I am a believer that setbacks are not failures as long as you learn from your mistakes. By reviewing the mistakes I made and endeavoring to learn from them I hope to put my first attempt at a CNC machine under the "learning from experience" column.
There are four major points that lead to the failure of this machine. I will address them individually and propose simple solutions for their remediation.
Issue- The reason this project was scrapped was because the base of the machine had sagged over its long (4') unsupported length. Without a flat base as a reference point there was no chance to have an accurate machine. This flaw would not have been as severe had the base been constructed using plywood but I had used MDF which I did not have much experience using. Its real drawback is a low specific stiffness (stiffness in relation to weight). I had a 3/4" sheet over a 4' span with the 3/4" spacer blocks glued on top not providing any stiffness. The design had full 3/4" sheet on top of this which would have increased the thickness from 3/4" to 1 1/2". The top sheet was not add as I was putting this off until the entire machine was complete to prevent it from being damaged
Solution- MDF can still be used as a construction material (and is in other designs) if used wisely. MDF should not be used as single sheets but laminated up into engineering sections such as I-beams, L-sections and C-channels. These shapes maximize the amount of material away from the neutral axis where it is more effective. My new design has a base fully supported by the surface the table sits on and a gantry with a L-section backbone.
2) Bearings over Bushings
To try and make things as simple as possible I used bushings on all axes. I knew my design decision would lead to more friction than a rolling interface but I underestimated how high the friction would be and the binding issues. To eliminate play on the axis four bushings were required. When one bushing is used there is a small amount of play allowing for it to run smoothly. To eliminate the play I had all the bushings pressed against the rod. Imperfections with this design caused the bushings to bind and friction to skyrocket.
The solution for this problem proved to be far too easy. Simple bearing blocks as used on the Joe's CNC can be easily made from Aluminum angle, skate bearings and 5/16" bolts. These allow the bearings to pressed against the linear rail eliminating play while maintaining an extremely smooth linear slide. (Bearings on my new design allow the gantry to roll when the base is tilted by 1/8" over a 3' length)
3) Drive mechanism-
I began by trying to use 3/4" threaded rod as a drive rod for the X and Y axes. Just picking up these massive things makes me shake my head at how silly this way. They proved too heavy to even rotate as they had too much inertia to overcome due to their significant weight. My way ahead was to use 1/4" threaded rod. This way very effective on the Z axis where the span as only 8". For the Y and X axes the rod would whip violently when spinning at a reasonable RPM.
Solution- The 1/4" threaded rod remained an effective solution with very high resolution for the Z axis. The X and Y axes require another drive method. Lead screws are the best option but far outside my budget. This left timing belts and roller chain as two equally suitable alternatives.
Issue- My CNC did not incorporate any method to compensate for slight imperfections in construction for alignment. The holes located the bushings in one set location whether the guide rails matched this position or not.
Things were not all bad with the initial design. The Z axis, driver board and limit switches all works as designed.