Price on an LED driver board is set by four things: layer count and size, whether the board is potted, how many magnetic components need to be characterized, and whether firmware has to come out of an MCU. A bare single-sided constant-current driver is one of the cheapest jobs in LED driver PCB reverse engineering. A potted outdoor driver with a DALI interface and a locked microcontroller is one of the more involved ones, and takes meaningfully longer. Send photos and we’ll tell you which one you have.
What actually moves the price on an LED driver board
Most LED drivers are electrically simple and mechanically awkward. The schematic is often 40 to 120 components — a bridge rectifier, a PFC or single-stage flyback stage, a controller IC, an optocoupler feedback loop, a current-sense resistor, and an output filter. That part is straightforward. What costs money is everything around it.
- Layer count and copper weight. Low-cost indoor drivers are frequently single-sided FR-4 or CEM-1 with jumper wires. Higher-power and metal-core boards run 2 oz or heavier copper, and aluminum-backed boards have to be handled differently during layer imaging because the substrate isn’t transparent to the usual techniques. Heavy copper also changes how we document trace widths, since the width-to-current relationship in IPC-2152 is what a rebuilt board has to respect if it’s going to run at the same temperature rise.
- Potting and encapsulation. Silicone or polyurethane potting in an IP65/IP67 housing is the single biggest swing factor. Removing it without tearing pads off a thin single-sided board is slow, manual work, and it consumes at least one sample.
- Magnetics. The flyback transformer, PFC choke, and common-mode choke aren’t catalog parts. Getting turns counts, wire gauge, winding order, interleaving, tape layers, core material and gap out of them means unwinding a sacrificial unit. Budget for this. It is the item most cheap vendors quietly skip, and it’s the reason a “copied” driver comes back with the wrong output ripple.
- Component identification. Lighting OEMs de-mark or house-code controller ICs more often than almost any other product category. Identifying an 8-pin SOIC with a three-character house code takes measurement and inference, not a database lookup.
- Creepage, clearance, and safety spacing. A mains-input driver has an isolation barrier that has to be reproduced exactly, not approximately. Documenting the primary-to-secondary slot, the optocoupler footprint, the Y-cap placement, and the actual measured spacings takes deliberate work, and it’s the part of the job that keeps your rebuilt board compliant.
- Deliverable set. Gerbers and a BOM cost less than Gerbers, BOM, netlist, a readable hierarchical schematic, a magnetics specification sheet, and pick-and-place data. Tell us which ones you actually need.
- Firmware. Dimmable, DALI, 0–10 V, and smart drivers usually carry a small MCU. If you need the code, that’s a separate workstream with its own feasibility question.
We’d rather quote your specific board than publish a number that will be wrong for you. What we can promise is that the quote breaks the cost out by line item, so you can drop the parts you don’t need. If you want to see how the same variables play out on other lighting hardware, our LED driver board copy work follows the same cost logic, and the broader picture of what drives reverse engineering cost applies here too.
Two boards that look similar and price very differently
A 24 W indoor downlight driver: single-sided, no potting, non-dimmable, one small EE-core transformer, a well-marked controller in an SOIC-8. Gerbers, BOM, and a schematic on that board is close to the floor of what this work costs.
A 150 W outdoor floodlight driver: two-layer board with 2 oz copper, fully potted in a die-cast housing, PFC front end plus LLC output stage, three custom wound parts, a de-marked controller, and an 8-bit MCU handling a 0–10 V dimming input. Same product category, several times the effort, and the firmware question hasn’t even been answered yet.
How long LED driver PCB reverse engineering takes
Lead time tracks the same variables as cost, but not proportionally. Here’s the honest shape of it.
| Board type | Relative lead time | What sets the pace |
|---|---|---|
| Single- or double-sided non-dimmable driver, no potting | Shortest | Netlist tracing and component ID; a few days is realistic |
| Same board, potted | Add a meaningful block of time | Depotting is hands-on and can’t be rushed without damage |
| 2–4 layer dimmable driver with MCU, no firmware needed | Moderate | Inner-layer imaging plus digital section tracing |
| Driver with custom magnetics fully characterized | Moderate to long | Unwinding and documenting each transformer and choke |
| Driver plus MCU firmware extraction | Longest, and least predictable | Chip family and protection state decide everything |
The board work and the firmware work run in parallel, so a combined job usually isn’t the sum of both. What we won’t do is give you a fixed date before we’ve seen the board. If depotting reveals conformal coating over the pads, or a controller with no recoverable marking, the schedule changes and we’ll tell you the day we find out. Our general turnaround guidance holds for LED drivers, with depotting as the usual reason a job lands at the slower end.
What happens to your board, in order
Intake and non-destructive capture
We photograph and scan both sides at high resolution before touching anything, record board outline, thickness, hole sizes, and copper weight, and log the housing label — input voltage range, output current, output voltage window, and any dimming protocol. That label tells us what the circuit is supposed to do, which makes every later inference faster. If the label says 700 mA constant current, the sense resistor value has to agree with that, and if it doesn’t, we’ve either misread a marking or there’s a programming resistor somewhere we haven’t found yet.
Depotting, if needed
Potted units get opened mechanically and chemically, working from the housing inward. This is where having more than one sample matters. One unit stays intact as a functional reference; another gets sacrificed. Soft silicone usually comes out cleanly. Hard-filled polyurethane bonded to a thin single-sided board is the worst case, because the resin holds the pads better than the pads hold the laminate.
Component identification and value measurement
Every part gets identified by package, marking, and in-circuit or out-of-circuit measurement. Electrolytics get read for capacitance, voltage, and ripple rating. The current-sense resistor gets measured precisely, because it sets output current and a 10% error there is a 10% error in your LED drive. Safety-critical parts — X and Y capacitors, the fusible resistor, the input fuse, the isolation-barrier optocoupler — are called out separately in the BOM with their safety class, not just their electrical value.
Copper and layer recovery
Single-sided boards are imaged directly. Multilayer boards go through controlled layer separation so each layer is captured on its own, then the layers are registered against the drill map. Vias and the primary-to-secondary split are checked against the original board, not assumed. Output copper on high-current drivers gets its width and thickness recorded rather than redrawn to a default rule, and the finished artwork is reviewed against IPC-A-600 acceptability so what we hand you can actually be fabricated.
Magnetics teardown
The transformer comes apart on the bench: core type and material, bobbin, gap, turns per winding, wire gauge and strand count, winding sequence, insulation tape layers, and any shield winding. You get a wind sheet a transformer vendor can quote from. Skipping this is how a rebuilt driver ends up with the wr
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