One of the reasons is that the drive gears on the LMS lathe (a SIEG SC4) are potentially hazardous to get working for fine pitches, since the only way I could get the gears to work had them sticking out of the lathe cover, where they could grab a sleeve or do something worse. I put up a video explaining the problem and then another showing the way I got it to work. My field expedient fix was to cover the exposed gears with a cardboard box.
The Sherline has its own inconveniences, although it doesn't expose you to gears with a motor running because it's hand cranked to thread.
My post last Friday showed that the results of a test which should just give some scratches on the blank I used and they appeared off. They were supposed to be perfectly overlapped but were spread out. By coincidence, a thread appeared on CNC threading on one of the forums I follow and that guy helped improve his results by increasing his motor RPMs and feed rate. I tried that and in that test, the five overlapping passes came out perfectly spaced and lying on top of each other. It took me a while to realize the scratches weren't as deep as they should have been, but after a couple of days, I found the combination and by this afternoon had successfully cut about 10 turns (half an inch) of 1/4-20 threads.
The commercial 1/4-20 nut threaded right on to it. Success.
The nature of computer work is that once you can walk one path you can walk many others, so I decided to get closer to the kinds of parts I did on my flame eater project and made a 10-32 screw in aluminum. I cut the blank to the outer diameter manually, not under CNC, and then rewrote the threading routine for the different sizes. Just before hitting the "run cycle" button, I decided to video it. This is 2 minutes of the system running (with me dabbing away chips and adding cutting oil). First try was a success.
Under 10 or 20 power magnification the threads look pretty rough and that's the only thing keeping me from saying it's done and adding more features. A commercial 10-32 nut still threads on properly and feels pretty good.
Ultimately, I'd like to be able to take pieces that have to have special threads put on them and do those pieces on this lathe. In the engine, there was a piece that was to be threaded 6-32, but only was big enough to have 3 turns exposed for the nut. It's on the left in this picture.
The part has operations done to it besides turning the OD of the bottom piece, and the major diameter of the small threaded stud, but it would be conveniently cool to put a piece of, say, 1/2" aluminum rod in the lathe, use the CNC to turn it to 3/8 OD over all, then turn the little stud (top) down the major diameter of a #6 screw, and finally cut the three turns of thread. All without taking the piece out of the lathe.
Gimme another couple of months.
Ahhhh....feels good, doesn't it?
ReplyDeleteMachining aluminum has a few tips, as I'm sure you know. As my Dad used to pound into me..."Speeds and Feeds"!
Any idea why the previous try's paths weren't on top of each other, and why speeding it up kept it in sync better? My guesses: one flash per revolution isn't a lot of resolution, and maybe there's a PLL in the controller that works better when it runs faster. Maybe the slower spindle has more RPM jitter from loading, or worse motor performance.
ReplyDeleteNot really. I'm also not 100% sure it's gone. And I'm not 100% sure it wasn't that the chuck wasn't tightened enough.
DeleteMy first suspicion was jitter, because some is visible on the pulse. The length of the pulse varies while I watch on the scope. It's not what seems to be a lot, say about 8% total, but it doesn't make a lot of sense.
I have the option of taking the motor and headstock off my other lathe and seeing if it changes. That's probably the next experiment.
I'm still in awe of CNC can do. Changing threads on my small Grizzly means going to the chart, swapping gears, and usually finishing with a die nut to get clean threads.
ReplyDeleteA second look reveals that at least some of the threads have a strongly asymmetrical profile. Not sure how that might happen. Maybe the threads are getting pushed to one side?
ReplyDeleteExactly. Toward the left end of the 10-32 screw one thread almost looks cut in half vertically.
DeleteIt almost seems like perhaps the system is missing the sync pulse, or perhaps messing up its timing.
I looked for problems in the pulse and don't see any errors more than a couple of percent. It doesn't seem like 4% timing error, say, could make that big a difference.
That's the target of research now.