Contact Us
Blog / PCB Stackup for SMT Assembly: Thickness, Copper, Finish — What You Must Specify (and Why)

PCB Stackup for SMT Assembly: Thickness, Copper, Finish — What You Must Specify (and Why)

Posted: April, 2026 Writer: NextPCB Content Team - C Share: NEXTPCB Official youtube NEXTPCB Official Facefook NEXTPCB Official Twitter NEXTPCB Official Instagram NEXTPCB Official Linkedin NEXTPCB Official Tiktok NEXTPCB Official Bksy
  1. Table of Contents
  2. Introduction
  3. What Is a PCB Stackup Specification?
  4. Standard 4-Layer Stackup: The Starting Point
  5. Stackup Parameters That Directly Affect SMT Assembly
  6. Surface Finish: The SMT-Critical Choice
  7. Controlled Impedance: When You Need It and What to Specify
  8. How to Document and Submit Your Stackup Specification
  9. Common Stackup Specification Mistakes
  10. Frequently Asked Questions
  11. Wrapping Up

Introduction

Most engineers treat the stackup as a signal integrity problem. Pick a layer count, set impedances, route the board, and move on. That logic works fine until your boards come back from assembly warped, or your fine-pitch ICs don't solder cleanly, or the factory emails you asking what stackup they should use — three days after you thought the order was confirmed.

The stackup specification connects your electrical design to the physical manufacturing process. It's not just a document for the PCB fab shop. It directly shapes what happens on the SMT line: how the board behaves in a reflow oven, whether the surface finish is compatible with your component pitches, and whether the board stays flat enough for pick-and-place to work reliably.

This guide covers what a stackup specification needs to include for SMT assembly, why each parameter matters, and what to get right before you send files.

What Is a PCB Stackup Specification?

A stackup specification is a document — usually a table or drawing — that defines the physical structure of your PCB layer by layer: the order of copper layers and dielectric layers, the thickness of each, the copper weight, the base material, and any controlled impedance requirements. It also typically includes surface finish selection and board thickness tolerance.

For 2-layer boards without impedance requirements, most manufacturers apply a default stackup, and you don't need to provide one. For any 4-layer or higher board, or any design with controlled impedance traces, you must specify the stackup explicitly. Without it, the manufacturer fabricates with whatever parameters meet your layer count and thickness — which may or may not match what your signal integrity calculations assumed.

The stackup also feeds directly into SMT process decisions:

  • Board thickness determines fixture and conveyor settings on the assembly line
  • Copper weight distribution affects how the board absorbs heat during reflow
  • Laminate Tg determines whether the board can survive multiple reflow passes without delamination
  • Surface finish determines solderability for your component mix
  • Stackup symmetry determines whether the board stays flat through the reflow oven

None of these are purely electrical concerns. They're process constraints, and the stackup is where you communicate them.

Standard 4-Layer Stackup: The Starting Point

For most digital and mixed-signal designs, a 4-layer board with a 1.6mm total thickness is the default choice. The standard layer assignment looks like this:

Layer Name Function Copper Weight
1 F.Cu (Top) Signal / Components 1 oz
Prepreg Dielectric
2 In1.Cu GND plane 0.5 oz
Core Dielectric
3 In2.Cu PWR plane 0.5 oz
Prepreg Dielectric
4 B.Cu (Bottom) Signal / Components 1 oz

NextPCB 1.6mm 4-Layer Standard Stackup Diagram

NextPCB 1.6mm 4-Layer Standard Stackup Diagram

Why this arrangement? Layer 2 as GND provides a continuous return path directly beneath the top signal layer — critical for controlling trace impedance and minimizing loop inductance. Layer 3 as PWR sits adjacent to GND, forming a distributed capacitance that helps with power delivery network (PDN) performance. Both inner planes are at 0.5 oz because they're not carrying high current; keeping them thinner allows more room in the dielectric budget for controlled-impedance prepreg thickness.

For power designs or boards carrying more than 2A on inner planes, inner layer copper should move to 1 oz. Note this in your stackup specification.

