• Welcome to engineeringclicks.com
• # Flex PCB design

Discussion in 'The main mechanical design forum' started by adih, Mar 20, 2013.

Joined:
Mar 2013
Posts:
5
0
Hi
I have to design a Rigid-Flex-Rigid PCB. There are many guidelines about bending radius of the flex. this issue (of the radius) is understood and obvious.

I checked about making each flex layer in a different length (book binder), and I've got from the manufactures that this is a very problematic and expensive process.

My problem is that if all flex layers are same length, the bending radius won't be as expected, especialy the important dimensions that I've attached in the figure (actual_bending.jpg).

https://skydrive.live.com/redir?resid=718E79F4BDE0B70!209&authkey=!AFEmaMzHCVjUnL4
(3 images: 1) original rigid-flex-rigid, 2) desirable bending form and 3) actual bending form with X and Y dimensions

1. how do I calculate the X and Y dimensions? (see actual_bend.jpg)
2. is the bending form I'll get is legit?

2.
3. ### Bob BurnsMember

Joined:
Mar 2013
Posts:
5
0
Hey Adih, I looked at your images. Are there 10 layers of circuitry going across the flex layers, or just five? Did you estimate the bend radius, of each of the curves, in the innermost kinked flex arm? The rule of thumb for single and double sided flex minimum bend radius guidelines is 6X the thickness of the flex. If those kinked curves are above 6X, you should be fine.

The other question, is this for a high reliability application? If the board is going to be in service for a long time, and cycling on and off, the expansion and contraction will create a dynamic flex environment, and will put stress on the outer most flex layer, on the circuits where they meet the rigid boards.

If it is 10 layers of circuits in the flex, and if it is a high reliability application (life critical), you may want to look at a bookbinder. If it is a commercial application, industrial, consumer, computer, etc. it should not be a big issue. You can have the outermost layer be a ground plane, so that it takes the most stress rather than individual circuits. If it is five layers of circuitry, you may wish to do a triplet and a doublet, with the doublet inside.

There are some ideas for you - Bob Burns

Joined:
Mar 2013
Posts:
5
0
Thanks Bob,
My application contains 5 leaves of dual layer flex, so 10 "electronic" layers but 5 "mechanical" layers. It for static industrial application.
when I calculated the flex length, I took it as x10 of its total thickness. But, as I showed, because all flex layers are same length, it won't bend as a full radius but as a kinked bend, as you called it. In this case, how can I estimate the "real life" radius and dimensions X and Y.
Just to be clear, the photo is a sample I've got from a different project, so I can't measure it. It is not the same. I must calculate especially the X and Y dimensions for the new project

5. ### Bob BurnsMember

Joined:
Mar 2013
Posts:
5
0

OK, so you have 5 flex leaves, each with two layers of circuitry.
The kinking you see is normal. If it is a static flex to install assembly for an industrial application, I would not be concerned about the kinking. It is to be expected whenever you have that much circuitry going across the flex layers, and splitting the layers into individual two layer flexes, gives the smallest bend radius possible. I would just try to make the outermost layer a plane layer if at all possible.

You asked how to calculate the X and Y dimensions so that the kinked flex radii don't fall below the minimum bend radius recommendations of 6X flex thickness (DuPont's and United States old Mil-Std-2118 recommendations for high reliability applications). I don't know how you would calculate that. I don't know how you would do it mathematically; and I don't know if CAD software would accomodate the multiple layer feature. Most CAD files I see lump them all together. Some people build mock up boards first, to see mechanically how they work, before building the actual circuit boards.

The 6X recommendation is for the individual flex layers, and not for all of them together, if that helps. The key is when you are done, the bend radii on the buckles not be less than the 6X minimum. There is an explanation of this in IPC 2223 5.2.3.1.

Also, you can have the flex go out into a shape like an incadescent light bulb, reducing the bend radii on the internal layers and reducing the mechanical stress on the outermost layers.

I hope that is some help Adih,

Bob Burns

6. ### Mark StapletonActive Member

Joined:
Jan 2012
Posts:
38
0
Adih, it's not clear (to me) from your description whether this flex circuit is actually going to actively flex in use or whether it only flexes when being installed and de-installed. If there is continuous flexing then of course there is a potential problem, but I wonder whether the flex layers can be arranged side-by-side along the length of the bend instead of one on top of another as your earlier application has it pictured? In other words, you may have 10 rigid board flex layers, and you need a flex circuit for every double layer, but I'm not sure that means that they have to be on top of one another. Why not BESIDE one another? I understand that this would necessitate some trace routing on the rigid layers which could be challenging (to say the least), but that would remove the problem with flexing entirely. At any rate, as the two rigid boards become laminated I'm not sure that you even CAN have different lengths among the five flex layers IF you have them on top of each other. How would it be possible to laminate those rigid boards? They would have to do it in two different processes on two different angles in order to get a reliable result, would they not?

Last edited: Mar 21, 2013
7. ### Bob BurnsMember

Joined:
Mar 2013
Posts:
5
0
Hey Mark, good points. If it is possible to separate the circuits so they are not all running parallel to one another, that would be good.

You can have different length flex arms. They are called bookbinder rigid flex boards. They are rare, not too many folks build them, and they are expensive, and you have to purchase custom tooling.

There is a picture of a bookbinder rigid flex shown on page 15 of a pdf, that is located here: http://www.printedcircuits.com/portfolio/PCi-Mil-Aero-Portfolio.pdf

Bob Burns

8. ### Michael RossWell-Known Member

Joined:
Mar 2012
Posts:
52
1
If this is your first flex PCB (or any other technology really), you should be in close contact with your provider. If they are worth working with, they will have good engineers or sale engineers that can help you make a really excellent component.

I guarantee you cannot do as good a job on your own, and not with even the excellent advice you will get in this forum.

9. ### Mark StapletonActive Member

Joined:
Jan 2012
Posts:
38
0
Bob, that's an awesome collection of flex circuit applications, and Adih should note that they're all military or avionics or both. This is why the flex circuits as justified (high reliability under rugged conditions including vibration and heat). There's a reason why few industrial applications use flex circuits -- high cost. In most instances if two boards are being connected at a 90 degree angle, as in Adih's illustration, it is accomplished via right angle header connectors. I'm not surprised that the "bookbinder" flex circuit would require specilized tooling. That alone will mean that you've either got to be working on a very high dollar application or you're going to have to expect to sell a large volume -- or both.

Joined:
Mar 2013
Posts:
5
0
Thanks Mark and Bob.
This flex only flexes when being installed and de-installed.
I cannot do the flex side by side. The reason for the flex is that I need to route about 600 wires in a small width. I have area problem so, I cannot use right-angle connector. It takes too much layout area. That is why I have 5 flex layers.
I'm less concerned about the radii or the reliability of the flex. My main concern is how I determine dimensions X and Y, for the right mechanical design and to eliminate stress on the flex

Joined:
Jan 2012
Posts:
38