|I started out by creating a material model that could represent crushing / grinding of a coffee bean, based on this paper.|
Then I did an explicit dynamics simulation of a few beans entering the grinder. This showed that the inlet geometry wasn't good enough, and I improved it. The rotor is hidden here.
This took 4 days to run, and was limited to 32k nodes+elements count because it's the academic version.
|The first iteration of the grinder was just two cones that I actuated by hand to grind beans. It was surprisingly effective given that it was entirely made of plastic.|
|The next iteration had some bearings and was a complete assembly.|
|You can adjust the grind size by switching out a plain bearing.|
|Two grind settings magnified 10x.|
|An early iteration with 3D-printed worm. The plastic worm heated up too quickly (low thermal capacitance and conductance) and was not stiff enough to avoid skipping. I added a reaction block to stiffen it, but the thermal problems were unavoidable.|
|I later added a steel core by pounding a nail coated in thermal paste into the hollowed shaft. That actually worked fairly well at low input power (< 30 W) but absolutely ate itself at 70 W.|
|A transient thermal analysis showed that the steel-cored shaft (upper) could survive about 7 W of frictive heating, while the all-plastic shaft (lower) could not (192 °C vs. 225 °C). So the nail solved the stiffness problem and at least partially solved the thermal problem.|
|A drawing I used to define the requirements of the shaft. I originally ordered it from Xeometry, which didn't work out well (they delayed 6 weeks and then didn't deliver). I redesigned it to be more manufacturable so I could use my friends workshop, and it turned out really well. I had to change the pitch to a standard value (2.5 mm) and the pressure angle to 25° because we only had 50° cutters. I noticed that the Volvo windshield wiper worm drive has a pressure angle of 0°, which would reduce frictive losses by about 10%. Because heat is a big design concern for worm drives, this could be an important enhancement.|
|The nearly finished shaft.|
|A rendering of a cross-section of the grinder.|
|A test of the new driveshaft. I discovered that the speed controller is asymmetric fwd/rev, and that at startup voltages under about 7 V it enters a "limp home" mode. At 12 V the controller hits 35k RPM! The skate bearings resonate - it's too fast for them.|