Typical Thickness Breakdown (1.6mm total)

Layer Thickness
F.Cu (copper) 35 µm (1 oz)
Prepreg 0.10 mm
In1.Cu (copper) 17.5 µm (0.5 oz)
Core 1.265 mm
In2.Cu (copper) 17.5 µm (0.5 oz)
Prepreg 0.10 mm
B.Cu (copper) 35 µm (1 oz)
Total ~1.6 mm

This is the stackup most manufacturers apply by default for 4-layer FR-4 boards at 1.6mm. If you don't specify, this is probably what you get — which is fine for most designs. The problems start when your design assumes a different prepreg thickness for impedance reasons, or you need a thinner core for controlled coupling between layers, and you never put it in writing.

Stackup Parameters That Directly Affect SMT Assembly

Board Thickness and Conveyor Compatibility

Thickness stops being a purely mechanical parameter once you're on an SMT line. At 0.8mm, boards start to flex on standard conveyor rails — enough to cause pick-and-place misalignment and stress solder joints during cooling. At 2.4mm and above, they stop fitting standard magazine stacking for automated loading, and the extra thermal mass changes what the reflow profile needs to look like. Neither deviation shows up in your schematic, but both show up immediately in production. If your design requires a non-standard thickness, call it out in your stackup spec — not as a footnote in a purchase order.

Copper Weight and Reflow Heat Distribution

Heavy copper behaves like a heat sink during reflow — it pulls heat away from pads and into large copper areas. On 2 oz outer layers, you'll often see pads lag behind the reflow profile: the oven temperature is climbing, but the pad hasn't reached liquidus yet because the surrounding copper is soaking up the heat. The result is cold joints that look fine visually but fail in thermal cycling. If you're running 2 oz outer copper on a power design, document it in the stackup spec and flag it to your assembly house. Most can tune the profile for it — but only if they know to look for it.

Stackup Symmetry and Bow/Twist

IPC-6012 allows up to 0.75% bow and twist for most SMT assemblies. In practice, many assemblers tighten this to 0.5% for BGA-heavy designs — and it's usually where they push back first if your stackup is asymmetric. Out-of-plane warpage causes BGA balls to lose contact with pads before reflow is complete, which is exactly the kind of failure that doesn't show up in visual inspection.

The primary cause of bow and twist is asymmetric stackup — copper layers that are not mirrored about the board's mid-plane. An unbalanced copper distribution creates unequal stress during the lamination and reflow thermal cycles, and the board warps.

Rules for maintaining stackup symmetry:

  • Mirror copper weights about the center layer. If Layer 1 is 1 oz, Layer N (bottom) should also be 1 oz.
  • Mirror dielectric thickness. The prepreg above and below the core should be the same material and thickness.
  • Balance copper fill ratio between top and bottom. If the top layer is dense with signal traces, add copper fill (hatched pour) on the bottom to equalize thermal stress. Aim for ≥30% copper coverage per layer.
  • Avoid stacking all heavy copper on one side. This is the most common asymmetry error in power board designs.

In your stackup specification, include an explicit note confirming that the stack is symmetric, or flag any asymmetric layers and explain why they're necessary.

Laminate Tg and Reflow Cycles

FR-4 laminate comes in standard Tg (~135–140°C) and high-Tg (≥170°C) grades. For lead-free reflow (SAC305, peak temperature around 245–260°C), standard Tg FR-4 can handle the process — the Tg isn't the melting point, and FR-4 is stable well above its Tg at reflow timescales.

What Tg actually governs is how many reflow passes the board can tolerate before the laminate starts to degrade:

  • Standard Tg (135–140°C): Fine for boards going through reflow once or twice. Double-sided SMT means two reflow passes; most standard Tg boards handle this without issue.
  • High-Tg (170–180°C): Recommended for boards going through three or more reflow passes (double-sided with a rework step, or boards with press-fit connectors that require additional thermal processing). Also preferred for automotive or industrial applications where the board will operate near 130°C in service.

Specify the required Tg in your stackup document. Most manufacturers default to standard Tg unless you ask otherwise. High-Tg adds a small cost premium — typically 5–15% on material cost.

Surface Finish: The SMT-Critical Choice

Surface finish is defined as part of the stackup specification, and it has a direct impact on solderability, component compatibility, and shelf life before assembly.

