Just a repost of my previously posted fabmoment. 
For a project at the V2_lab we had to develop a transparent touchpad in textile. For this we needed to create circuit boards that are thin, small, easy to mount on a textile substrate and more or less flexible. To create such a circuit board we started to experiment with cutting copper foil on the CAMM-1 at Protospace fablab. 
Design
The cutting of circuits in copper foil turns out to be very doable but some extra care when designing the board is needed.
No vias
This speaks for itself. A work around is to use WireWrap wire to make wire bridges.
Corners should be rounded
Copper foil does not stretch as the vinyl does. Because of this if the corners are too sharp it is possible the rotation of the knife lifts the copper from its backing. This might cause problems if you have dense packaging and/or small traces.
Minimal 0.6 mm space (degrades quickly when blade gets blunt)
After some experimentation we found that the minimal space between cuts is ~0.7 to 0.6 mm. This can be lifted a bit as long as the amount of cuts is limited and cutting speed is very slow. This also depends on the foil brand, sharpness of the knife, knife force etc.. It is good practice to cut a test before cutting the rest to fine tune the settings. 
On above picture it is very clear what happens if spacing is too small.
Don’t do tight packaging
If the circuit board needs to be flexible don’t package the components too tightly. The solder joints are the most susceptible to mechanical stress. If the traces are longer they can take much of the stress away that would otherwise work on the joints.
Component sizes
As far as component sizes concern anything smaller than 0805 and SOIC would be pushing it. Even if it can be cut soldering it will be practically impossible as the traces are able to move. We used 1205 packages to be on the safe side; there is enough space for a bit of movement. We did our board design in CadSoft Eagle and exported as a bitmap with all layers turned off except Top, Pads en tStop. 
Board design The bitmap was then imported into Adobe Illustrator, scaled to its actual size (this has to be done very accurately) and paths drawn over the traces. It is now very easy to adjust the sizes, round the cornes and do some more tweaking that might need to be done. Only keep in mind that the pad distances stay the same and are accurately aligned. The last step is to outline all paths (Object -> Path -> Outline stroke) and merge all shapes that form a single path using the Pathfinder’s ‘add to shape area’ tool. Inversely to create holes use the ‘subtract from shape area’ tool. 
Board outline ready to be cut.
Cutting
For cutting the circuit we use Cutronic copper foil (http://fablab.waag.org/content/techsoft). This is an A4 size copper foil with an adhesive backlayer. It is specially made for this type of usage (although they don’t advise to do SMT with it 😉 ). Cutting the foil needs some different settings of the CAMM-1.
Pen Force
Should be set somewhere between 60 to 80 gf. This also depends on the shape the knife is in and the type of foil used. Experimet with the build in test pattern before proceeding. Cuts should be well through the foil but only make a little schratch in the backing. If it cuts through the backing force is way too high. If it does not scratch the backing sometimes it won’t cut through the foil meaning more rework afterwards.
Speed
This should be set to quite a low value, 1cm/s is the slowest and recommended for complicated designs with tight packing. It also gives some time to push back little pieces that came loose with some fine tweezers.
Blade
The amount the blade should stick out depends on the foil used. For the Cutronic foil somewhere near ~1mm gives good results. It is recommended to use the original Roland blades as they are expensive but harder than the cheaper blades and will last longer. The blade originally supplied with the CAMM-1, the ZEC-A1005, is quite suitable for cutting copper foil. See the attached pdf for an overview of CAMM-1 blades. 
A cutout circuit board, left ready to be soldered (after tidying the little corner on the left), right as it comes out of the CAMM-1.
Soldering
First thing to do is peel of the excess copper so only the circuit itself is on the backing paper. Soldering can be done while the circuit is still on the paper, it will get a bit yellowish but won’t burn or melt. One thing to watch out for is the glue loosing adhesive strength when heated (it will stick like before when cooled down). This can make soldering a bit tricky. Loose pads will try to stick onto the soldering iron so great care and thinking about soldering order is needed. Also a set of very fine (SMT) tweezers which are anti-magnetic is a must. When all that is kept in mind, with a bit of practice, soldering on the foil is surprisingly easy. 
Finished board (black blobs are carbonized glue, grey wires are conductive threads) Some components which have very tiny leads (like SOD diodes) will have to be reinforced as the leads will break from the package under stress. We used some drops of polyurethane glue to do the job (the blobs on the left and right in above picture). Only use small drops as it will expand. With a little penknife, tweezers and care it is fairly easy to remove the circuit as a whole from its backing. Carefully slide the knife under the circuit bit by bit while holding it with the tweezers until the whole circuit comes off. The circuit can now be stuck on something else; textile in our case (see for example our first flexible circuit on felt test)
End result
The end result was a transparent flexible touchpad. The circuits have been connected using conductive thread and carbonized glue. 
Almost finished touchpad Due to the limitations mentioned above we could not cut the A/D board as the A/D chip has a TSOP-16 outline which is too small to cut. As a compromise we etched it on a very thin epoxy board.
