Trial Wooden CNC Table

Trial CNC MachineThe design of this first wooden table was based on the basic moving gantry system.

It was built as an experiment with CNC. As such it was built as inexpensively as possible with shop scraps and framing lumber.
The rails were gas pipe, and skate bearing trucks were used on all axes.
The leadscrews were threaded rods, and the leadnuts were rod couplers.
Back of wooden gantryA plywood angle was made for the gantry.  Right image.
The gantry end plates were also sheathing plywood.

The machine used 420 oz.in. servos which were driven by Gecko 320s. These were to be recycled into a larger metal machine.
The leadscrew pulleys on the servos gave a 1:3 reduction (12 and 36 tooth)
This ratio did not take advantage of the servos' power. It should have been 1:2.

Those were the kinds of mistakes that justified the building of this trial-run machine.
Pipes attached to gantry endsPipe Supports

The rails were supported by rings cut with a hole saw.
Tee nuts with bolts in the rings permitted rail alignment.

This system did not incorporate rail supports; therefore, the longer axes flexed under heavier cutting loads.

...................

There are pictures of the different bearing systems on the bearings page.
Bearing on Z axisFor the short travel Z axis, angles with bearings were used.  Black and white photo on right.

This system can be difficult to tune because it gains precision by tensioning the bearings against the opposing rails, rather than against each bearing's neighbors on its own rail.

This bearing system loads-up longer unsupported rails and causes them to flex.

It became clear that for the shorter Z axis, it is better for the rails to move and for the bearings to be fixed.

The rails would add rigidity to the moving components, and the bearings could be tightly tensioned.
.................

The leadnuts were made of rod couplers as shown below.
Leadnut made of rod couplerThe coupler was cut almost all of the way through and tines were welded onto each side of the cut.
A bolt was threaded into one of the tines and it free-wheeled through the other.
Tightening the bolt bent the rod coupler and removed the play between the rod's and the coupler's threads. It was clunky but it seemed to work. However, this turned out to be a complete waste of time.

A rod coupler by itself can give backlash in the 0.003 range, and any adjusting of the system in the picture just added drag.

.....................
Cobbled limit switchLimit switch made of scrapThe images on the right are of limit switches made from old push button switches and the finest quality coat hanger. 

They served well though they were not consistent in stopping the table at the exact same spot.
Since they were not home switches, it did not matter as long as the axis stopped before slamming into the ends.
It only takes one runaway servo to make even the crudest switch worthwhile.
....................
Small carving in plastic
Here is an image of a small plastic sign the table made with a 0.03 inch bit chucked in a Dremel.
The quality of the cuts was better than expected. The lines are a bit wavy (not even visible in this image) but considering the workmanship and quality of components in this table, this detail was a pleasant surprise.
Problems with CNC software
However, heavier work was of low quality since everything could bend.
The image on the right shows two problems.
One is the wavy lines permitted by the machine's flex.
The other is a software problem that sometimes occurred when an old version of ACAD was used with Ace Converter.  Extra lines were added that ACAD did not show. Interestingly, these lines were visible in CorelDraw.   The joys of cobbling software.
..................
Building this machine showed that this popular design is inherently flawed for a number of reasons.
  • Unsupported rails do not give enough rigidity. Flexibility permits chatter.

  • The long distance between the X rails and the gantry gives the table a mechanical disadvantage that is only remedied with expensive components.
    This is discussed more on the Mechanics page.

  • The sides of the gantry prevent wide stock from being placed on the cutting surface. This makes it impossible to carve an address in a door, for example.

  • The leadscrew underneath the table gets in the way of supporting the table bed, so the bed cannot carry heavy stock.

  • The space under the machine is lost to other uses from storage to vacuum hold downs or fourth axis components.

  • The center leadscrew does not support the ends of the gantry. All racking has to be prevented by an overbuilt gantry and commercial bearings. Again, unsupported rails are hopeless unless they are quite large, and homemade bearings are unable to offer the required support.

There had to be a better way and the metal table was a result. The lessons learned from it were then incorporated into the prototypes that were built for the plans.

............................
The 25 x 37 machine in the plans weighs a fraction of this wooden clunker, yet it has better specs even though it uses lighter materials.

The rack/leadscrew machine weighs about the same as this trial table, but is a far more durable machine that is capable of much better work. Similar materials in a different layout make all the difference.