Relieving the Strain

Should you add a strain relief bead to your rigid-flex circuit?

You are putting the finishing touches on your new design and feeling pretty good – you have made both the mechanical and electrical teams happy. At least mostly happy. One nagging issue remains: Should you add a strain relief fillet or bead to the interface between the rigid and flex portions of your circuit?

IPC-2223 discusses strain relief and considerations for and against. An estimate may be that about 50% of all designs include stain relief, and the rest do not. Keep in mind that the guidance below applies to both rigid flex and also to flex with FR4 or metal stiffeners. How do you decide if yours needs it?

First, let’s discuss the role of the strain relief bead. At its core, the goal is to avoid a very sharp bend of the flex right at the rigid/flex interface. Some are concerned resin may flow out from the rigid material that could create a sharp edge, and the flex may be bent sharp enough to be damaged by the sharp rigid or resin edge.

Application of the bead causes the bend to start some distance away from the edge and distribute the bend along the arc. There are several reasons you might use strain relief and some where you may not.

When to use it. The primary use case for strain relief is when the flex is very thin and flexible, the bend radius is tight, and there is high likelihood the bend will take place right where the flex exits the rigid section. This is a fairly small percentage of designs. It typically means the flex is only one or two layers, or maybe up to four layers if unbonded. Flex layers are usually signals, and at most one plane layer. These thin flexes are at the greatest risk of some type of damage due to localized excessive strain, either bending sharply or even twisting stresses. The strain relief bead adds value in these cases.

The other good case for the strain relief is when the flex must bend sharply as soon as it exits the rigid in order to fit within the assembly or chassis enclosure.

Bend radius is also a determining factor. Generally, if the flex is bent less than 90°, there is no need for strain relief. The value is far greater on parts with 90° and 180° installations. Keep in mind that sometimes you may need to bend to 135° or 180° just to do the installation, even if the result is 45° or 90°. So consider the act of installing, not just the end state.

And when to not to. Like any feature, however, strain relief comes at a cost. It is a manual process performed one part at a time. So, if it is not needed or advised, it is a possible opportunity for cost savings. When can we avoid it?

When the flex length between rigid sections is quite short, adding strain relief will reduce the total flex available to complete the required arc. This will lead to a smaller effective bend radius, which may negatively impact compression and tension strain on the inner and outer diameter. For example, if the flex length is 0.500″ (12.7mm) or less, the strain relief applied at both ends could consume upwards of 40% of the total flex length. IPC-2223 suggests a bead width of 1.0-2.5mm.

You may be able to require a small bead on your drawing, but even if it were half that width, it would consume 20% of the flex length. Driving to a smaller bead width naturally diminishes the value of the bead itself, as it protects less and less of the interface zone. Keep in mind, to achieve a smaller bead, the supplier must carefully apply and monitor the bead application. The viscosity of the material permits it to flow and spread to some degree prior to curing. As you drive the requirement smaller, you will be driving the unit cost higher.

IPC-2223 also advises against applying a bead when the vertical height from the flex surface to the top of the rigid (Layer1) is less than 0.010″ (0.254mm). Reason: Keeping the bead height that low is very difficult. As a result, the bead will protrude above the surface of the outer layers of the board and may interfere with the solder paste screening process.

Beyond these two examples, there are other cases where the bead is not as valuable.

When the flex is three or more layers bonded together, the natural stiffness of the flex will cause the flex to bend in a full arc from rigid-to-rigid section unless you specifically bend it in a particular location. As layer count rises, the ability to bend the flex tightly near the rigid edge is very limited, and the bead adds little value.

There are also those rigid-flexes with six to 20 layers of flex made up of unbonded pairs of layers. As the layer count rises, the cumulative effect of all the material impacts the natural radius as the part is bent. The strain relief has little impact on high-layer-count flex.

Copper content in the flex must also be considered. If the flex uses heavier copper, like 2oz. (70µm), it is harder to bend. Similarly, if multiple plane layers are in the flex delivering power and ground or shielding, they will be far more mechanically robust than sparse signal layers. The plane or shielding layers protect the internal signal layers. They reduce the potential risk from twisting or torsional strain. The load on the exterior plane layers would need to be tremendous to cause damage to a trace on an internal flex layer.

Finally, if the bend is naturally located well away from the rigid-to-flex transition, then the bead is serving no mechanical purpose and can be omitted in this case.

Still not sure what to do? Get your fabricator’s opinion. I often suggest that you start without the bead and add it if validation testing indicates it is needed. Bending tests can be run on samples to determine if the strain relief is providing a margin of safety that you need in the application.

Nick Koop is director of flex technology at TTM Technologies (ttm.com), vice chairman of the IPC Flexible Circuits Committee and co-chair of the IPC-6013 Qualification and Performance Specification for Flexible Printed Boards Subcommittee; nick.koop@ttmtech.com. He and co-“Flexpert” Mark Finstad (mark.finstad@flexiblecircuit.com) welcome your suggestions.