Backlash Elimination in Precision Linear Axis Bearing
This is my Precision Linear Axis with a preloaded slider for backlash elimination.
A couple of weeks ago, I attempted to actuate the precision linear axis that I had built, using a self-backing leadscrew and nut that I had found. Unfortunately, I did not succeed and ended by breaking the slider. (Refer Week 4 Hardware)
I blamed this on the lack of proper support for the leadscrews at the two ends. Over the past week, I worked on re-attaching the actuator to the slider as well as preloading the slider to eliminate backlash.
The original preciison linear axis that I built in Week 3 consisted of a box rail and slider of length 11inches each. This was done to match the length of the slider determined from my error apportionment at the time. Thus, there wasn't any distance the slider could travel when fully encased in the box rail. In order to allow for actuation of the slider when fully encased in the rail, I decided to reduce the length of the slider by half.
I knew that halving the length of the slider would reduce its repeatability, so I measured this once again after cutting the slider in half. I made sure to secure the laser pointer to the slider using hot glue. I also clamped the slider to a table.
Overall, I achieved an angular spread of 3.44deg.
Next I was on the lookout for a rubbery material of low enough stiffness that I could use to preload the slider. I first considered cutting out a strip from my rubber flip flop, but luckily I managed to find some polyethylene foam. Unable to locate the exact type of foam, I couldn't look up the compression online to determine the stiffness. So I conducted a simple experiment by adding a know load to the strip I would use and measured the compression. This was not a very accurate test. I estimated the stiffness to be 20N/mm.
I then cutout a thin strip of plastic (0.019") and glued it to the top of the strip using epoxy. This was done to reduce the coefficient of kinetic friction at the interface. The plastic being a lot stiffer than the foam would ensure that the preload forces is a result of the foam compressing only. I then glued the strip to the slider, again with epoxy.
Last week, I determined that the preload displacement for the compliant material to be 1mm. The width of the box rail is approximately 84.15mm. Using the table saw, I shaved off the width of the slider until the overall width was about 85mm. (During this process, I accidentally shave off more than required but I compensated by increasing the thickness of the foam) This gave me preload compression of 0.85mm. (Assuming the inner faces of the rail are perfectly parallel which they are not).
The preload force was thus 17N. If the slider was to be used for my rising desk, this would have corresponded to approximately 6% of the actuation force.
Before, inserting the slider into the rail, I waxed all the inner faces of the rail to reduce friction. ( Coefficient of kinetic friction for wood on wood is 0.15 with lubrication and 0.25-0.40 without!). The slider was able to rather smoothly travel inside the rail without significant force! I was very pleased with the result.
Following this, I was ready to install the actuator mount as well as support members for the ends of the lead screw.
For the actuator mount, I drilled a 0.625" hole through the centre of a small wooden block. I then pressed the nut through and secured the nut to the block using 3 wood screws.
I initially wanted to support the lead screw at both ends, ie with 2 bearings at one end spaced ~3D apart and a 3rd bearing at the other end of the shaft. However, I wanted to keep the slider open at one end to be able to swap the preloaded slider with a regular one to get a feel for the difference in motion. Hence, I decided to support the shaft only at one end.
I managed to find some skateboard bearings with 8mm inner diameters and turned one end of the shaft down to 7.97mm for a nice sliding fit tolerance. I then carefully drilled a hole for the bearing on block of wood and gued this to the rail while ensuring that the lead screw was perfectly level.
Finally, I performed a repeatability test to gauge whether my method to eliminate backlash was effective.
Disclaimer: I was unable to hot glue the laser pointer to the slider as I couldn't find the hot glue gun in the Hobby Shop.
My angular spread for the experiment was 1.50deg. This is nearly half the spread obtained before preloading the slider! Although this result is good, I believe it can be better. Hence I plan of re-doing this test with securely fixing the laser pointer to the slider.
Overall, I successfully designed and fabricated, using deterministic design principles, a preloaded slider in a precision linear axis system, to eliminate backlash!
With regards to eliminating backlash from the actuator itself, I feel gravity itself would eliminate this, given the wight of the desk!