Aligning the Y-axis ways
5 Bears Home Homebrew CNC bench mill
As I had mentioned previously, there is more to these linear guides than simply bolting them all together and suddenly having a perfect, low-friction way. To illustrate, I will show what I did for the Y-axis. The X-axis, and the Z, will be mounted with similar care. Reference this print for a rough idea. The shoulders don't have to be exactly like this, but some method must be used to create a fixed (and positionally accurate, i.e. parallel to the table edge) master rail, and an adjustable slave rail set.

The Y-axis blocks are mounted to a 3/4" thich, accurately milled plate which will be eventually sandwiched (via dowel pins) to a nearly identical plate which will contain the X-axis blocks. In the picture below, the grid of holes between the blocks are to secure the Y-axis ballnut carrier. I drilled an excess in a repeating pattern so that I may alter the position of the ballnut relative to the plate if I so choose or need to do. The holes in the corners of the plate (6 total) are for the locating and securing shoulder bolts, which act as both a dowel pin and a SHCS.

To begin the exercise, the blocks were mounted on the rails, with the master side of the block oriented inwards. Each block has two side edges... only one is the master, or reference, edge, and that is the one with no printing on it. The unreferenced edge has the block's size and style printed on it, in this case, THK HSR25. In use, the master edge is clamped or otherwise pushed against a shoulder, and the blocks are then secured. Looking at the picture above, the left rail, viewed from the end, is the master, and is mounted firmly against the shoulder of the plate. Being the left rail, since the sandwich is upside down, we are looking at the rail to the right.
To keep everything clear in my own mind, I sloppily stamped "MASTER" on the Y-axis sandwich plate. It is hard to see, but the blocks are hard against a milled rim of ~2mm height. In this position, the 8mm SHCS which secure the blocks to the plate are alternately tightened so as to fix the master rail and block set firmly to the plate. All of this work is done with the rails in place, as they force the two blocks each side into linear truth.

The other set of blocks is tightened just enough so as to allow for movement under friction, within the limits of the bolt mounting holes. This is why it is important to provide clearance in the holes from the plate to the blocks. The hole to bolt clearance is fairly sloppy. We want to be able to shift the blocks a bit.

A sensitive dial indicator (mag mount) was set to display the parallelism between the two rails. The far rail is the master, whose blocks are firmly mounted against the plate's master datum edge. By moving the rail through the blocks, I can check a section perhaps 8" in length.
A "tenths" indicator is best! Tenths indicators are humbling beasts. What would be a minute ripple even with a 0.0005" indicator becomes massive needle flailing with a good tenths dial. Be sure the runout is repeatable before you begin banging on the slave rail... sometimes these indicators stick a bit.
The master rail was slid back and forth, and the divergence was noted; the "slave" blocks and rail, at the front, were then gently manipulated with a rubber mallet (LIGHT taps) to minimize runout. When the smoke cleared, I had .00025" divergence from parallel over 8"... this is pretty close to 0, and good enough for this mill. This doesn't mean the axis will be "off", it simply means we won't be loading the balls excessively as we move the axis.
In other words, if the rails diverge from parallel, the blocks have a certain ability to "absorb" error. As the axis moves, and the rails diverge, you can actually feel a bit of stiffening as the internal balls begin to load. I'd guess over 8" with these blocks, one could have upwards of .0015" of divergance from parallel with little ill effect. This applies also to other planes; hopefully, since the block seating areas on the plate were milled at the same time, with the same cutter, to identical heights, we have taken care of that.

Once the rails are clamped onto the base, the Y-axis plate will track perfectly. Here is the other end of the test, with the .00025" or so divergence.

The previously overbuilt reference edge for the base is secured with a pair of 6mm SHCS and T-nuts. The mill base is tipped on its side in this view.

And the mill base is ready to accept the aligned Y-axis rail system.