Sunking PCB

Flex Circuit Shielding: What It Changes in the Stackup

Design Guide2026-10-21Sunking Circuits Technical Team
Flex Circuit Shielding: What It Changes in the Stackup

The short version: a shield is not something you bolt on at the end. It becomes a layer of the stackup. The moment you decide to shield a flex circuit, you have also changed three other things: the reference plane your signals see, the total thickness and neutral-axis position, and the physical path the ground return has to take. Those map onto impedance, bend life, and whether the fabricator can release the job at all. Treating the shield as "one more layer we add last" usually ends in impedance that came back low, bend-area cracking earlier than expected, or a fabricator email asking where this ground is supposed to connect.

1. The shield enters the stackup, not the enclosure

A flex stackup is already thin: base film, adhesive, copper, coverlay, all stacked within a very small total height. At that scale, nothing you add can stay out of the structure.

A shield participates in at least three ways:

  • Electrically, it becomes a new reference plane for the traces. Whether you had single or dual reference, and how far that reference sits from the signal, both change.
  • Mechanically, it adds thickness, and because it sits on one side, it pushes the neutral axis off center. Layers further from the neutral axis see more strain when the part bends.
  • Process-wise, it has to be tied to ground, and that tie is made by a coverlay opening, a via, or an edge structure. Each one is a manufacturing action that has to appear on the drawing.

That is the problem this article addresses: which of the three each option costs you, and in what order to decide. Overall flex stackup and process limits are covered separately; here we assume you already have a buildable flex stackup and now need to put a shield into it.

2. Three approaches, three different places the cost lands

In practice there are three routes. They are not ranked by "how well they shield" — they differ by where the cost lands.

1. Conductive shielding film (applied)

A composite film with conductive adhesive, applied over the coverlay or replacing part of it.

  • Cost to the stackup: added thickness, and that thickness is an adhesive-bearing composite whose behavior under heat and pressure differs from the base film.
  • Cost to the electrical side: it becomes the reference plane, but it sits behind coverlay and adhesive — layers whose dielectric behavior is not the same as the base material. You cannot simply model it as "one more ground plane."
  • Cost to the bend: relatively mild, which is the main reason it is popular. Film follows a bend better than copper.
  • Cost to manufacturing: placement accuracy, where the boundary terminates, and how it grounds all have to be stated.

2. Solid copper shield layer

Treat the shield as a real copper layer that goes through normal imaging and lamination.

  • To the stackup: it is a copper layer. Layer count goes up, and press cycles plus structural symmetry have to be reconsidered.
  • To the electrical side: the most deterministic reference plane and the most controllable impedance, because what sits between it and the signal is known base material and adhesive, not an extra composite.
  • To the bend: the heaviest cost. Solid copper is the least welcome structure in a bend area — thick and continuous means strain concentrates.
  • To manufacturing: the most conventional, since it uses processes already in place.

3. Hatched (mesh) copper shield

Making the shield copper a grid instead of a solid plane is a very typical flex compromise.

  • To the stackup: occupies a layer like solid copper, but with less effective copper.
  • To the electrical side: this is the one to watch. A hatched reference plane is discontinuous. For low-frequency and general EMC cases the effect is limited, but for controlled-impedance high-speed traces, once the opening scale approaches the trace scale, the plane can no longer be treated as uniform.
  • To the bend: clearly better than solid copper, which is exactly why it exists. The grid gives the copper room to deform.
  • To manufacturing: the hatch pattern has to work with bend direction and trace routing. It is not something a generic fill solves.

The point: the difference is not "which shields better" but "would you rather pay in impedance, in bend life, or in process complexity." Decide which one you cannot give up, then work backwards.

3. Change the reference plane and you must recalculate impedance

This is the step most often skipped and the one that generates the most rework.

Controlled impedance is a relationship between the trace and its reference plane. Adding a shield changes that plane, in at least three ways:

  • Number of reference planes. You may have gone from ground on one side to ground on both. That drives impedance down noticeably. A trace width calculated for single reference will come back low if reused unchanged.
  • Distance to the reference. How far the shield sits from the trace depends on the coverlay and adhesive stack, not on the distance you assumed while drawing.
  • Continuity of the reference. Solid copper is continuous, hatched copper is not, and a film's conductive layer has its own structure. Continuity is what makes the "reference plane" assumption valid in the first place.

So the order has to be: choose the shield → calculate impedance → set trace width. Any other order is rework.

Even with the right order, impedance still carries fabrication variation from etching, lamination and dielectric thickness tolerance. That part is broken out in its own article, and how to pick and state tolerance classes has one too. The one thing to carry away here: the reference-plane change caused by shielding is a change in design input, not fabrication variation. Do not expect the fabricator to absorb it inside the process window.

