Flexible-Rigid PCB Connection Methods

[epsilon0]

I'm looking for some insight from those more experienced in the industry on the different ways to connect a flexible circuit to a PCB.  I'm going to minimize the details here, because I'm more concerned with hearing different opinions/styles, rather than a solution to a specific problem.

Given a single-sided flexible circuit with a large number of positions (>256), how would you connect this to a PCB?

[RobK_NL]

I don't think any meaningful answer can be given with such meagre information.

One could think about using a couple of these things:


But 256 connections on a single sided flex ... 

[Siwastaja]

FFC. Possibly multiple FFCs, because AFAIK, >256 position FFC connectors do not exist (after 80, things start to get expensive). This has the added benefit that they can be connected to the circuit from different sides, possibly allowing layout optimizations compared to all 256 signals coming to a single point.

OTOH, if you need to put >256 signals through connectors to another subassembly, you most likely should think about building it in a different way, with increased integration so that more happens at fewer places. If there is some specific requirement which prevents this (such as abnormally high temperature where the signals are needed), it most likely also limits the choice of connectors, as well.

Flex circuits can be soldered in permanently, as well. Not sure what kind of equipment is used for this in industry, though. Tiny amount of solder paste and reflowing with hot air should work as long as you have a means to clamp the connections down. Some Kapton tape could work for this in small scale.

[jbb]

You can use standard flex PCB connectors (preferably a locking bar type rather than a friction fit) and split the flex PCB into multiple sub-tails (e.g. 50 at a time). It'll waste some space (because the connectors always have some dead space at the sides) and be annoying to assemble.

... if you need to put >256 signals through connectors to another subassembly, you most likely should think about building it in a different way, with increased integration so that more happens at fewer places.

Yes!!

Flex circuits can be soldered in permanently, as well. Not sure what kind of equipment is used for this in industry, though. Tiny amount of solder paste and reflowing with hot air should work as long as you have a means to clamp the connections down. Some Kapton tape could work for this in small scale.

I believe this is done with a hot bar soldering tool.  I understand that it's quite difficult to get one set up just right, and makes rework/repair really hard.

[Ian.M]

There's also anisotropic conductive adhesive tape.   

[JS]

Have you heard about serial busses this days? 

But 256 connections on a single sided flex ... 

I'd like to see the layout and assembly 

You can use standard flex PCB connectors (preferably a locking bar type rather than a friction fit) and split the flex PCB into multiple sub-tails (e.g. 50 at a time). It'll waste some space (because the connectors always have some dead space at the sides) and be annoying to assemble.

... if you need to put >256 signals through connectors to another subassembly, you most likely should think about building it in a different way, with increased integration so that more happens at fewer places.

Yes!!

What they said!
LCD panels have something to do with this thou.

Flex circuits can be soldered in permanently, as well. Not sure what kind of equipment is used for this in industry, though. Tiny amount of solder paste and reflowing with hot air should work as long as you have a means to clamp the connections down. Some Kapton tape could work for this in small scale.

I believe this is done with a hot bar soldering tool.  I understand that it's quite difficult to get one set up just right, and makes rework/repair really hard.

Yes, hot bar is used, factories equiped with them will do it, no worries. There are some attachments for soldering irons but I'm not sure to be picturing the size and density of 256+ contacts in a flex. Let's say 4mils density so it's not impossible to align but already a pain to do it, that's over 2" for this many contacts. Not impossible but hard for sure, first align and then keep even pressure and temp on all the contacts, quite a task for a hand made prototype, I guess not a problem for a real factory. Tear down a few LCDs and see how it's done, I guess the older the better but YMMV.

JS

[Heartbreaker]

There's also anisotropic conductive adhesive tape.   

As already mentioned by others - ACF bonding.  Making a 2D structure in stead of a 1D structure reduces the footprint size significantly.

An alternative solution could also be a flex-regid PCB.

/Sverre

[epsilon0]

Thank you all for your input and recommendations.

For what it's worth, this summed up exactly how I feel about this project as well.

But 256 connections on a single sided flex ... 

I'll provide some more details now to satisfy those who are interested.   I've attached an image of their current design as well.

I recently joined a team of researchers developing a flexible and transparent passive electrode array.  We are manufacturing the flexible circuit in house and the goal is to route this to multiple 36 pin connectors, which interface with our stimulation/recording controller.  I was recently asked to survey and compare other methods of connection, which is why I requested ideas rather than a specific solution.

My concerns with the current design are:

• Friction makes it nearly impossible to separate PCBs without bending the pins
• Creates holes in fragile substrate, could lead to tears
• Overly complex - 6 layer PCB, large amount of pins, routing is a nightmare
• Wire to PCB (left side) - The group is having trouble finding an assembler that can solder the 36 position, 30 AWG connectors to the PCB 

From your recommendations it sounds like these are the options (some of which I have looked into already).

• Split flex into multiple smaller FPC-able branches
• ACF bonding (they've tried this in the past, but were using a flat iron in house...)
• solder-paste and reflow
• Flex-rigid PCB
• Re-work the design

Again, I appreciate the input.  If these new details spur any additional ideas, I would be happy to hear them.

[CopperCone]

Has flat flex reliability changed in the last 15 years? Ifound that i particularly dislike flat flex connectors that yoj cannot disconnect.

I have seen situations in apc backup batteries where it seems that the solder joints failed even though there was some kind of silastic strain relief about an inch from the solder location. To me it seems that there is something funny about strain distrivution in a flat flex cable.

But i never saw a problem with flat flex that is used in connectors and i found that i actually, in well mechanically engineered situations prefer it to flat flex cables. I have had many issues with flat flex cable terminations but none withflat flex. However it is a complete dog to repair unlike ribbon cable due to the very high pin densities with flat flex.

The insertion force seems kind of ridiculous with ribbon cable and it causes a fair bit of board flex.. However i have seen ribbon cable like connectors that had very nice terminations, like sata for instance, which i am actually really impressed by..

Unless the enclosure is super engineered for being minaturized i think i would prefer sata like connectors connected to nice ribbon cables..

[JS]

  You can have chips soldered on to the flex, so you do some workout in the flex and then go out with less pins. The flex should be well suported where it have the chips, not ideal to be flexing around with the chips on top.

  I don't know what your electrodes are doing, what type of signal you are receiving, how conditioning is done, etc. so I can't tell if this is a reasonable approach. Maybe you could add a few chips on the flex and get out with 30 contacts instead of 300.

JS