High-Tech PCB Reverse Engineering Serices PCB Clone & IC Unlock

BGA PCB Copy: Escape Routing, Pad Recovery & Rework

Sep 3, 2026  /  PCB COPY

bga pcb copy: BGA chip being removed from a PCB using a hot-air rework station

Why BGA Packages Make PCB Copy Harder

Ball Grid Array packages hide their connections underneath the component body. When you reverse-engineer a standard QFP or through-hole board, every pad and trace is visible under a scanner or microscope. A BGA board is different: the escape routing disappears beneath hundreds—sometimes thousands—of solder balls, and the vias that carry signals to inner layers are often plugged, filled, or capped. That makes BGA PCB copy one of the most technically demanding tasks in board-level reverse engineering.

The challenges cluster around three areas: accurately mapping the pad array and its fanout, reconstructing escape routes through multiple layers, and handling rework when the original board is your only sample. Miss any one of these and the cloned board either won’t assemble or won’t function.

Understanding BGA Pad Arrays and Ball Maps

Exposed BGA pad array on a PCB showing dog-bone via escape routing

Before a single trace can be drawn, you need a correct ball map. The ball map defines the X-Y grid of solder pads on the PCB land pattern, their pitch, pad diameter, solder-mask opening, and—critically—which balls are signal, power, ground, or no-connect.

Pitch Classes and Their Impact

Pitch (mm) Typical Ball Count Minimum Trace / Space Escape Layers Needed
1.27 100–400 6 / 6 mil 1–2
1.00 200–900 5 / 5 mil 2–4
0.80 300–1500 4 / 4 mil 3–6
0.65 400–2000 3.5 / 3.5 mil 4–8
0.50 500–2500+ 3 / 3 mil 6–10+

As pitch shrinks, the number of traces that can escape between adjacent pads drops—often to zero for inner rows. That forces signals onto deeper layers, which is why fine-pitch BGA boards are almost always high-layer-count PCB copies with complex via structures.

BGA Fanout and Escape Routing Explained

Escape routing is the pattern of vias and short trace stubs that move signals from the BGA pad array out to the wider routing channels of the PCB. The most common strategies are:

  • Dog-bone via: A short trace stub from the pad to a via placed in the open space between four adjacent pads. Simple, reliable, but consumes channel space.
  • Via-in-pad (VIP): The via is drilled directly in the center of the pad, then filled and plated over. Saves space, essential for ≤ 0.65 mm pitch, but harder to reverse-engineer because the via is invisible from the surface after filling.
  • Staggered fanout: Outer rows escape first on upper layers; inner rows drop to progressively deeper layers. The “peeling” pattern is the signature of a well-designed BGA escape.

During BGA PCB copy, the engineer must determine which strategy the original designer used—often a mix—and replicate it exactly. Changing the fanout style can shift via positions by fractions of a millimetre, which in turn moves return-current paths and alters impedance.

Interactive Fanout Planner

Use the calculator below to estimate how many routing layers your BGA fanout requires, based on pitch, ball count, and your fabricator’s trace-and-space capability.

PCB design

BGA Fanout and Escape Routing Planner

How many rows escape per layer, and whether you can avoid via-in-pad. Decide this at schematic time, when the package is still a free choice.

mm
mm
mm
mm
mm
mm
Traces between adjacent pads
Layers needed to escape
Gap between pads mm
Via land diameter mm
Rows escaping on the top layer
How this is calculated

The arithmetic is simple: subtract the pad diameter from the pitch to get the copper-free gap, then see how many trace-plus-space channels fit in it. Each channel clears roughly one more row of balls per internal layer, plus the two outer rows that escape on the surface.

The result decides your layer count and often your entire fabrication class. At 0.8 mm pitch you can dog-bone out with standard vias. At 0.5 mm the via land no longer fits between pads and you need via-in-pad, which means filled and capped vias. At 0.4 mm you are building an HDI board with microvias whether that was the plan or not.

Delayering a BGA Board: Layer-by-Layer Recovery

The standard approach to BGA PCB copy involves removing material one layer at a time—chemically or mechanically—and imaging each exposed copper plane. Under a BGA footprint, this process has extra complications:

  1. Filled vias look like pads. After the top copper is etched away, a filled-and-plated via-in-pad appears as a solid copper dot, indistinguishable from a through-hole pad unless you cross-section or continue delayering.
  2. Micro-vias span only one or two layers. A laser-drilled micro-via connecting Layer 1 to Layer 2 will vanish once you remove Layer 1 copper. If you didn’t image it first, the connection is lost.
  3. Ground and power planes under the BGA may be split. Large BGA processors often sit above partitioned power islands. Missing a 0.1 mm anti-pad gap between VCC_CORE and VCC_IO will short two supplies on the clone.

