Flex PCB material selection (high flexibility)

[Yansi]

Hello, flex PCB newbie here:

I would like to get rid of connectors and wires in a product, by using a flex pcb. However, there are a lot of materials of different substrate thickness and types and I kind of do not know how to pick a proper one.

What I need is to get the most flexibility, or maximum amount of flex cycles respectively.  It is a product with a movable part operated by a user, that needs to be connected by several wires to it.

I understand that to obtain the most cycles of the flexible part, I have to keep maximum bending radius to keep the least stress on the PCB. What I currently do not know and can't find much info about is what type of flex PCB substrate is suitable for such repeated bending.

For example a local PCB fabhouse offers Dupont Pyralux AP 85xx series and AP 91xx series in different thicknesses (1 to 5 mil), with different plating thicknesses. Also some other material ThinFlex A20xx and A40xx series. (2 and 4 mil). That is what I have to go with.

I'd guess that for achieving the most flexibility and flex cycle count I should choose a rather thin material (1mil substrate?) and to also choose a thinner copper plating for minimum stressing of the materials.

Do you have any recommendations on what standard materials and thicknesses are for flexible links, what to be aware of, etc. ?

Thank you for any hints,
Y.

[T3sl4co1l]

I don't think it's too onerous.  I've seen pretty ordinary looking cable in printers before, either ~1oz and probably around those substrate thicknesses, or the embedded-flat-wire kind (the stuff that's rather substantial feeling, seems like the copper is thicker than the carrier; don't know if you can get PCB made that way, but they're commonly available as plain cables at least).

Some supporting material may help enforce the bending radius constraint.  For example, thicker but also more flexible tape, or foam, glued to the cable (one or both sides).  Don't forget strain relief, so the cable, with any stiffeners/springiness-ifiers, are anchored to mounts, keeping flex away from the connectors.

What kind of cycle life are you expecting?

I think you could calculate cycle life, based on the moduli of copper and polyimide (or whatever substrate), layer thicknesses, and the fatigue limit or curve of copper.

The copper foil won't be dead-soft, but; well, hm.  Raw foil might be annealed, but is more likely supplied half-hard or thereabouts?  Plated copper, isn't actually dead-soft (as you might think from the purity?), but contains internal stress due to plating and additives, and, is probably somewhere in the half to full hard range in terms of hardness, but is weaker than a rolled product of the same hardness.

Hah, wow, never thought of that before but it's amusing -- if the stackup is all polyimide and other high temperature resins, I bet the copper could actually be effectively annealed by heating, without compromising the substrate strength too much.  Polyimide is rated to crazy temperatures, right around where copper starts to anneal I think.

Anyway, a bit rambly, but should be good for insight and leads on what to look for.  Cheers,

Tim

[chickenHeadKnob]

I have a roll of  7 mil I can't remember how thick double sided 1 ounce copper on mylar which I purchased from an ebay surplus vendor who claimed it was much tougher than the typical polyimide substrate and was preferred for high flex strength durability in applications like battle-bots. Mylar has significantly lower melting point however so that is the trade off. I don't know where the seller sourced the stuff, it was more than 10 years ago, sorry.

[KL27x]

IME, the substrate is not what is going to break from the repeated bending. It's the traces. Yes, I think thinner the traces the better. Dunno how they make FPC jumpers, which you can bend in half no problem. But sadly FPC board doesn't act like that.

I don't know a reason to go crazy looking for a specific brand of polyimide.

Thinner substrate is better for less stress on the traces during bending.

So thin 1 mil substrate and thin half oz copper is what I choose for this, but I could be doing it wrong. Much of the thickness of the final stackup will be in the soldermask/coverlay. So I don't know what I'm doing, but that's how I do it. If you're counting votes.

[Yansi]

Hello,

Thanks all for any hints. I spoke with a tech from a local pcb fab, had a bit of talk about the design.  We agreed on the substrate thickness to be minimum (1 mil), with also least amount of copper thickness (0.5 oz).  Thin coverlays are also available (0.5mil if I've looked correctly).

