There are two aspects associated with flex circuit bending: one is static or one-time bending, the other is dynamic flex involving multiple bending operations.
Today, technology trends are increasingly toward flex circuits or a combination of rigid-flex circuits for wearable/IoT PCB designs. You can say those trends put us on a different footing, so to speak. Therefore, it's important to get a handle on new design terminology and things that need to be factored in as you move to this next level of embedded design.
Board modulus refers to its structure—a low modulus means a softer structure, while high modulus refers to a harder board with stiffener. A stiffener that provides rigidity to the flex circuit for stable soldering is shown at the top of Figure 1; components mounted on the opposite side of the stiffener are shown at the bottom of Figure 1. Stiffeners are an inexpensive way to rigidise certain areas on the flex boards, such as SMT areas, pin areas, or hole pattern locations for component mounting.
Figure 1: A stiffener provides rigidity to the flex circuit for stable soldering (top) and components mounted on the opposite side to the stiffener (bottom) (Source: NexLogic Technologies)
SMT areas don't always need stiffeners depending on the components being installed at that location. However, adding a stiffener is going to add very little cost to the assembly. Stiffeners are used to reinforce solder joints and are sometimes used to force bend lines in the selected areas. Stiffeners can be made from FR4, polyimide, copper, or aluminium-based materials.
Coming around the bend
Regardless of the application, a flex circuit must be pliable and bendable, but the question is: How pliable and bendable can it be? The jury is still out on the preciseness of bendability (or "bendableness"). As of this writing, the IPC is being rather conservative with its call. So, in essence, the exact definition or gauge of bendability hasn't been pinned down and probably won't be due to its nebulous nature. The best advice given is to rely on an experienced EMS Provider that has several wearable/IoT PCB designs under its belt and has a storehouse of critical nuances associated with flex circuit bendability.
Having said this, it's good to know terms like bend radius, bend ratio, and strains, all of which are inextricably intertwined. Bend radius, as the name implies, is how far you can bend that flex circuit before something breaks or incurs a latent fracture. It's also necessary to know that the measurement of the bend radius is performed from the bend's underside surface. Again, it's advisable to partner with a savvy EMS Provider to analyse and resolve bend radius questions and issues.
The second term, bend ratio, takes into account the ratio of the bend radius to the thickness of the flex circuit. For example, the bend ratio for a multi-layer flex circuit for a medical electronics wearable device is at least 20:1. By comparison, for the single and double-sided flex circuits the bend ratio should be at least 10:1. Tighter bends may create the risk of circuit damage. It's always preferable to use more gradual angles rather than a right angle bend with a sharp radius. Bend radius is calculated by measuring the distance from the inside surface of the bend to the centre of the radius.
It's also important to know there are two aspects associated with flex circuit bending. One is static or one-time bending; the other is dynamic flex involving multiple bending operations. The bend radius for static bending should be at least 10 times the thickness of the circuitry and the strain on the critical layers should be 2.2% or less. On the other hand, the bend radius for dynamic flex circuitry should be 25 times or less.
The bend radius for dynamic flex circuitry, such as the example populated with µBGA packages shown in Figure 2, should be less than 0.8% for 50,000 cycles, less than 0.6% for 100,000 cycles, less than 0.4% for up to one million cycles, and less than 0.2% for a million cycles or more.
Figure 2: Flex circuit populated with µBGA packages (Source: NexLogic Technologies)
As was noted above, the bend radius, bend ratio, and strains created by the flexing action of the boards are inextricably intertwined. In the case of strains, these are already built in when the manufacturer produces the flex circuit. In other words, strain is inherent in the different circuit layers and can be mitigated with strain relief devices such as stiffeners.
Dielectric materials within the flex circuit can cause more strain depending on their thickness. Dielectrics differ in their ratio of stiffeners to thickness. Choosing a dielectric material according to the underlying application gives the finished flex circuit the quality it needs. In terms of impedance controlled designs, the conductor widths and dielectric thicknesses can be adjusted to meet the required impedance results.
As previously noted, high modulus PCBs are hard boards with stiffeners. Here, bend radius is extremely important and needs to be factored in because the calculation of bend ratio should also incorporate the thickness of the stiffener, thereby increasing the total thickness for the flex circuit. Keeping the bend ratio small increases the flex circuits' reliability.
This means that it's important to understand the strain at different levels within flex circuit layers. In turn, this means knowing which layers use what amounts of copper. Changing the amounts of copper has the most adverse effect on strain difference.
For example, take a flex circuit with a half-ounce of copper weight. This will bend with a particular strain amount with a specific bend ratio. However, if the amount of copper was to be doubled to one ounce, the flexibility would be considerably reduced, and the bend ratio would be limited because the copper thickness has doubled, thereby creating an overall thicker flex architecture. All of this means that you have to calculate the bend ratio very carefully.