For boards with ten or more layers, the delayering accuracy directly determines project cost and risk—a reality explored in depth for impedance-controlled board reverse engineering where dielectric thickness matters as much as copper geometry.

Recovering Damaged BGA Pads

Because the donor board is often the only sample, it frequently arrives with pads damaged by previous rework attempts, corrosion, or mechanical stress. Pad recovery is not optional—if a pad is missing, the cloned board’s land pattern will be wrong.

Common Pad Damage Scenarios

  • Lifted pad: The copper pad has separated from the substrate. The outline may still be visible under magnification, allowing dimensional recovery, but the via connection beneath it is suspect.
  • Cratered pad: The pad and one or more resin layers have torn away, exposing inner copper. This actually helps reverse engineering—you get a free cross-section—but the board can no longer be used as a functional reference.
  • Solder-bridged pads: Residual solder obscures pad boundaries. Careful solder removal with low-temperature alloy (e.g., ChipQuik) or solder wick is needed before imaging.
  • Thermally darkened pads: Overheated pads change color, making automated optical recognition unreliable. Manual measurement under calibrated magnification is required.

Pad Dimension Recovery Workflow

  1. Remove the BGA component using a controlled hot-air or IR rework station.
  2. Clean residual solder with flux and wick at ≤ 260 °C to avoid further damage.
  3. Image the exposed pad field at ≥ 2400 DPI with calibrated scale reference.
  4. Measure pad diameter, pitch (center-to-center), and solder-mask opening on undamaged pads.
  5. Extrapolate damaged pad positions from the known grid; verify against the component datasheet.
  6. Cross-reference via locations with inner-layer images obtained during delayering.

If the board was originally photographed before disassembly, those images can supply missing geometry. Our guide on reverse engineering a PCB from photos explains what surface images can and cannot tell you.

Rework Considerations for BGA PCB Copy

Rework enters the BGA PCB copy process at two stages: removing the BGA from the donor board, and assembling the BGA onto the cloned board for functional testing.

Removing the BGA From the Donor Board

The goal is to remove the component without destroying either the component or the board underneath. Key parameters:

  • Profile temperature: Follow the component’s MSL (Moisture Sensitivity Level) rating. Most BGAs are MSL-3, meaning they must be baked before reflow if exposed to ambient humidity for more than 168 hours.
  • Bottom-side preheat: Bring the entire board to 150–180 °C before applying top-side heat. This reduces thermal gradient and the risk of pad lift.
  • Nozzle size: The rework nozzle should match the BGA package outline within ±2 mm per side. An oversized nozzle heats neighboring components; an undersized one creates hot spots.
  • Lift timing: Use a vacuum pickup to lift the component the moment all balls reach liquidus. Waiting too long allows intermetallic growth; lifting too early tears pads.

Assembling the BGA on the Cloned Board

Once the clone is fabricated, the BGA must be placed and reflowed. If the original BGA was recovered intact, it can be reballed and reused. Otherwise, a new component is sourced. Either way, the cloned board’s pad geometry must match the original within ±25 µm for fine-pitch devices.

After assembly, functional testing with golden-sample comparison is the definitive check. X-ray inspection catches bridging and head-in-pillow defects, but only a powered test confirms that every signal actually reaches the BGA die.

Via Structures Under BGA: Micro-Via, Blind, Buried

PCB cross-section revealing stacked micro-vias beneath a BGA footprint

Modern BGA designs use a hierarchy of via types, each with different reverse-engineering implications:

Via Type Typical Diameter Spans RE Challenge
Through-hole 0.25–0.35 mm All layers Easy to image; visible on both sides
Blind micro-via 0.075–0.15 mm L1–L2 or L1–L3 Visible only from one side; may be filled
Buried via 0.10–0.20 mm Inner layers only Invisible from surface; found only during delayering
Stacked micro-via 0.075–0.10 mm Multiple sequential spans Each segment must be imaged at its layer pair

A 0.50 mm-pitch BGA on a 12-layer board may use stacked micro-vias for inner-row escape, blind vias for mid-ring rows, and through-hole vias for the outermost ring. Capturing every via type accurately is what separates a successful professional PCB reverse engineering service from a failed attempt.

