Flexible PCBs; Any interest, and any tips? (NOT a discussion about specific PCB manufacturers, that will get reported, so no spam please : )

This is an interesting topic and some great questions.  Unfortunately, I don't have any answers given a lack of experience in this area, but I do look forward to hearing as I have found myself wondering about if flex boards might be needed somewhere in my future.

shabaz said:

Could these be useful for perhaps creating membrane-style keyboards?

Perhaps, but I'd be a bit nervous given the flat structure. Most membranes have a thick mylar sheet with holes or formed "indents" to increase separation. I'm not sure the mask will be enough in the long run, as the copper can elongate under lots of bending.

shabaz said:

For instance, are there particular components or connectors that lend themselves to such use, and are some components best avoided?

Master the use of stiffeners under flexible PCBs - they're basically like glued on bits of substrate that can keep sections of the flex flat and relatively rigid. This is where I would mount my components to - any component mounted above flex that has no stiffener will likely peel off given enough bending, so usually you'd use the flexy-part as an interconnect ribbon (similar to how an inkjet printer uses them to connect the print head to the mainboard). If you design the end right, you could even get it to fit into a flexible flat connector socket for an ordinary rigid PCB, simplifying connectivity, as soldering while possible is often finnicky and risky as overheating is much more damaging to flex.

shabaz said:

Has anyone tried soldering flex PCB directly to a FR4 PCB?

I have, it works as long as you're moderately skilled at soldering ... my students, however, make a mess of this.

shabaz said:

bent a lot (or even repeatedly)?

I can't be considered an expert, but one has to be really careful of the bending radius if they want to repeatedly bend. Folding is generally only good for a few times at the most or the copper will develop fatigue cracks. I tend to make the copper traces as thick as possible to reduce failure risk especially in fatigue, but I've seen very fine traces work okay.

Just think about the fact that hard disk head stacks are connected to the mainboard using a flexible printed circuit board and they have to last the lifetime of the disk in terms of seek operations (could be billions). I've never seen it fail - in part because there's enough slack. But those differential pairs going to the head are quite thin ...

- Gough

We do a lot of rigid/flex where the component areas are FR4 and the flex allows the rigid sections to be at different angles.

We have noticed a couple of manufacturers have dramatically reduced prices.

Large components on flex, especially BGAs probably should have stiffeners glued under the components on the other side of the flex.

Our flex PCBs usually have rigid frames to maintain dimensional accuracy during CCA component stuffing. (The frame adds to the cost)

I think it is recommended to have copper on only one side of flex where it is repeatedly flexed.

One recommendation I would offer is to use MLCCs with "flexible terminations." Different vendors name it differently, but it is generally a layer of silver epoxy that allows the terminations (solder points) to flex slightly. On rigid PCBs, this pliable layer helps resist cracking when the board flexes. (The dreaded "board flex" failure.)

Also, if you have a specific bend point in mind, I would recommend keeping MLCCs as far from it as possible. If you have to get close, keep the terminations in parallel with the radius so that both terminals (hopefully) see the same stresses.

Or you can do what automotive customers do: put two caps in series but turn one of them 90 degrees.

I think you're right, for a keyboard, the gap between the contacts would be way too thin with just solder resist. A layer of "something" more thicker in the sandwich is needed.

Regarding copper elongation over time in such a keyboard, I was thinking an option might be to use the flex PCB just as a way of electrical contact delivery and not for if's springiness, for instance, by having a circular cut-out around each pad for about 340 degrees or so, and then sticking a firmer thin sheet of plastic on top with the legends, so that the top sheet is being used for its springiness, rather than the flexible PCB. But that could cause its own issues, for instance, tearing at the traces over time since they would flex, maybe buckle a tiny amount each time. Maybe it's only feasible for non-critical stuff with a built-in lifetime.

The manufacturer doesn't offer the rigid portion capability on that service, I suspect that would cost a lot more, but I guess there are DIY techniques that could be used, e.g. combine the flex PCB with a separate thin FR4 board for the portions where the large components need to go. I guess that's the best of both worlds, combine a flexible PCB design with normal FR4, using the flex for all the more interconnect areas and maybe smaller components if needed!

Good point. I've hit this issue with normal FR4 and hot air tools too : ( To me 0402 decoupling capacitors (100nF or higher) seemed especially delicate compared to say 0603, and I've hit situations where the hot air didn't heat the component evenly when reflowing, and presumably micro-flexed the component, and ended up with shorted 0402 capacitors : ( After that I've also tried to avoid capacitors on regions of a board that may flex slightly.

Neat idea with the orientated dual capacitors for some protection!!

Some research via Google images..

The style using metal springy domes is quite common, and the top surface can be plastic with raised bumps. I don't want a clicky tact-switch style keypad, but if that was needed, then this looks like something that could be done with a rigid PCB too, although the benefit with the flexible PCB is that the connecting flex is integrated.

The spiral thing on the right doesn't use a dome I think. No info on any of these, all I could find were low-res images.

Next I looked at microwave control panels. Interesting how one layer has solid rectangle pads, and the other has the meandering pattern. I think the contacts are covered in conducting carbon.

Perhaps the meandering pattern allows for more pressure on the edges, to make a better electrical contact.

Here's another microwave panel, this time with a pattern on both sheets. I think the pattern is horizontal lines on one sheet, and vertical lines on the other, making it look like square hatches. Again this could be to increase the contact pressure at those interfacing points. The pads look like they may be silvered.

Old-fashioned home computer keyboard:

This membrane was covered with plastic keys on springs, and I think the larger circles are insulating deposits of rubbery ink (on one sheet, in the sandwich filling area), and the springs sat on top of that circular area. When the key was pressed, only the center portion was pressed. I could be wrong, I'm going by memory. The weird shaped cut-outs are where the base plastic had protrusions. I think they prevent the sheet from slipping or expanding too much across the whole length perhaps.

Back in time,  I did a collection of design and best practices articles.

hackaday.io/.../164461-flex-pcb-design-information

And probably the most impotant lesson I have learnt is, that a 2mil polyimide with single sided copper traces flex stackup is flexible. A 5mil with two-sided 2oz copper planes not so much. 

Hi Wolfgang,

That's brilliant.. I've downloaded all the documents to read! After all that, I think I will have decent confidence to move forward. I will probably create a test board with different features on it, to get to know the material, it will have various membrane button attempts on it too. It will be nice to get some experience with the material.

I have a sheet of double-sided copper-clad flexible PCB which I've occasionally been using small offcuts of without etching, and the one I have is not thick copper but does feel somewhat stiff due to the large copper areas, so I can only imagine that 2oz would have indeed been quite stiff.