The Golden Screw

I am still working on my CNC mini-lathe.

It seems that one of my big problems was misunderstanding the purpose of CNC. I thought of CNC as something intended to allow machinists to cut shapes which are difficult or impossible in manual machining. There is some truth to that, but CNC is also intended to save time.

Back when I was working on a kibbutz in 1984, I stupidly chose to be a grapefruit picker. The kibbutz had a CNC factory, and I could have learned a lot there. One of the volunteers said there was a machine they called “the golden screw.” You put a piece of brass in, and shortly thereafter, you pulled out a finished part with threads on it. A hungover teenager from Germany could do it. Machinists were not required.

I suppose it illustrates the deep purpose of CNC: to avoid paying any more for labor (or anything else) than absolutely necessary. Paying a skilled machinist a high wage for performing lots of operations on a brass part could raise the retail price to a hundred dollars or more. Paying a kid virtually nothing to shove parts into a machine and take them out is cheaper. And the faster the machine works, the cheaper it will be.

Earlier this year, when I started trying to get my lathe to work, I was confused because it seemed my references had a strange obsession with speed. There was a lot of information about making the lathe zip around like the Flash looking for an outhouse after a pie-eating contest. I thought it was bizarre that a person who made parts in his garage would care that much about speed. Of course, I failed to consider the fact that most CNC users are not in garages. They’re trying to make money.

If you have a $100,000 CNC lathe, I’m sure speed is no problem. It should be made to take it. But if you convert a manual machine, things are different. The machine has a limited tolerance for jerking and accelerating.

I had problems with the lathe skipping steps, sua sponte, in Gcode programs. I learned that it somehow decided to omit steps it couldn’t handle. Also, the couplers that turn the screws rely on friction, so if you make the motion too snappy, they can spin without turning the screws.

Last night I was working on getting my x-axis steps-per-inch and backlash figures right. I tried to cut a rounded end on a piece of Delrin. The lathe wandered off and crashed into the work. It skipped the steps that pulled it back from the work, so it kept moving forward.

The reason for this was that I had raised the motor’s velocity. The lathe decided to skip certain steps. When I lowered the velocity, everything worked.

Now it appears that I can machine things to within a few thousandths of spec, which is an improvement. The lathe works well enough to make parts, and that means it works well enough to be used as a teaching tool. So I have to have new goals.

First of all, I have to get on top of CAD and CAM. Mach3, the program that runs the lathe, offers prefab wizards for simple operations, but you can’t type in “door knob” and get a part. If you want to make anything more unusual than a screw, you have to draw it in CAD and send it to the lathe.

I am capable of drawing parts in Fusion360, Autodesk’s incredible free program. Now I have to find out how to get those parts to my contoller.

I also need to learn about tool placement. When you turn on a CNC lathe, the tool has to know exactly where it is, and that’s not as simple as it may sound to a person who isn’t a machinist. Locating something to within a thousandth of an inch is a job. If you have to use more than one tool on a part, you have to locate more than one tool.

Some people make revolving tool holders. You can put a right-hand tool in one mount and a threading tool in another, and you can program the controller to rotate the appropriate tool into place at the correct time. It won’t work unless you know where the tips of the tools are.

I don’t need a rotating tool holder; I should be able to get by with a turret or quick change tool post. But even then, I’ll need to be able to index everything accurately and tell the lathe where the tips will be.

By “index,” I mean there will have to be something on the lathe that forces the tool holder to fit perfectly into a preset position. Whatever this reference object is, the tool holder will “index” against it.

I don’t think a motorized tool holder makes sense for me. It will cost much more time than it will save. I’m not in a hurry, so I don’t mind programming pauses into my programs that will allow me to change tools manually.

I’m closer to getting the threading problem fixed. If I had known that my KFlop controller required two inputs for threading, I would have bought something else, because Mach3 likes one input. I have the KFlop, so I have to make it work.

I’ve had problems trying to find optical switches for it. I will need to mount a disk on the spindle with a slot in it, and I’ll need two slotted switches mounted at 90 degrees to each other. I kept trying to find a switch that would provide the KFlop with the required 3.3-5V input. I looked at datasheets, and I couldn’t understand why they didn’t list output voltage.

Of course, I was looking for a figure that doesn’t exist. I was looking at switches, and switches put out whatever voltage you feed into them. If you buy a rocker switch, the literature will list a maximum voltage, but beyond that, V-in equals V-out, so the datasheet won’t give you an output voltage.

I figured this out after asking some questions on a forum. It should have been obvious. The problem is that I had preconceived notions about transistors. Optical switches use transistors to provide output, and I am used to thinking of transistors as voltage amplifiers with a gain higher than 1, so I just assumed the switches made their own decisions about output voltage.

After all this is done, I’m going to want ball screws. The screws I have now are just too aggravating to work with. I keep trying to pin down the correct figure for steps per inch, but it’s elusive, and I think the screws are the reason. A setting that works fine in one trial will be slightly off the next time.

You would think you could rely on simple math, starting with the pitch of the screws, but I got bad results that way.

I am told I need fat screws, because they work better than skinny ones. I keep checking Ebay. It looks like it’s going to be a minimum of $150, all told. But it beats chasing numbers I will never catch with ordinary screws.

The lathe is not going to be the answer to my turning prayers. It will be wonderful for threading and for turning shapes that aren’t plain old combinations of cylinders, but it won’t be a milling machine. It seems like a milling machine would have been way more useful.

I suspect that even threading would be easy on a mill. You could make a threading cutter and make it go around a vertical workpiece, or you could put the workpiece in the spindle and run it past a stationary cutter. Maybe. I know they make CNC rotary tables for mills, so you should be able to mount workpieces in them and rotate the work that way. Anyway, the lathe is not looking like the optimal choice.

Oh boy. I was right. I just found a video.

Who wants to buy a lathe?

I exaggerate, but you can see what I’m saying.

Regardless of whether it makes sense, I will get this thing working. Maybe some day I’ll be able to make a little CNC milling attachment to go on the carriage, allowing me to mill things as big as 2″ by 2″. Hooray.

The C programming is going fine, although I had a strange problem. I wrote a program and got it working. Then I modified it, and it refused to change. It compiled fine, but it continued to do what it did before I modified it. Deleting the executable file didn’t help. I can’t figure that out.

I changed the source file name and compiled it, and it works. I can modify it.

It’s like the executable for the program that won’t change is stuck in my computer somewhere, refusing to leave.

If I produce anything useful, you can bet you will hear about it here.