At the conclusion of our webinar, Tips for Designing Flex Circuits for Medical Applications, we had several questions submitted to our presenter, Zachary Walker, Product Manager of Flex and Rigid-Flex Circuits at Epec. We have compiled these questions into a readable format on our blog.
Q&A From Our Live Webinar
Quick Links:
- What is the bending lifespan of a dynamic flex? Can I expect it to fail after 10,000, 100,000 bends?
- I need to use LCP for my flex material due to the material requirements. Can this be done?
- Why is RA copper used?
- My design calls for 85-ohm impedance. Do you have a stack-up for that impedance?
- Can Megtron-6 or a better material be used with rigid flex?
- If I have a static (under 100 bends - during assembly) rigid-flex design, what is the bend guideline? Anticipating 90-ohm, impedance-matched, 4-layer flex, should I still need RA copper there, too?
- Since the minimum bend radius is 100x the board thickness, what is the typical range for board thickness?
- I believe you said 100x the board thickness is for dynamic flexes. Is the minimum bend radius different for static bends?
- What's the best approach to handle overengineered boards when design requires higher specs?
Watch the Recording Below:
Question: What is the bending lifespan of a dynamic flex? Can I expect it to fail after 10,000, 100,000 bends?
Answer: That's a little bit of a difficult question, and the reason being is IPC, when they consider dynamic flex, they like to use the term infinite. It's not truly infinite, but that's the term they like to use. So, it can be kind of difficult to nail down to say if a board's good for 100,000 bends or 1,000,000 bends. It could be difficult to nail down an exact number. And it can even be more difficult because you start having to consider the layer count, how dramatic of a bend is it, or, the thickness of the copper, anything of that nature. So, you can't exactly nail down a lifetime and failure point for bending. Usually, that's only done through testing. And with testing, it could be difficult to do and would require pure mechanical field testing, if possible.
Question: I need to use LCP for my flex material due to the material requirements. Can this be done?
Answer: Unfortunately, we cannot support LCP, which is shortened for Liquid Crystal Polymer. We are always welcome to looking over designs that may require it. However, we have in the past experienced designs that have been brought to us with LCP as the key performance material that they may need. But generally, since we can't supply boards with that material, it kind of circles back to us evaluating the designs and seeing, OK, does the design truly need LCP? Can we deviate to just standard polyamine? Sometimes the answer is yes. Sometimes, material considerations are too strict, and you end up finding that standard polyimide works just as well as LCP, and that you don't need to spec out an expensive, exotic material just to have your design function. Sometimes it will just function normally with polyimide.
Question: Why is RA copper used?
Answer: RA copper, which stands for rolled annealed, is used primarily because it's better for bending. RA copper is a copper that, when compared to ED, electrodeposited copper, or treated copper, RA has been heat-treated and treated in such a way that all of the grain direction is oriented along one axis, which can allow for bending and can improve bending even in certain cases. And so, many flex designs will end up using RA copper because sometimes you need that one or two percent extra flexibility. It's not a quantifiable extra flexibility, but sometimes you need that little bit of guarantee. So, many designs will use RA copper, and many designs will even specify grain direction in their drawings for RA copper.
Question: My design calls for 85-ohm impedance. Do you have a stack-up for that impedance?
Answer: I believe we do have an 85-ohm stack-up. Generally, we will provide stack-ups and be able to supply stack-ups for things like 50-ohm single, 75-ohm differential, 100-ohm differential. There are certain standard "impedances". However, some designs might require unusual ones that may end up going outside of those for whatever reason. We may not have the tried and practiced stack-ups that we've manufactured for those impedance values, but that doesn't stop us. We can fall back on the calculator just as easily and be able to calculate it, and then at the end of our design, be able to test it to ensure impedance. And so, in that case, 85, I believe we have a stack-up. But if you require a special impedance beyond that and we don't have it on hand, we could, at least, say that we can calculate it for you, and then at the end, provide testing to ensure that we're within tolerance for those impedance values.
Question: Can Megtron-6 or a better material be used with rigid flex?
