FR4 flex? (2023 Update)

No, but I have done 3-layer boards.

I did have a 2 sided board 0.4mm that we used as poor-man's flex but it was surprisingly rigid and had to be heated with a hot air gun to get it bendy enough to adopt desired shape, once cool it retained shape without tendency to straighten. I suspect it will not be conformable enough for your application but is cheaper to try than flex.

piglet

Good wisdom, thanks.

Cheers

Phil Hobbs

not tried to use in real application, but seen and bent in my hands:

cheers Michal

A 3-d curved surface isn't ideal for any transverse-stiff material (it has to stretch or shrink) in a thin plate. You can relieve the stretch/shrink stresses by a circumferential kerf cut (think make-a-spiral) or some variant.

Conductive elastomers are an unlovely, but possible, variant circuit technology.

I have this vision of a matrix of spring-loaded pogo pins.

I'm not sure whether to be worried by that :)

Thanks. Dunno if we can make do with 2 layers, but maybe.

Cheers

Phil Hobbs

Thanks.

The board's aspect ratio is 5:1, and we expect to cut it up into paddles joined together with long skinny 'cables'. So it's basically a 1-D problem. It ought to bend fine--my worry is stuff like traces cracking or progressive damage to the glass fibres.

The current plan is to use L2 and L3 for the 'cable' wiring to reduce metal fatigue compared with L1 and L4.

Using Piglet's suggestion for hot forming the average curve might be a win--we obviously need to accommodate some range of curvatures, but they should all be convex at least.

We need to solder actual circuitry on them though--several photodetector/TIA pairs, as I mentioned, including largish through-hole photodiodes. So the choices seem to be

1. Serious 4-layer flex circuit$$
2. Rigid-flex circuit$$$$$
3. Semi-flexible FR4.

So #3 looks like a good option if it can be made to work reliably.

Cheers

Phil Hobbs

I might do my first rigid-flex. I'll have a metal block with an e/o modulator bolted to the top and a heater on the bottom. I'm thinking about machining a pit in the block and putting a little pcb in there, pushing against the bottom of the eom, with a thermistor wheatstone bridge. A shallow trench could let the flex get out to the outside world and not conduct much heat. The other end of the flex could run to a connector on my control board.

I wonder if an FFC jumper could be soldered directly to a board somehow.

Probably a hassle.

We use some LEDs that surface mount on the top of a board but shoot light down, through a hole to the bottom. That backlights a front panel.

Will you need to sterilize the rig between uses? Throw it away?

The real one will be potted in flexible black silicone with clear silicone for windows. We imagine using a peelable layer for asepsis, but it might be possible to get it cheap enough to be disposable.

Cheers

Phil Hobbs

There is more than one kind of flex (either as part of rigid-flex or alone). If you need to survive dynamic flexing there are rules to follow and things to specify or it will die early.

It isn't a printhead application--the bending will be quite gentle and relatively infrequent.

Any references for the flex rules?

Cheers

Phil Hobbs

It's a bed-mails-test fixture. Most of the boards made at Cambridge Instruments got one, and the production engineers, the printed circuit layout drafts-people and the design engineers would haggle about the little circles of copper that the pogo-pins stuck into. Production wanted each node to have it's test point, and there were nodes where the extra capacitance was lethal.

Some of the pogo pins had quite a long throw.

The Hewlett-Packard X-Y recorder used flexible cables to accommodate the excursions of printing head along the X and Y axes.

The write-up in the Hewlett-Packard Journal laid some emphasis on keeping the flexible cables symmetrical, and putting the metal in the middle to get million excursion life-times.

Not if HP was right. But you aren't going to flex the circuit anything like as often as the X-Y recorder did.

If. There are other printed circuit materials - we used a layer of isocynate resin-bonded Teflon cloth at one point - purely for it's lower dispersion - but it would have been softer than epoxy glass.

Search for design guides and maybe include "dynamic flex" about how copper should be arranged. Multiple layers are much worse than two double layer flex circuits on top of each other. Also, probably obviously, take care where flex emerges from rigid to prevent bending at the joint. The suppliers can use a better form of copper too (may not be necessary in your case). Did a rigid-flex design a couple years ago for a space instrument that had to work in a vacuum- some of the sensors were gimballed so we had to take all that stuff into account.

I've got samples of minimal thickness FR4, it's okay for small bends, maybe. I imagine mounting parts on flex might involve some additional tooling.

Nope, but a crazy idea, can you cut some slots and such in FR4 such as to make it more bendable at two or three places? Do you need bendiness on more than one axis?

George H.

Around 2003, we made these XFP transceivers:

< >

ROSA and TOSA are attached with a flexible Kapton tape. The metal bar behind the solder location was absolutely necessary to make it work.

I don't think that FR4 might work. Epoxy does not like to bend nor does glass fiber around a tight corner. I don't think that Kapton costs too much, they use it in mass market things like printers and there even seem to be pre-fabricated "cables".

BTW, these ROSA/TOSA tapes were a major 3D electromagnetics simulation nightmare. Endless HFSS and ADS runs. The hot signals and GND had to flip sides somewhere, at 10 GBPS without damaging the eye diagrams.

I had some bad mechanical experience recently with Rogers TMM-6. Bends easily and then breaks suddenly. It feels somehow like a rubber eraser. Then we made a multilayer with TMM6 + FR-4. The board house swapped the sides. The uwave synthesizer and the microstrips ended up on FR-4, the DC stuff on TMM-6. :-((

Gerhard

We did some boards with mixed lams, microwave for the top layer and FR4 for the rest. They were curled like potato chips.

I'm ordering some fast boards now with Isola called out for the top dielectric and FR4 optional for the rest, but with a flatness requirement.