If the trace is also high-speed, copper surface condition enters as well — roughness versus insertion loss is an independent variable that compounds with the shielding choice.

4. Total thickness and neutral axis: can the bend area carry this shield

Bending is the entire point of a flex circuit, and it is the first thing sacrificed when a shield is added.

Two mechanical facts decide the outcome:

First, the neutral axis. When a board bends, the inside is in compression, the outside in tension, and somewhere in between strain is zero. The further a layer sits from that axis, the more strain it takes. Adding a shield increases total thickness and makes the stack asymmetric, so the neutral axis moves. A signal layer that used to sit near the neutral axis can end up on the tension side purely because a shield was added.

Second, how continuous copper behaves in a bend. Thick, continuous copper is the least favorable structure in a bend area. That is the whole justification for hatched shields — they trade reference continuity for bend margin.

Which leads to a few practical handling rules:

  • The shield does not have to run the full length. Many designs cover only the segments that need it and keep clear of bend areas. Segment boundaries then have to be explicit on the drawing.
  • If the bend area must be shielded, hatched copper is usually the only realistic option, and the pattern has to suit the bend direction.
  • Do not terminate the shield boundary inside a bend area. Any layer's start/stop boundary is a stiffness discontinuity, and a discontinuity placed where the part flexes repeatedly is where cracks start.
  • Respect symmetry where you can. Shielding one side offsets the neutral axis more than shielding both.

Bend radius and copper foil type are treated in more detail in their own article, and bend-area routing rules have one as well. This article only adds what neither covers: how a newly added shield layer participates in all of it.

5. How the ground connects: this is what makes the file releasable

A shield that is not grounded is not a shield, just a floating sheet of metal. On flex, how it grounds is precisely the item that most needs to be stated, because every option carries a manufacturing action:

  • Coverlay opening plus conductive contact: open the coverlay where the shield has to meet ground, so it contacts a ground plane or pad. Opening location, size relationship and contact method all need stating.
  • Via connection: tie the shield to an internal ground layer with a via. Via placement on flex is constrained, and definitely not in a bend area.
  • Edge or end connection: carry the shield to a board edge or the connector end and ground it there. Cleanest electrically, but the layout has to leave a path.
  • Conductive adhesive contact on a shielding film: the adhesive's own conductive path reaches the ground structure below. The contact area has to be defined.

An engineering consequence that often gets missed: the position and number of ground points is not purely an electrical choice, it is simultaneously a mechanical one. Every ground tie is a local stiffness change and a stress concentration. A ground tie inside a bend area is a hard point placed exactly where the part needs to be soft.

So the drawing has to state at minimum:

  1. Where the shield sits in the stackup and in what layer order (which layer, inside or outside the coverlay).
  2. Its form (film / solid copper / hatched copper), with hatch pattern intent if hatched.
  3. Coverage and boundaries (which segments are shielded, which are not, what the boundary references).
  4. Grounding intent (where, by what method, how many points).
  5. Which dimensions are functional requirements and which the fabricator may adjust within the process.

Miss any one of them and the fabricator either asks, or builds to their own interpretation — and the second is the real risk. Conventions for stating material roles and layer order are visible on the materials page and the tolerance page.

6. On rigid-flex, where does the shield stop at the transition

If this is a rigid-flex build, the shield raises one more mandatory question: what happens at the rigid-to-flex transition?

Three handlings, three consequences:

  • Shield runs through the transition: electrically continuous, but the transition is already the most stress-concentrated region and another layer tightens it further.
  • Shield terminates before the transition: mechanically safer, but the shield is interrupted there, so grounding and continuity need separate arrangement.
  • Form changes at the transition (solid in the rigid section, hatched in the flex section): a compromise, but the junction is where two structures meet and the handoff location must be explicit.

Whichever you choose, give the location explicitly rather than letting the fabricator infer it. Transition-zone manufacturing constraints are documented separately, and overall rigid-flex design inputs are in the rigid-flex design guide and on the rigid-flex product page.

7. Coverlay, adhesive and shield: the stacking order

Order is not cosmetic; it changes both the electrical and mechanical outcome:

  • Shield outside the coverlay: coverlay and adhesive sit between shield and trace, so the distance is larger and the dielectric stack more complex. Impedance impact is relatively smaller, but the shield is on the outermost surface and needs protection considered.
  • Shield inside the coverlay: closer to the trace, acts more strongly as a reference plane, larger impedance impact, but better protected structurally.

Both are standard practice. What matters is that once chosen, the layer order is written down, and that impedance is calculated against the actual order rather than a simplified "it's a ground plane either way" model. Coverlay material selection logic is covered separately.

