Precision Desktop Lathe
We designed and manufactured a precision desktop lathe able to cut with a <50-micron repeatability after surviving three "death tests": dropping the machine from waist height, hitting the spindle with a sledgehammer (where the workpiece is held), and standing on the cross-slide of the lathe while actuating its motion in both horizontal directions.
This included a wide variety of analysis, testing, design, and redesign, with further details available upon request. Below are two key demonstrations of performance: cutting, and stage motion. Further description introduces results from our error budget, influences of nonsystematic effects on error, and an added cooling system.
Error budget: We evaluated the deflection of the shaft using an HSM model in MATLAB to reach the target 50 µm cutting repeatability requirement.
Here, we investigated the final nonsystematic sources of error (not controlled by the design itself, and changing over time or cutting conditions). A SolidWorks thermal FEA analysis paired with thermal imaging (below) were used to quantify deflection from thermal influences.
Thermal study using infrared camera. We located the most heat generation from the motor itself, and found steel components to heat up relatively much more than aluminum.
After identifying thermal influences and excessive forces as possible limitations, we implemented a coolant system to evaluate how it influenced performance in both aspects. Ideally, it would cool and lubricate cutting.
With cooling, deflection would decrease 1 µm/°F in both y and z axes. With lubrication, deflection would decrease 13 µm in y and 7 µm in z per 1 lbf decrease in cutting forces.
In practice, it also resulted in smoother cuts.
Left: cutting without coolant
Right: cutting with coolant
We monitored the energy draw of the lathe during cutting to observe peak, variation, and range of the energy draw for a single cut.
Here, forces were around expectations.
The effect of coolant on cutting steel and aluminum was similar, and reduced the variation between cuts but did not decrease the mean value.
Coolant reduced surface roughness to 2/3 the value without coolant, with constant effect over large range of coolant application rate.