Kickstarter Campaign Produces Large Affordable CNC Cutting Machine

They claim 1/64th "real world precision". It is supposed to be closed loop dc servo motors, which, if they close the loop back to the software (with the virtual encoders most servo drives produce), obviously means any positioning error is in the drive mechanism or pulse generation failure/signal error.

The encoder resolution sucks (8148 steps/rev), but since it's a fairly direct drive (i'm assuming 1 revolution is an inch for simplicity), you can lose a few steps on the generation side and not get too bad of results.

But the drive mechanism seems like it's impossibly sloppy. So i'm not sure i believe either the precision (i would have to believe the precision depends on speed), or that there is pretty much any repeatability.

I'm also assuming you are using no dust hose here, since the way they've done dust collection will deflect the x+y axis if there is any tension in the hose.

As for the rest you said, yeah, i'm with you. Getting good fitting joinery in woodworking is 0.005" or less (more than that and it's either too loose to fit well and will display very visible gaps).

Obviously, for cutting curves/etc, you can sand parts in woodworking (less easily in metal :P), but still .. if your box joints are slightly misaligned due to positioning error, it's a junk sheet.

While you are right that it will be nice for large crafts, i worry about two things: 1. The process of going from "drawing a thing on a napkin" to "cutting it out with this machine" is just never going to be that simple. Even experienced woodworkers using simple software like sketchup are nowhere near as fast at creating their design as they are just drawing it on paper. The reason we use computer programs is because they reduce error (IE i can see that i screwed up a calculation and the joints don't align). Or, in my case, because i'm nuts and when designing a high chair for a friend, i use the stress analysis module in solidworks or inventor to make sure it can support the weight ;)

Going from that to CAM (even 2.5D CAM like this thing supports), or figuring out what's wrong with the gcode, or ...

I think folks are seriously overestimating their desire to essentially debug a bunch of software just to cut a random thing. People spend tons of time, money, and energy getting to simple workflows.

That's, IMHO, pretty much not compatible with "want to do a one off thing occasionally"

In that realm, i think the shaper origin is a much better bet.

2. In the cabinet shop world, when the sheet comes out wrong they simply throw it out and do it again, because it only wastes 5 minutes and they are buying sheets in volume. Somehow, i don't think people really want to ruin that many pieces, either cost wise, or time wise. At 1/64th, it may ruin a lot of joints.

3. You can't nest which means, a lot of waste.

For those curious, here's what, for example, trying to make booth seats looks like:

http://www.woodweb.com/images_forums_public/cnc/Nest.png

The blue is all of what gets cut out.

The normal way this is done is vacuum hold down. The vacuum can pull vacuum despite the massive volume of emptiness there, and hold the parts (these are usually 20hp vacuums).

Since it will now be vertical, and there is no hold down, it's not possible to do anything close to this (without another mechanism, like screwing non-cut parts of each thing to a plywood sheet behind it, etc)

They are mostly CNC, but they have horizontal heads in addition to the vertical ones.

By that token, it will literally never happen that you can put horizontal boring heads on this thing.

So you'd have to settle for "ikea-like" furniture.

There your main problems are not CNC related at all. They use a lot of bent laminations, etc.

I don't think the CNC is really going to help you much!