Furthermore, you have to check copper thickness at different layers within the flex material. This is because the thickness affects the bend ratio and the strain factor. You can use certain types of flex material for certain applications, so it's not a case of "one-size-fits all."
When it comes to placing vias, there are certain rules you have to follow since vias can experience a lot of fatigue due to bends and curvature. Again, the greater the number of layers, the more difficult it becomes for vias to maintain their integrity because they have to maintain adherence with the circuit's multiple layers. Take a six-layer flex board, for example. You need to make sure each layer internally adheres to all others while the circuit is stationary, as well as when it's flexing and bending.
It's important to place vias appropriately, keeping in mind the flexing movement of the circuitry. In terms of spacing, it's recommended that you keep a 20-mil clearance between other vias that are being placed on the same board, as well as 20 mils from the vias to the edge of the board.
When you're placing the vias, you can define areas on the PCB layout where the circuit is not going to be bent—or where any bending is minimal—and then place your vias in these areas.
In the case of a rigid-flex board, as illustrated in Figure 3, you can place a minimum number of vias in the flex circuitry and try to keep the majority of the vias in the rigid section. Also, when you're placing the vias, the typical via size used for flex circuit is 5 mils; however, depending on the application and aspect ratio, different via sizes can be used.
Figure 3: Rigid-flex board in a panel form (Source: NexLogic Technologies)
The rule of thumb is to keep the vias in the proper place so that they are performing their main function, which is to carry current. At the same time, the flex circuitry must have the integrity to keep the connection and be able to withhold the fatigue of bending.
Regular vs. annealed copper
As a wearable/IoT product designer, you may not come across this particular area since manufacturers produce ready-to-use flex circuits. However, it's something that'll make you savvier and possibly avoid design issues. For starters, don't use electrolytically-deposited or ED copper in your flex circuit. ED copper is normally used for rigid PCBs; for flex circuits, it's best to use rolled annealed copper.
Rolled copper is a considerably better, more pliable material. Its surface is treated to make it smooth, meaning it's more amenable to bending and flexing. Having said this, some ED copper versions that are characterised by special grain structures can be highly effective for flex circuit bending. In most cases, however, these ED coppers aren't cost-effective for the majority of wearable/IoT devices.
Flex layer core thickness plays an important role, as does keeping the same finish thickness on all rigid areas. As for rigid circuitry, you want to avoid having 32 mils on one side and 62 on the other, for example, otherwise the sequential lamination process of rigid-flex circuit fabrication comes into question and poses difficulties, so it's prudent to keep the same finish thickness in all rigid areas.
Normally, in a rigid board, you have even number of layers. By comparison, in a flex board you can have even and odd numbers. For example, you could have six layers on the rigid side but only three layers on the flex side.
Layer construction when designing the flex is asso very important. You have to make sure you are minimising the thinnest possible construction for the bend radius to improve flexibility. If you have a five-mil Kapton material versus two-mil Kapton, flexibility and bend radius will be better for the two-mil Kapton.
Also, when designing rigid flex circuitry, you have to ensure both flexibility and mechanical reliability. You have to take into account the fine balance that comes with experience to make sure that the board being designed is flexible enough to perform its function and reliable enough to endure the flex and bend cycles that are being calculated for its life cycle. Normally, you use half-ounce copper for flex boards. In extreme cases, when high capacity is required, you can use one ounce, but this is the exception, not the rule.
One thing you need to do is perform a paired structure. For example, if you're working with an eight-layer rigid board, you can have four flex layers or two layers of flex circuits.
Also, it's best to try to offset the traces from layer to layer in the bend area. This is because multiple traces going into the bend area jeopardise the flex circuitry over the long term when it is flexing and bending an inordinate number of times. If there is an offset, then all the stress and strain is not concentrated at one point, but is instead distributed throughout the circuitry. This means the stress and bend area are considerably more flexible and reliable over a longer period.
When it comes to impedance controlled design, sometimes offsetting trace layers may not be possible. The reason for this is you need to have the trace in the proximity of a solid reference plane, which may not allow you to implement an exact offset of traces. Impedance controlled design may make it challenging to maintain staggered traces, which directly affects mechanical flexibility and reliability.
What can be done is to offset impedance controlled traces with subsequent layers. For example, you can run one trace on layer 3, you can run the reference plane on layer 4, and you can run the other matching staggered trace on layer 5. Thus, you can offset the traces between different layers, but you still need to keep the reference plane in mind because impedance is a function of the signal's distance from the reference plane.
Offsetting impedance control traces with subsequent layers is but one of several design considerations that need to be factored in when creating a flex circuit-based wearable and/or IoT design. The points elaborated on in this article are the major ones that need to be considered, including bend radius, bend ratio, strains created at different locations, and via placement. However, as you move along in your designs, you'll find others to factor in depending on whether you are targeting a consumer, industrial, military/aerospace, or medical electronics application.