Finish Solderability Flatness Shelf Life Best For
HASL (Pb-free) Good Poor 12 months General through-hole, large-pitch SMT
ENIG Excellent Excellent 12 months Fine-pitch SMD, BGA, QFN
OSP Good Excellent 6 months High-volume production, cost-sensitive
ENEPIG Excellent Excellent 12 months Wire bonding + SMT mixed assembly
Immersion Silver Good Excellent 6 months RF boards, high-frequency design

HASL deposits solder unevenly across pads — the surface profile has peaks and valleys that become a reliability risk at pitches below 0.65mm. On 0.5mm pitch QFP or QFN, HASL pad coplanarity variance can prevent the stencil from seating correctly during paste printing, leading to bridging or insufficient paste. For prototypes with only 0603 passives and SOT-23 discretes, HASL is fine and saves money. For anything with fine-pitch ICs, use ENIG.

ENIG provides a flat, oxidation-resistant surface ideal for SMT. The nickel barrier layer prevents gold diffusion into the solder joint, and the gold layer provides excellent initial wettability. One failure mode to be aware of: black pad syndrome — a phosphorus-enriched nickel layer that forms under the gold during plating and causes brittle, non-wetting joints. If you're working on high-reliability designs, it's worth asking your supplier about their ENIG process control or past black pad incidents. Most reputable manufacturers have this dialed in, but it's the kind of thing you verify once and then stop worrying about.

>> Recommend reading: HASL vs ENIG: An Ultimate Guide on Surface Finish

OSP is the right choice for high-volume production where boards are assembled quickly after delivery. The organic coating burns off cleanly during reflow, leaving bare copper that wets well. The shelf life constraint (typically 6 months, sometimes as short as 3 months after opening) means OSP is risky for prototype boards that may sit on a shelf for months before assembly.

For most prototype and low-volume PCBA orders, ENIG is the default recommendation. It handles the widest range of component types, stores well, and the cost premium over HASL on small quantities is minimal.

Controlled Impedance: When You Need It and What to Specify

Impedance control means the manufacturer measures trace impedance on test coupons and adjusts dielectric thickness or copper etching to hit your target. Not every design needs it, but these signal types always do:

  • USB 2.0 / USB 3.x differential pairs: 90Ω differential
  • HDMI, DisplayPort: 100Ω differential
  • PCIe (any generation): 85Ω or 100Ω differential
  • DDR3 / DDR4 / DDR5: 50Ω single-ended, 100Ω differential
  • RF / antenna traces: 50Ω single-ended (microstrip or coplanar waveguide)
  • Ethernet (100/1000 Base-T): 100Ω differential

If your design includes any of these, your stackup specification must include:

  • Target impedance value (e.g., 50Ω SE, 100Ω differential)
  • Tolerance (typically ±10% for digital, ±5% for RF)
  • Reference layer (which plane layer provides the return path)
  • Trace width (as-designed) and whether width adjustment by the manufacturer is permitted to meet impedance
  • Which nets or signal groups the requirement applies to
  • Coupon requirement — specify that impedance test coupons must be included in the panel

A stackup specification without these details forces the manufacturer to guess your intent, or to fabricate without impedance control and simply hope the default works.

 

How to Document and Submit Your Stackup Specification

The stackup specification doesn't have to be a formal document. In practice, most manufacturers accept:

  • A table in your Gerber upload notes (for simple 4-layer boards)
  • A dedicated stackup PDF (for 6-layer+ or impedance-controlled designs)
  • A stackup file exported from your EDA tool (KiCad's Layer Stack Manager, Altium's Layer Stack, or similar)

At minimum, your stackup document should include:

Parameter What to Specify
Total board thickness e.g., 1.6mm ±10%
Layer count e.g., 4-layer
Layer sequence Layer name, type (signal/plane), copper weight per layer
Dielectric thickness Prepreg and core thickness per layer
Base laminate e.g., FR-4 standard Tg, FR-4 high-Tg (170°C+), Rogers 4003C
Surface finish HASL, ENIG, OSP, etc.
Controlled impedance Target value, reference layer, tolerance, affected nets
Tg requirement If above standard
Bow and twist limit If tighter than IPC default 0.75%

For 2-layer boards without impedance requirements, a note in your order saying "standard stackup, 1.6mm, 1 oz outer copper, ENIG finish" is sufficient. The manufacturer's default handles the rest.

