Sorry for the lag.....gainful employment comes first.
Believe me, I am not in any way casting dispersions towards those of you who might see too many "warts" to this method or would rather deal with paint-on or vinyl masks and etchant but I believe this way shows a lot of promise, is quicker than any other way, not messy, fairly low-cost and still has some more improvements to come. More expensive cutters may do a better job but, for now, I'm focusing on the ones that target the cost-sensitive scrapbooking hobby market - it's just too good a tool to pass up for re-purposing.
As requested, I have uploaded a movie "cutting" a simple PCB.
http://www.youtube.com/watch?v=sQEncvJLrro&feature=youtube_gdataI also posted some microscope photos on my web page
http://research.microsoft.com/en-us/projects/papercutterpcbs/. So you can see (or verify for yourself if you have one of these machines) that this does, indeed, cut through copper clad. I continuity-checked this board and, before checking and correcting under a microscope, found 100% continuity and one short that was easily fixed.
Now for some nitty gritty:
I have posted (and will continue to post) the methods and equipment I use(d). As you can see by the zoomed-out picture of the board, the lines are somewhat wavy. I'm not sure what causes this or if it can be improved. It may be due to the mechanical integrity of the cutting head (or lack of) and drive belt (too much compliance?) or that additional cutter compensation is required. This problem limits the minimum trace width to something larger than a well-executed, chemically etched PCB can do.
I used 1/32" 0.5 oz copper clad PCB material available from wwww.digikey.com (and elsewhere). I will experiment later with the more common 1.0 oz but for now, the 0.5 oz will cut with a single pass of a not-new blade.
Note that if you get a clean mechanical and electrically isolated etch, there is no need to "weed" or remove the unwanted copper unless it's for high voltage or special RF work. Many times, extra ground plane can lower noise. You just have to remember to ground those islands.
Something I've been working on is front-to-back registration for double sided PCBs. The cutter I'm using has the ability to lay down registration marks which can be read by the cutter head. More later....
The knife can leave behind tiny shards of copper that will have to be removed with fine tweezers under an eye-loupe or cheap microscope. This is also generally required in the workflow of the desktop milling machine approach to mechanically etched PCBs. These shards will either cause immediate shorts or, worse, cause a short later on so you need to look at the etched PCB under magnification and remove these or correct any obvious problems with a knife or tweezers. If you still detect a short with an ohmmeter or an audible continuity tester (best) that you can't see, I use a ~10 uF capacitor charged up to around 30 volts and discharge across the offending circuit traces that will usually zap the offending shard while leaving the trace(s) alone. Experiment. I would suggest NOT scrubbing with ScotchBrite as you do with mill-etched PCBs as it may fold over the raised copper back into the cut trough, shorting the traces.
Some have asked about the life of the blade. My answer to that is to use 3rd party 45 degree, carbide blades (the vendor is also listed on my website). They're about $8.50 ea but are worth it. Consider the same material (single crystal carbide) that is in the "V" shaped milling bits used in the expensive mechanical etchers (TTech, LPKF, Accurate360) and spin at ~60,000rpm while the XY direction moves relatively slowly. The tip touches and is abraded by the glass in the FR-4 substrate in both methods. If you think about it, those spinning bits have a tangential contact length on the PCB material many orders of magnitude more than a single pass of the carbide blade in a CNC paper cutter. I do notice that the fine tip on my blade get broken off early in the cutting process but the blade still works and lasts quite long.
Something else I've noticed that may be a positive (but the jury's still out) is the "cup" formed by the furrowing blade that piles up copper on the edges of the cut. This cupping around pads to be soldered tends to reduce solder bridges as the solder wants to pool in the cup.
More to come if there's any interest.
Mike
[Non-text portions of this message have been removed]