Yes, you read that right. I reversed the entire gantry. "But you're giving up so much space"... "The router would just be hanging off the end of the table"... EXACTLY. It works perfectly along with the next modification:

        

I added a vertical clamping system at the far end of the X-axis (opposite the X-axis stepper) so that the ends of workpieces could be cut and the length of the piece could be virtually unlimited. The clamping system and the table mounts allow for pieces to hang vertically up to six feet and extend horizontally over ten feet (longer if I open the door to my shop). This allows me to cut proper dovetail joints, box joints, mortises and tenons with ease. The clamping surface is parallel with the YZ plane and has a built-in alignment stop at the far end of the Y-axis that is parallel with the XZ plane. This gives me a known X and Y coordinate for the workpiece, so that I can just touch the Z and go. The details are shown in the clamping system section.

Unfortunately this isn't an option for everyone, but I happened to have a 3'x6'x5' concrete ledge in the corner of my shop. The machine fits perfectly here with room for a dust collection system behind it. I purchased some brackets from 80/20's ebay store and used some tapcon's to secure the machine to the ledge. I used rubber backed washers on top and some rubber gaskets between the brackets and the concrete to help dampen the machine. The end result is an extremely strong and stable structure. It wasn't very easy to get it up there, and it's a bit more difficult to access, but it's out of the way and I never could have used the space this effectively otherwise.

The kit that I used included a low-profile support bushing for the Z axis and has since been eliminated from the revision 5 plans. I wanted the machine to have as much clearance as possible, since my Bosch Colt spindle limited the vertical reach. By removing the low-profile support and lowering the Z axis rail to the bottom of the extrusion, I extended my reach without compromising my clearance. I haven't had any vibration or screw whip and the Z carriage itself stabilizes the screw.

When I first assembled the machine I used the helical couplers that came with the kit. One of the first few tests that I did with the machine was to test for backlash using a "wiggle" program. This involved having each axis move back a forth relatively quickly over a very small distance (a few thousandths). I wasn't very thrilled with the results, but when I investigated the cause I discovered it wasn't the lead screw/nut, it was the couplings. During the "wiggle" test the helical couplings were just flexing like springs. So the steppers were wiggling, but the leadscrews were staying still. When the machine was off, I could easily turn the screws by hand, so I knew it wasn't excessive friction. However, the amount of static friction on the screws was enough to cause the couplings to flex during small movements. If I moved a larger amount, the tension in the "spring" would exceed the static friction and the screw would align itself properly. this type of performance was unacceptable to me, so I looked for an alternative coupling. I settled on the zero-backlash jaw/spider couplings from Ruland. They look like the Lovejoy couplings that used to come with the kit, but they perform MUCH better. The spiders are urethane and are under pressure when assembled. They also have clamp connections rather than the set screws of the lovejoy's. Switching to the new couplings completely eliminated the backlash problem and also helped dampen the motor vibration which should be easier on the steppers and the frame. Before I even put them on the machine I did a simple test that confirmed my suspicions: hold each coupling with both hands and twist each end in opposite directions. I can get the helical coupling to flex several degrees, but the spider couplings don't show ANY visible flexing.

I discovered fairly early in the assembly process that the ACME nuts are VERY sensitive to misalignment. Not so much that they won't work, but enough that it can increase the friction on an axis by an order of magnitude. The key is to let them align themselves before tightening. On the X-axis this is straightforward:

I didn't see this mentioned anywhere else, so I thought I would mention it here. Obviously assembling the kit to be square and level is extremely important, but the router/spindle needs to be aligned properly also. Since I use my machine for planing wood, if the bit is even slightly crooked it will cause ridges in the finished surface. I found that even though my router mount was square, the frame of the router was not parallel with the router's axis, so it need to be shimmed anyways. So how to you know how much to shim with a 3/4" bit as your reference? At first I did the "guess and check" method by doing parallel passes on wood or plastic and feeling the ridges. It was immediately clear that this would not be very effective or accurate. What I needed was a larger bit that would exaggerate any alignment issues:

So I made a 2' bit out of some 1/4" aluminum rod. I mounted it in the router and spun it around BY HAND. Do not turn the router on when this extremely large, unbalanced weapon is in the collet. It will hurt you and probably the router too. As I spun the bit around manually I checked the clearance in all directions and made adjustments. As I got closer to square, I would lower the Z-axis until it almost touched. Then to make the final adjustments I would listen for areas where it would rub. Keep in mind that you'll need to surface your table first since it will be used as the reference.