Common Stackup Specification Mistakes

  • Leaving out the surface finish. Some engineers treat surface finish as an ordering option rather than a design requirement. If you don't specify it, you may get HASL on a board with 0.4mm pitch QFNs. Specify surface finish in the stackup document, not just in the order form.
  • Specifying nominal copper weight instead of finished copper weight. Outer layer copper gets plated during PCB fabrication — a 1 oz base copper outer layer typically ends up around 1.4–2 oz finished after plating. If your impedance calculation assumed 1 oz and the finished copper is 1.8 oz, your trace impedance will be lower than designed. Confirm with your manufacturer whether copper weight specs refer to starting (base) or finished copper.
  • Not specifying prepreg type for impedance-controlled designs. Different prepregs have different dielectric constants (Dk). Standard FR-4 prepreg Dk varies from about 3.8 to 4.5 depending on glass style and resin content. For controlled impedance, specify the prepreg material (e.g., "Isola 370HR 2116 prepreg" or "standard FR-4 with Dk ≤ 4.2 at 1GHz") so the manufacturer can calculate trace width adjustments correctly.
  • Using a non-symmetric stackup and not flagging it. If your design requires an asymmetric copper distribution for electrical reasons, tell the manufacturer. They may be able to compensate with dummy copper fill on the lighter layers to balance thermal stress.

Frequently Asked Questions

Q: Do I need a stackup specification for a 2-layer board?
For a standard 2-layer board with no impedance requirements, no. The manufacturer will apply their default stackup (1.6mm total, 1 oz outer copper, standard FR-4). If you have specific thickness requirements, controlled impedance (e.g., an RF trace or USB differential pair), or non-standard copper weight, you should document it.

Q: What's the difference between prepreg and core in a PCB stackup?
Core is a rigid, fully cured fiberglass-reinforced epoxy sheet with copper foil on both sides. Prepreg (pre-impregnated) is a fiberglass sheet with partially cured epoxy — it's the adhesive layer that bonds cores together during lamination. In a 4-layer board, the two inner copper layers are etched onto a core; the outer layers are bonded to the core via prepreg during the lamination press.

Q: How do I know if my design needs controlled impedance?
Check whether you're routing any high-speed serial protocols (USB, HDMI, PCIe, DDR), RF or antenna traces, or any differential pairs operating above ~100MHz. If yes, those traces need controlled impedance. If your design is purely through-hole or low-speed (I2C, SPI at standard voltages, UART), controlled impedance is typically not required.

Q: What is PCB bow and twist, and why does it matter for SMT?
Bow and twist are measures of out-of-plane deformation in a PCB. Bow is a curvature across the board; twist is a rotation about the diagonal. During SMT assembly, a warped board doesn't sit flat on the conveyor or fixture — this causes pick-and-place misalignment, paste printing inconsistency, and incomplete BGA solder ball contact during reflow. IPC-6012 specifies a maximum of 0.75% for SMT assemblies. BGA-heavy boards often require 0.5% or tighter.

Q: Can I use Rogers or other high-frequency laminates in a standard SMT assembly process?
Yes, but with caveats. Rogers materials (4003C, 4350B, etc.) have different Tg values, CTE characteristics, and drill behavior than FR-4. They're compatible with standard lead-free reflow, but the assembly house needs to know the material so they can adjust their drill speeds, depaneling method, and reflow profile if needed. Always disclose non-FR-4 laminates in your stackup specification.

Wrapping Up

Getting the stackup specification right is one of the most leveraged things you can do before sending files for assembly. It's not a long document — for most 4-layer boards, a single table covers everything the manufacturer needs. But the information in that table — copper weight, dielectric thickness, surface finish, impedance targets — determines whether your boards come back flat, solderable, and electrically correct.

The engineers who have the most consistent results aren't necessarily the ones who spend the most time on layout. They're the ones who communicate clearly with the factory about what the board needs to be.

If you're working on a multilayer or impedance-controlled design, it's usually worth getting the stackup reviewed before fabrication — especially if the board will go through multiple reflow cycles or uses fine-pitch components. Most assembly houses can review the stackup alongside your Gerbers and flag potential issues early. If you want a second set of eyes on it, send it over.