arm | Adds modeling to match the recovered dielectric measurements. |
| ENIG vs. immersion tin vs. selective gold on flex fingers | Minor to moderate — mostly a finish call in the fab notes. |
| Firmware involved | Priced separately. Depends entirely on the part, not the board. |
| Number of good samples you can send | Sending two or three is cheaper than one, because we can section freely. |
Lead time follows the same shape. A two-layer flex with a single stiffener and a handful of passives is a short job. A ten-layer rigid-flex with two flex arms, buried vias, and a BGA is measured in weeks, not days, and most of the added time is cross-sectioning and verification rather than drafting. If you’re comparing against a plain rigid build, the honest way to think about it is one to two extra weeks on top of what an equivalent multilayer rigid board would take.
What we need from you to quote
Send this and you’ll get feasibility, lead time, and price back — not a request for a phone call:
- Photos of both sides, board laid flat on a dark background, whole board in frame, in focus, no flash glare. Phone camera is fine.
- One photo in the folded/installed state if you have it, so we can see the bend direction and approximate radius.
- Overall dimensions — rigid section outlines and flex arm length and width, in millimeters. Calipers preferred, a ruler in the photo is acceptable.
- The main chip markings, typed out. The full line of text on the largest IC, exactly as printed, including the date code line.
- One line on deliverables: Gerber only, Gerber plus BOM, full schematic, or turnkey through prototype.
If you know the layer count or have any fragment of the original documentation — an old fab drawing, a partial BOM, an assembly print — send it. It won’t reduce the verification work, but it often shortens the guessing.
Where rigid-flex reverse engineering gets hard, and where it sometimes doesn’t work
Being straight about this saves everyone time.
Single-sample projects with heavy potting. If the rigid sections are encapsulated in filled epoxy and you can only supply one board, we may be able to recover it, but the risk of losing the sample during decapsulation is real. We’ll tell you that before we start rather than after.
Adhesiveless vs. adhesive-based flex. Modern adhesiveless laminates are thin enough that the boundary between coverlay and base polyimide is hard to resolve even under a good scope. We can measure total thickness reliably; splitting it into exact sub-layer values sometimes carries a tolerance band rather than a single number, and we’ll say so in the stackup drawing.
Damaged flex arms. A cracked or creased arm often means the copper in that region is broken in ways that corrupt continuity testing. We can still image it, but net verification on those specific traces becomes inference instead of measurement.
Custom or house-marked parts. A relabeled ASIC or a custom power module with a house number cannot always be identified. We document the footprint, pinout, and observed function, and flag it as an open item — we don’t invent a part number to fill a BOM cell.
Firmware in a locked MCU. Board recovery and code recovery are separate problems. Whether a specific part can be read out depends on the family and its protection state — STM32 RDP settings, MSP430 JTAG fuse condition, AVR lock bits — and no honest lab will guarantee an outcome sight-unseen. We evaluate the part and tell you what we think before you commit.
Legitimate use, briefly
Most rigid-flex work that lands here is maintenance-driven: a medical instrument whose flex cable vendor closed, an industrial camera head whose original design files went with a departed contractor, a controller built around a polyimide interconnect that nobody stocks anymore. Obsolescence and repair are the normal reasons.
You need to hold the rights to the design you’re asking us to recover, and to comply with applicable IP law in your jurisdiction. Every project runs under NDA, and your files aren’t shown to anyone outside the assigned engineers.
FAQ
How much does rigid-flex PCB reverse engineering cost?
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