Z-axis Rail Mounting

5 Bears Home Homebrew CNC bench mill

The new NSK ballscrew was a success. It is now contained in a modular "pan", which fits snugly as shown in this drawing, between the two side plates. There is roughly 0.050" of clearance total between the ballscrew plate and the side plates... this will allow for some lateral repositioning during the final installation.

Any time you are building up a precision motion assembly, you need to allow for minute repositionings of the components during installation. No matter how careful you are measuring and planning, ultimately small errors will crop up, and without the ability to shift ball nuts, screws, rails, etc, the linear motion assemblies can and will load severely and prevent smooth and accurate operation of the axis.

My next step was to mill and tap the side plates for the NSK 25mm linear rails. Rail-to-rail spacing is exactly 120mm, and leaves plenty of room for the Z-axis ballscrew assembly. Also shown in the drawing is the spindle backplate, which forms the backbone of the spindle axis. Not shown is any portion of the Z-axis ballscrew assembly other than the ballscrew base plate, shown here as the semi-shaded portion between the rails.




From our previous installment, a modular ballscrew, nut, and plate. It will be mounted via t-nuts to the 90mm x 90mm vertical column, between the two heavy sideplates.
The Z-axis side plates are composed of 1" thick 6061 flats, each with 8 large steel T-nuts. Once attached, the mass of the vertical column easily trebles, and the rigidity goes way up as well. These had been machined long ago, and no photos were taken. The machining of these large plates was a real pain, as the surface which mates with the 90 x 90 extrusion had to be truly planar, and getting such a large surface into this happy condition is not the easiest task. The other three sides were not milled; rather, they were sanded with an orbital sander using 400 grit silicon carbide paper. This leaves a pleasing (but inaccurate) finish. Try it!

A very important aspect of this entire project is the ability to mount and machine larger pieces on my full-sized mill. Early on, I realized that I could not do a continuous milling longer than about 22", and the side plates were designed with this limitation in mind. The overall column height is 800mm, or 31", while the Z-axis NSK rails are 18" long. With 22" available X travel on my big mill, I can easily true both side plates for mounting of the rails.

Here, a .0001" DT indicator is used to position the Z-axis column + side plates in truth along the big mill's X-axis. I wanted it to be as close to 0 as possible. This was a pretty tedious operation due to the length of the setup. Due to this long length, it was impossible to get a true 0 TIR over the whole length, so I merely minimized runout, and probably got it to within 0.0004" over 22 inches.

Note the size of the CNC mill's Z-axis column relative to the 8" X 36" table of my ENCO mill. I had to remove the handle from the far end of the big mill's X-axis table, as the overhang of this assembly was interfering with handle rotation. A 3/4" carbide end mill will do the honors of cutting the NSK rail slots into the side plates. Carbide cutters are useful not only for the longevity of their edge, but especially for the stiffness they exhibit, and in general will produce a more accurate surface than a cutter of high-speed steel.

The mission here is to generate two flat areas, shown in the drawing in red, that are perfectly parallel to the back of the column, and perfectly coplanar with each other. This latter is achieved by using a locked quill for the final cuts... do not unlock the wuill depth when transitioning from one rail to the other.

Once the rails are secured to these milled edges, the truth of the Z-axis will rely upon the accuracy of this operation.

The master rail slot is milled first. On the master side, the rail will be snugged hard against a rim, while on the other side, roughly 0.010" clearance will be milled so that the slave rail will not be registered against a rim; rather, the parallelism of the rails will be generated, referencing the master rail, with an indicator. See this page if this is confusing.

As I mentioned in the opening of this installment, there must be a means of adjusting the installation of critical linear motion components during final assembly. If both NSK rails were "snugged" against registers in the sideplates, normal milling runout during this cut would cause an inaccurate mounting of the rails.

Here, I am checking the Y-axis depth of the cut which will create the master register rim on the sideplate. This rim for the master rail will be 3.5mm in width. I have a ways to go yet.

  Once the master rail side is complete (here, the sideplate to the left in the photo), I shift my attention to the slave rail side. The quill depth of the big mill (doing the cutting here) is not altered. I simply plow through the master rim to get to the other side plate. You can see the path that the cutter took in the very lower left corner of this photo.

Certainly, I could have raised the quill, moved the table, and lowered it again, but this way I have a guaranteed planar cut between the two sideplates; the rails will be at the exact same distance relative to the column backside.

The slave side is easier to mill, as the rail need not snug against a rim. Clearance is intentional. One pass is all it takes to mill the slave side plate rail mount.

Lots of swarf.

With the ugly stuff done, I get to drill the holes for the rail hold-down bolts, spaced 60mm apart. The holes in the rails are designed for 6mm bolts, but 1/4" work fine and I have more tooling for that hardware, so these were drilled and tapped for 1/4" X 20 SHCS.

The chuck is an Albrecht, purchased on ebay, quite cheap. I have never had an Albrecht chuck before this one, and wouldn't pay the price for a new one, but I realize now why they are so expensive... they are accurate, exceptional chucks.

Note the drill which is screw-machine length. With this stubby drill, I can forego any spot or center drilling to get the hole started true, which really speeds up the process due to the large number of holes to drill and tap.

  The tapping is done by hand, and gets very tiresome, very quickly. The old toothbrush is a handy tool to get the swarf out of the tap flutes between holes.

I need a tapping head!

(Postscript: I have begun to make use of my mill under power to tap holes 1/4" and larger. To do this, make the tap hole significantly deeper than necessary. Set the spindle speed to its lowest setting. Zero the quill depth DRO with the first tap contact with the hole. Under power, lower the tap and allow it to register and begin to self-feed. Turn off the spindle at roughly 8 turns, or 1/2 of the hole depth, whichever occurs first. We are not tapping to full depth, just starting the tap, but it does start true, and greatly simplifies the subsequent tapping job. Please note, this is not a novice exercise. Don't do this without some experience and knowledge. It is very easy to snap a tap!)

The rails are mounted in a procedure very similar to that performed for the Y-axis. The Master rail is first snugged into place, registering against the rim. Machinists clamps are put to good use to force registration against the rim, and the SHCS are sequentially tightened down. The milled rim forces the rail along a true line.

The slave rail goes on next, and the SHCS are installed, but only tightly enough to still allow shifting of the slave rail. A .0001" indicator is installed with a mag mount on the master block, and the slave rail is indicated parallel to the master. Surprisingly, the rails are not straight. They easily vary up to 0.0004" or so along 18". To produce a parallel set of rails, proceed as follows.

First, the middle of the slave rail is located by eye so that its cap screw is centered in the rail's hole, and only this cap screw is tightened to any degree. Use the indicator to measure the position of the rail at this location. The block which carries the indicator's mag mount, located on the master rail, is then tracked to the next hole, where the cap screw there is gently tightened. Any deviation of the indicator from the first hole is corrected with a rubber mallet with gentle taps, with the cap screw tightened enough to allow adjustment, but not so loose as to allow the rail to spring back afterwards. With an indicator reading identical to the previous hole, cap screw #2 is snugged a bit firmer. The process continues from hole to hole until the indicator runout is close to 0 along the full travel of the block. At this point, sequentially tighten the cap screws firmly. When all is complete, do one final check of parallelism with the DTI. Correct any deviation if need be by loosening the errant cap screw and perhaps its neighbors, manipulating the rail, and tightening once more.

It sounds tedious, but it's actually a pretty fascinating process and fun to do.