Turns out that drawing I posted of the spindle nose is incorrect. It shows 41.54mm over 2.38mm thread wires. I made 2.42mm thread wires, and my spindle is only measures 40.90mm with them. Martin from the CNCecke also confirmed that drawing is incorrect with thread wire on his spindle as well.
Today I cut the thread in the spindle guage. First ground up a 55° bit with 5° of top rake on the Clarkson. Set the top slide to 27.5° and zeroed it, scratched off on the work with the cross slide and zeroed it as well. From then on all cuts were make with the top slide.
I turned with the slowest back gear. 73RPM.
I made quite a few mistakes, so I'll have to make another one, but at least it shows the Boley can cut it's own spindle thread in Meehanite.
The mistakes:
- I had the alignment boss spot on size, within a 0.0001" of the spindle's boss, and nicely polished with 600 grit sand paper. Thought I could improve the finish with 1000 and 2000 and went 0.0003" under. The alignment boss is 41mm h5 which only gives a tolerance of 0.0003" (11 micron). I need to get some metric micrometers. My 1-2" Starett also can't be trusted for absolute values, as zeroed on a 1" standard it's 0.0003" out on the 2" standard (or one of my standards is out). Still, it works fine for comparison measurements like this.
- Setting up, I slipped on the keyboard and did a G0 Z-41 instead of G0 Z-1, and put a thread into the alignment boss.
- In the middle I got impatient and tried a couple of cuts at 140rpm. I suspect that LinuxCNC accelerates at max rate for the axis when starting a G33 thread move, and then once at speed maintains sync. Speed variation during the threading are compensated for, but speed variation during acceleration aren't. Since the rotation angle travelled doubles during that acceleration with double the spindle speed, the thread then has a slightly different track. Need to pick a speed and stick to it for the whole thread.
- I started off doing 0.2mm infeed on the compound each cut. At about 0.8mm in, I had brain fade and infeed to 1.4 mm or so. Stalled out with a slipping spindle belt. Luckily it was early enough in the cut that most of the scars were cut out later.
- I dropped to 0.1mm infeed per cut at about 1mm, and to 0.05mm per cut at 2mm. Got impatient and tried 0.2mm when already at 2.2mm. Stalled again, but less harm down. Luckily did not lose sync, as LinuxCNC synced the Z stepper down to a stop following the spindle.
- Didn't stop to measure often enough. Went to 3.30mm infeed. I cut the thread about 0.15mm too deep.
This one is a bit too small to be useful as a guage. A back plate cut to have a nice fit to this is probably still too tight for the real spindle.
Mark