8. How to verify the shield was actually built

Shielding is one of the few structures where "was it done right" is hard to judge visually, so the verification method has to be fixed at drawing stage, not improvised when boards arrive.

Workable directions:

  • Continuity and ground confirmation: verify the shield actually reaches the ground the design intended rather than floating. The most basic item and the most frequently skipped.
  • Layer order confirmation: confirm by cross-section that the shield sits where the stackup says it does.
  • Coverage confirmation: shield start/stop boundaries match the drawing, especially segments deliberately kept clear of bend areas.
  • Impedance verification: for controlled-impedance traces, verify against the shielded stackup. Pre-shield data does not carry over.
  • Bend-related verification: verify under the product's actual bend conditions, not static appearance alone.

The point: most shield verification items are structural conformance, not performance figures. What a fabricator can deliver is consistency of structure and connection. Shielding effectiveness itself is a system-level result influenced by enclosure, cabling and the grounding system, and should not be assigned to the fabricator alone. General verification methods and release logic are on the quality page; capability boundaries are on the capabilities page.

9. Decision order: ask which one you cannot give up

Collapsing the sections above into a usable sequence:

  1. Is this trace controlled impedance? Yes → reference determinism comes first; lean toward solid copper or a well-defined film, and recalculate impedance against the shielded stackup.
  2. Does this trace flex repeatedly? Yes → bend life comes first; lean toward hatched copper, or keep the shield clear of the bend area.
  3. Both? → this is the real conflict, usually resolved by zoning: keep bend areas and controlled-impedance areas from overlapping, use the right structure in each, and state the junction explicitly.
  4. Fix the ground path, and check whether any ground point lands where the part should stay soft.
  5. Put it on the drawing: layer order, form, coverage, grounding intent, which dimensions are functional.
  6. Define verification items, especially ground continuity and layer-order conformance.

Outline and opening machining also interact with shield boundaries — contour cutting and punching constraints are worth reading alongside. Flex product scope is on the flex PCB product page.

FAQ

Q: Which shields better, film or copper?

A: There is no general answer from the manufacturing side, because shielding effectiveness is a system-level result influenced by enclosure, cabling and the grounding system. What manufacturing can answer is where the cost lands: solid copper gives the most deterministic reference plane and the worst bend behavior; hatched copper is bend-friendly with a discontinuous reference; film is mild on bending but puts a more complex dielectric stack between shield and trace. Choose by "which one you cannot give up," not by "which is better."

Q: Can I keep the impedance I already calculated after adding a shield?

A: No. Adding a shield changes the number of reference planes, the distance to them, and their continuity — that is a change in design input. A trace width calculated for single reference will come back low on a dual-reference structure. Recalculate against the actual shielded stackup, then set trace width.

Q: Is a finer hatch always better?

A: A finer hatch brings the reference plane closer to a continuous plane, which helps impedance; but the more continuous the copper, the less bend margin remains. That is exactly where the compromise sits. The hatch pattern also has to work with bend direction and trace routing rather than being treated as a generic fill. Specific values are design input and depend on what this trace actually requires.

Q: Can a bend area be shielded?

A: Yes, but the form has to change. Solid copper is the least favorable structure in a repeatedly flexed segment; if the bend area must be shielded, hatched copper is usually the realistic option. Also keep the shield's start/stop boundary out of the bend area — a boundary is a stiffness discontinuity, and a discontinuity where the part flexes repeatedly is where cracking starts.

Q: Can ground points go anywhere?

A: No, and the constraint comes from both sides. Electrically they determine whether the shield works at all; mechanically each tie is a local stiffness change and stress concentration. A ground tie in a bend area is a hard point placed where the part most needs to flex. Position, method and count all have to be stated explicitly.

Q: How completely does this have to be specified before a fabricator can release the job?

A: At minimum five items: the shield's position and layer order in the stackup; its form (film / solid / hatched, with pattern intent if hatched); coverage and boundaries; grounding intent (where, how, how many points); and which dimensions are functional versus adjustable within the process. Miss one and the fabricator either asks or builds to their own interpretation.

Q: How do I verify the shield was built correctly?

A: Most manufacturing-side items are structural conformance: ground continuity (the shield really reaches the intended ground and is not floating), layer-order conformance (confirmable by cross-section), coverage and boundary conformance, impedance verification against the new stackup for controlled-impedance traces, and verification under the product's real bend conditions. Shielding effectiveness itself is a system-level metric and should not be assigned to the fabricator alone.

---

Prototyping a flex or rigid-flex board?

Bend radius, coverlay and stiffener placement decide reliability. Tell us the bend cycles and assembly method and we will propose a construction with pricing.