I was also kind of assured it ain't be cheap, or not at least thaaat cheap.

I got also a suggestion to put copper traces on the outside part of the cycled bend in the flexing part, as to not compress the copper, rather the kapton, which has a bit of give, unlike the copper. However I forgot to asked about the hardness of the copper on the substrate, whether it is annealed or what.

Now back to the drawing board and design the final shape.

[KL27x]

I was also kind of assured it ain't be cheap, or not at least thaaat cheap.

Dunno where you live, but you will get bent over for FPC by most "local companies" unless you live in Shenzen, lol.

Unless you are under a time deadline, you are almost assuredly going to save a boatload of money by making your own mistakes and ordering from China, directly. After discussing my first FPC design with a lot of local companies, I found the curious coincidence that they all take roughly 3 weeks, but can't say exactly. 3 weeks is how long it takes for someone to order the thing from China. So I would say 99.99% of the companies I contacted take your Gerber and order it from China.

The "big" legit "FPC company" I found in my country (because they advertise FPC expertise) don't make FPC, either. They have experience and expertise in the design and in fabrication of FPC/rigid composite boards, and the like. I got the distinct impression they make money from big medical/government contracts.

[Yansi]

This should be a serious manufacturer, that indeed should have the technology to make them. They can do a lot of specialties (including Rogers, wild stack-ups, extreme copper weight - once they made a 2layer 0.5mm thick buried copper PCBs for a high power motor controller, etc)

But the are EXPENSIVE. Hence why I don't use them and avoid them for personal use. This time, company pays and time is ticking (they can't set up sensible deadlines, because mechanical guys are in command of the electronics design dept... ...)  so money are currently not the issue. Meeting deadlines with crazy ideas out of reality is.

[jbb]

I’m sure you’re on top of this already, but: use a single copper layer.
If you have stubborn EMI problems, there are special EMC layers which are very thin conductive films (2um vs 18um for half ounce copper).

In my limited experience, a major piece of the puzzle is managing the location of the stress. You need to look at your connectors, fasteners and any guides, slots etc. I had one design that failed after 10,000 cycles because the FPC was flexing against a cut edge of a rigid PCB. Adding a little bit of tape to control the bend location took that up to > 500,000 cycles. Try to engage with a mechanical engineer on this task because it’s actually a mechanical issue. Some clamps or tapes or (shudder) glue may be required.

If you desperately need a component or connector soldered onto the FPC, you don’t need to go full on rigid flex. You can get the FPC manufacturer to fit a stiffener on the back side (polyimide or FR4 is available).

Finally, you’ll need to work out how to manufacture it without putting a kink in the critical bending region. Kinks usually concentrate the stress and speed up the fatiguing.

[Yansi]

Yes, only single layer on the outside of the bend, that was also what I was told. If desperately needing ground on the other side, I was suggested to use just a fine mesh,  not a solid pour.

Only FPC, no flex-rigid - I think I haven't even mentioned that option. The main PCB in the device is just 2 layer job, not worth the trouble investing into space tech.  The FPC alone is expensive enough by itself.

So yes, I will also need to figure out that end with the FPC connection.  (Probably an 8pin 1mm pitch job there).

[wraper]

IME, the substrate is not what is going to break from the repeated bending. It's the traces. Yes, I think thinner the traces the better. Dunno how they make FPC jumpers, which you can bend in half no problem. But sadly FPC board doesn't act like that.

Actually it's usually opposite, at least when inspected under microscope it's rare to see no damage to substrate on faulty flex. Either substrate cracks starting from sides or gets some sharp deformations and only then traces give up. I would say that the most important is considering shearing forces applied to the cable as they are the most dangerous. When folding phones were popular, they had way more flex cable failures than sliders due to that. Bending with small radius is also extremely bad. I marked with red places most subjected to damage by shearing with folding type. IMHO if you make something where cable will go through a hinge, you should use thicker material as it should be less susceptible to shear. If you make sort of sliding stuff or cable will be straight, not going through a hinge, use thinner substrate as it can be bent safely with smaller radius.