Answer: From what we have experienced, we have not had any success using Megtron-6 material in a rigid-flex. Many materials will need to have low flow properties, such that the prepreg used does not flow into undesirable sections of the flex during lamination, which can limit material selection. As for any materials that are ‘better’, there may be such materials that can be used for rigid-flex, but at this time, we can only recommend sticking to standard FR4, since it is the easiest to manufacture and most widespread in the industry. If you do want to explore more specialized materials, please feel free to reach out so that we may research the material and discuss it together.
Question: If I have a static (under 100 bends - during assembly) rigid-flex design, what is the bend guideline? Anticipating 90-ohm, impedance-matched, 4-layer flex, should I still need RA copper there, too?
Answer: For static bends, you can anticipate a bend radius of 6X the thickness for 1-to-2-layer designs and 12X the thickness or greater for more than 3 layers of flexible circuitry. If your design is a standard 4-layer flex with 90-ohm impedance on the surface, we can expect an approximate bend radius of 0.141”, which will be subject to change depending on copper weight and required overall thickness. RA copper is advised, but not as necessary for a static bend as it might be for a dynamic bend. Should you be close to the bend radius, we will always recommend using RA copper, as sometimes that little bit of extra flexibility can make a difference.
Question: Since the minimum bend radius is 100x the board thickness, what is the typical range for board thickness?
Answer: The typical range will vary from a 1/3rd oz copper, 1-layer design at 64um, all the way up to a 2oz copper, 2-layer at approximately 340um. However, I advise against meeting the upper or lower portions of that range, if possible, since being too thin, the board becomes difficult to manufacture, and being too thick can be difficult to bend at all.
Question: I believe you said 100x the board thickness is for dynamic flexes. Is the minimum bend radius different for static bends?
Answer: The minimum bend radius is defined further for static than dynamic per IPC-2223. While dynamic is limited to 1–2-layer designs, static bends can be applied to those layer counts and above. For 1-to-2-layer designs, the radius will be 6X the thickness, and for 3 or more layers, the bend radius is 12X or greater, with greater being reserved for more layer counts on a case-by-case basis.
Question: What's the best approach to handle overengineered boards when design requires higher specs?
Answer: When a design requires higher specs or if it is over-engineered, the first place I look is the material callouts. If material equivalents exist, or if substitutions can be made if the design is going too exotic for materials, such as calling out specific types of FR4 when, as long as the IPC slash codes are the same, they may have an equivalent.
The next place I look would be via structures. Via structures are the easiest way to overcomplicate designs, increase cost, and lead time. The first idea is to find common electrical nets and see if it’s possible to consolidate a blind/buried via to an existing through via connection and eliminate it. Alternatively, sometimes adding extra layers can help by dividing traces among multiple layers rather than trying to consolidate them across a single layer and requiring blind or buried vias.
Ultimately, it’s difficult to create a ‘silver bullet’ to simplify all complex designs, so the best case is to provide it to a manufacturer and work with them to see if there are ways to simplify the design.
Key Takeaways
- Flex lifespan depends on design details, not a fixed bend count: Dynamic flex circuits do not have a guaranteed number of bends. Lifespan is influenced by layer count, copper thickness, bend severity, and overall construction, and must be validated through mechanical testing.
- Material choices should be justified, not assumed: Exotic materials like LCP are not always necessary. In many medical flex designs, standard polyimide can meet performance requirements at a lower cost and with better manufacturability.
- RA copper improves bend reliability, especially for dynamic flex: Rolled annealed copper is preferred for flex circuits due to its grain structure and improved bend performance. It is critical for dynamic applications and still beneficial for static bends when margins are tight.
- Bend radius rules vary for dynamic vs static flex designs: Dynamic flex designs follow much larger bend radius guidelines, while static bends allow tighter radii based on layer count. Understanding IPC-2223 guidance early prevents reliability issues later.
- Early manufacturer involvement helps simplify overengineered designs: High specs often drive unnecessary complexity through material callouts and via structures. Working with the fabricator early can uncover material equivalents, simplify vias, and reduce cost, risk, and lead time without sacrificing performance.
