I feel like updating the saga of my CNC progress. I have learned a great deal about CNC, but in relation to the amount that I need to know in order to be competent, I know just about nothing.
First thing: do not build a CNC lathe. Do not even waste your time. A mill will do nearly everything a lathe will do, plus a million things a lathe can’t begin to do. You can thread with a mill. You can turn parts with a mill. You don’t want a lathe. Believe me. And if you have a lathe, no one will want to help you, because only about 10% of CNC hobbyists use lathes. Even software makers ignore us.
Second thing: precision is expensive. You can fix a machine so the computer compensates for backlash. But it doesn’t actually work. Say you’re milling a round hole instead of boring. You have to make the cutter travel in a big circle, as well as rotating. Every time an axis changes direction, you’re going to get error due to backlash. The only way to get rid of it is to use ballscrews, and cheap ballscrews don’t really work, although they may make you feel good. On the web, people with cheap screws talk about backlash figures between 0.005″ and 0.010″, which is huge.
When I got ready to build my lathe, I figured the machinery itself would be simple to create. The plans I bought didn’t mention ballscrews. Then I got it running and found out I had 0.018″ of backlash on the z screw. The software compensates on simple parts, and that’s very good. It’s worth a lot. I can do a lot of stuff I could not do with a manual mill, and I can do it with good accuracy. But sooner or later I’ll want to cut a part that requires a sudden change in direction on z, and I won’t be able to do it well.
Another annoying lathe issue: you have to tell the computer about the cutters you use. When you use a tool, only a tiny part of it will touch the work. The computer has to know where that contact occurs. Tools have rounded corners, so there won’t be a sharp, defined point where the tool makes contact. It will vary as the shape of the work and the direction of tool movement change. Lathe inserts vary a lot, so you may have to have a whole bunch of tools defined. I’m not sure HSS is useful at all, because you can’t grind it precisely. If you try to tell the computer the radius on a tool made from HSS, you’ll definitely have error.
I have not used a CNC mill, but it stands to reason that it should be less complicated. Say you’re using a half-inch end mill. You know exactly where the lower surface is, at all times. You know where the sides are. Not complicated. Maybe I’m wrong, but I can’t really see myself spending days telling the computer about the small number of tools I’ll use.
I’m going to keep the lathe, because even with limited precision, it can be useful, but I would not waste my time building another one. I’m hoping to use it for threading and tapers, as well as curved parts like dial handles. Things like that will justify its existence until I get a mill going.
People are telling me I should have bought a used machine. There are a lot of old CNC machines out there that have obsolete electronics. I may be wrong about this, but I will relate what I understand to be true. Years ago, CNC required big dedicated computers and expensive controllers. In 2014, a secondhand PC can handle most of that stuff. There are old machines out there with screws and motors that still work, but they have heavy, useless built-in electronics. Commercial users want to get rid of them, so hobbyists buy them and bypass the ancient circuitry, running them with Mach3 or LinuxCNC.
I would love to have an old vertical machining center, which is just a very fancy milling machine with an enclosure that keeps chips and coolant contained. But they’re gigantic. There’s one for sale down here, and it weighs over 8000 pounds. I’m crazy, but not that crazy. Also, what if you buy one, and the screws are no good? You’re out maybe six grand, and then you have to put another two or three thousand in. Arrgh.
I am thinking I should get a new Chinese mill, like an R45 clone or a Grizzly G0704, which is a little smaller and less rigid. I would have to put decent screws on it, which would give me gastric distress, but it would do just about anything I want to do. I still have the ability to add 6 more axes to my controller, so I should be able to keep the lathe and add a mill with no problems.
I bought Dolphin CAD and CAM, which is a pretty good commercial program. They gave me a big discount. I haven’t gotten to where I can actually use it, but I’m working steadily. I have managed to get Mach3 working (most of the time), so I use it to fine-tune the machine. Yesterday I cut a #2 Morse taper in aluminum, just to see if it would work. Tapers have to be very precise, but the geometry is simple, and the tool doesn’t change directions while it’s on the work, so I figured there was some hope that my lathe would pull it off.
Below is a photo of the taper. It fits fine in the center bore of my rotary table. I put Sharpie ink on it and tried to rub it off on the inside of the bore, to check for high and low spots, but I couldn’t see any problems in the result. It would definitely be sticky enough to hold a drill chuck in the lathe’s tailstock.
If I can make tapers that actually work, it will be a nice ability to have. I should cut an R8 and see what happens. My experience with lathes is that even with error in the chuck, you can get a very nice, round part as long as you don’t move it around while cutting it. If you take it out and put it back in, it won’t be in the same position. I don’t think it’s unrealistic to try to make an R8 taper, but the dial indicator or test indicator will tell me for sure.
I am enjoying the lathe. I plan to keep working on it until it can thread. But compared to a mill, it’s…handicapped. That’s just how it is.