A bend radius lower than the recommended minimum needs a closer look at the details.
It seems self-evident that designers create a flex or rigid flex because the part must bend. That does not tell the whole story, however. We need more information to ensure the design works in the application as intended.
For starters, are we talking “Use A” (flex to install), or “Use B” (dynamic flex)? I estimate that 95% or more of the flex designs are Use A; truly dynamic uses are rare. Let’s talk first about flex to install.
For flex to install, rules of thumb are well-defined. IPC design guidelines establish that for 1- and 2-copper layer circuits, the bend radius should be at least 10 times the thickness of the flex. For circuits with three or more bonded flex layers, the rule of thumb is 20 times the thickness. The key for parts with multiple unbonded flex layers is to select the thickest set of bonded flexes for the radius assessment. For example, if the part is five flex layers but is unbonded between the second and third flex layers, making a 2- and 3-layer section, do your assessment on the thickness of the 3-layer portion, not all five layers.
The guidelines ensure the copper in the stackup does not experience significant plastic deformation during the bend process. In many cases, a small amount of deformation may be acceptable, especially if the application is truly “bend to install” and does not move again. If you expect some type of movement for servicing, it’s better to use more conservative values.
Keep in mind, these guidelines are intentionally conservative because they consider all possibilities, including bend angle: 45°, 90°, 180°, etc. Material selection and artwork patterns on each layer can also impact performance. If the bend radius ends up lower than the recommended minimum, take a closer look at the details.
Bend angles less than 90° are more or less a non-concern. I don’t recall ever seeing failure on a part bent less than 90°. For 90° to 180° bends, use a fixture to control the bend process to ensure a repeatable process with a controlled radius.
Figure 1. Bend angles less than 90° are typically a non-concern.
Generally, we want the layer with the narrowest traces to be in the middle of the stackup along the neutral axis of the bend. The neutral bend axis defines the point in the cross-section where no compression or tension exists. If signal traces can go there, the radius can get quite tight without risking foil deformation. This is unlikely, however. On a two-layer flex, the center core dielectric usually acts as the neutral bend axis, while the two foil layers experience tension and compression, respectively.
For the two-layer flex, we recommend putting the layer with the thinnest traces on the inside radius of the bend. Experience tells us that traces are more likely to crack under tension than compression. Also, plane layers are unlikely to crack unless severely creased.
In all cases, wider traces perform better. Wider traces withstand tighter bend radii and more bend cycles because any fatigue failures take longer to initiate and propagate on wider conductors. The wider trace can overcome certain grain boundary imperfections.
When working with multilayer flex, the same idea applies, and it is more common to see the signal layers sandwiched between plane layers for EMI and impedance purposes. This comes with a few advantages. First, the signal layer sits closer to the centerline of the stack and neutral strain. Second, the planes above and below protect the signal layer from torsion stresses on the outermost conductors. Put simply, you need to break or severely damage the plane layer before you can break the internal signal traces.
Some construction options can help reduce thickness. For example, using foils of 12µm and 18µm thicknesses allows thinner coverlay materials, shaving a few mils off the stack, which is a big deal when working with 10X and 20X multipliers. Using unbonded flex pairs increases flexibility and lets you define the radius based on two-layer construction and 10X rather than 20X.
We have observed that cases with very aggressive bends can be quite successful; however, doing your homework beforehand is crucial. This research can include microsectioning to evaluate the distortion in the radius, confirming that the end-result meets performance needs. Depending on the severity of distortion, you may need to implement some method to anchor the part in place after bending to avoid work hardening the foil.
Dynamic bending presents a different story. The 10X and 20X rules of thumb do not apply. Bend radii must be much larger to avoid work hardening of the copper foil. For dynamic applications, you are usually limited to one or two flex layers unless the bend radius is very large and the range of travel is very limited. There are many permutations of dynamic bending, but here are some examples:
- Coiling and uncoiling a spooled flex
- Opening and closing an S or U bend
- A unidirectional bend (0° to 90° and back to 0°) repeating
- A bidirectional bend (0° to 90°, back to 0° and then to -90°) repeating
- Other custom dynamic moves such as one end of a board staying stationary and the other moving slightly up and down.
All these are impacted by the number of expected cycles as well as the frequency and speed of the bends.
Dynamic bend designs clearly need the copper as close to neutral as possible and use the thinnest, most flexible construction practical. These are almost always validated with an endurance test. While a couple standards exist in the test methods, custom test systems often emerge to emulate the real bend mechanism. Generally, the ability for the part to meet the bend requirement is almost entirely design related. Manufacturing process variations are rarely a factor. So, it is very important to mind all the details when optimizing for a tight bend or dynamically moving flex. As always, if you have a good relationship with your fabricator, lean on them to help make the best possible choices.
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; This email address is being protected from spambots. You need JavaScript enabled to view it.. He and co-“Flexpert” Mark Finstad (This email address is being protected from spambots. You need JavaScript enabled to view it.) welcome your suggestions. They will speak at PCB West in September.