Signal Integrity Concerns in BGA Escape Zones

BGA escape zones are electrically dense. Traces run close together through narrow channels between via pads, and return-current paths on reference planes are interrupted by via anti-pads. During BGA PCB copy, two signal-integrity issues deserve special attention:

  • Impedance discontinuity at via transitions: A signal dropping from Layer 1 to Layer 4 through a via passes through multiple reference-plane transitions. If the cloned stackup doesn’t match the original dielectric thicknesses, the impedance at these transitions will differ, causing reflections on high-speed nets.
  • Crosstalk in escape channels: Two differential pairs escaping through adjacent channels may couple if spacing is reduced even slightly. The original designer likely tuned spacing to a specific crosstalk budget; the copy must preserve it.

For boards carrying DDR4/5 memory buses, PCIe Gen 3+, or multi-gigabit SerDes, these details are non-negotiable. A cost-reduction redesign that changes the BGA escape strategy to save layers will almost certainly break timing margins.

Industry Examples: Where BGA PCB Copy Is Common

BGA packages dominate in several industries where legacy board replacement is a recurring need:

  • Semiconductor fab equipment: Process-tool control boards often carry large FPGAs in 0.80–1.00 mm BGA packages. When the OEM discontinues the board, semiconductor equipment board reverse engineering becomes the only path to continued production.
  • Industrial automation: CNC controllers, servo drives, and printing-machine control boards increasingly use BGA-packaged SoCs. Copying these boards requires both BGA fanout recovery and careful handling of mixed-signal sections.
  • Telecom and networking: Switch and router line cards with 20+ layers and multiple BGA ASICs represent the extreme end of BGA PCB copy complexity.

Step-by-Step BGA PCB Copy Workflow

  1. Incoming inspection: Photograph the board from all angles. Record BGA package markings, pin-1 orientation, and any visible damage.
  2. Component removal: Desolder the BGA using a profiled rework station. Preserve the component if it will be reused.
  3. Surface imaging: Scan the exposed pad field and surrounding routing at high resolution.
  4. Delayering: Remove copper and dielectric layers sequentially, imaging each. Pay special attention to micro-via landings.
  5. Netlist extraction: Stitch layer images into a 3-D via map. Assign net names by tracing from BGA ball to destination.
  6. Schematic generation: Build the schematic from the extracted netlist; verify against component datasheets.
  7. Layout recreation: Redraw the PCB in CAD, matching pad geometry, escape routing, and stackup. Use the fanout planner above to double-check layer requirements.
  8. Fabrication and assembly: Manufacture the clone, place and reflow the BGA, then run X-ray and functional tests.

Have questions about any step? Our reverse engineering FAQ covers the 30 most common concerns buyers raise before starting a project.

Common Mistakes in BGA PCB Copy

The most expensive mistake is not a wrong trace—it’s a wrong via type. Substituting a through-hole via where the original used a blind micro-via changes the stub length, adds parasitic capacitance, and can violate the board’s impedance budget on every affected net.

  • Ignoring via fill status: A via-in-pad that was filled and plated in the original must be filled and plated in the clone. A simple tented via will leave a void under the BGA ball, causing solder to wick down during reflow.
  • Changing pad-to-via offset: Even a 50 µm shift in the dog-bone stub direction can push a via into an adjacent pad’s anti-pad, creating a short on an inner layer.
  • Misidentifying no-connect balls: Some BGA packages have depopulated balls for thermal or future-use reasons. Placing a pad where none should exist wastes routing space and can create unintended stubs.
  • Skipping X-ray verification: After assembly, X-ray inspection is the only non-destructive way to confirm that every BGA ball has formed a proper joint. Skipping it turns functional testing into a guessing game.

Choosing the Right Partner for BGA PCB Copy

Not every reverse-engineering lab has the equipment or experience for BGA work. Before committing, verify that the provider can demonstrate:

  • Micro-via detection capability (SEM or high-resolution cross-sectioning).
  • Controlled-impedance stackup reconstruction, not just copper tracing.
  • BGA rework stations with profiling and X-ray inspection.
  • Functional test infrastructure to validate the finished clone against a golden sample.

Browse our completed project case studies to see how these capabilities come together on real boards—from four-layer consumer devices to 24-layer telecom backplanes.

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

WhatsApp Send board details