C What I Have Done

More progress on the CNC front.

I’m sure I already said I ordered some acetal rod for CNC practice. If not, now you know. For less than ten bucks, Zoro Tools sent me six feet of it.

Acetal, also known as Delrin, is a very tough plastic. It feels and looks like polyethylene. It’s so tough, they make gears out of it. I used it this afternoon.

Today I had a couple of goals. One was to get my backlash correction working, so the lathe would not wander all over the place while it cut. The other was to get the initialization code fixed so the motors wouldn’t disable every time I looked away for ten seconds. I believe I accomplished both goals.

CNC is an interesting thing. I got very excited about it in 2014, because I saw that it should–SHOULD–change the world by giving ordinary citizens the power to make all sorts of things in their own homes. Guns, for example. But I may have expected too much from the technology.

There is no reason why a big outfit like Microsoft can’t come up with a program that allows relatively unskilled people to design fairly complex parts and turn them into code so computerized tools can make them from raw materials. There is no reason why a big outfit like Samsung can’t make a machine that reads the code and creates what you designed, without a lot of bugs and accidents. These things have not happened, though, because demand is not there yet.

Right now, real CNC is not plug and play. You can buy a 3D printer, which is one kind of CNC tool, and you can make stuff all day, but if you’re like 99.98% of the people who own printers, you won’t be able to make anything useful. You won’t be able to design new parts worth a damn, because CAD is hard, and even if you did, they would be printed in crude, flimsy plastic.

Recently, someone wrote an article saying 3D printing was “dying,” and he said something like, “Eventually you get tired of printing out little plastic skulls.” This is what people do; they buy printers, they download files other people wrote, and they print worthless knickknacks. You can do better, but you won’t do it on a $300 printer, and you won’t do it without acquiring some skills.

If you want a CNC machine tool, which is infinitely more useful than a printer, you will have to either lay out some serious cash or find an incredible Craigslist deal with a seller who has no idea what the tools are worth. If you can get a decent used mill for several thousand dollars, you’re doing very well. Then you have a giant mill that takes up half of your garage, and you have to learn manual machining, CAD, C, Gcode, CNC machining, and God knows what else.

You can get a CNC router, if you’re happy being unable to work in metal. I don’t see the point.

To make it work, you will have to get a ready-made CNC machine, or you will have to put a lot of stuff together. Then you have to learn how to make it run. I don’t mean making it make things. I mean you will have to learn things like how to make the motors run at the right speeds, moving in the right directions. It’s not easy. Believe me, there is no one-sheet “Quick Start” primer that will get you on your feet.

I didn’t buy a machine made for CNC. I did a conversion. I had to get two computer boards to make my lathe work, and then I had to make a cabinet for them, install a power supply, make the manufacturer’s software work with the boards, and make the software and boards work with the lathe. Then I had to make the software work with Mach3, which is probably the closest thing we have to Photoshop or Word for CNC. It helps people who aren’t programmers run CNC tools.

After all that, I had to make all of it work with Windows 10, which is a terrible operating system when it comes to connectivity and drivers.

I had to do all this, just to create a machine that functions. Making new parts with it that are more complicated than drumsticks…that’s still down the road.

The manuals for my computer boards are horrors straight from hell. The guy who wrote them is an engineer, which means he is not really a human being. They probably seem very straightforward to him. To me, they seem disorganized, vague, incomplete, and misleading. And I can’t say anything bad about him. He had a Herculean task to perform. It’s a wonder he did it as well as he did.

So anyway, home CNC has astounding potential to put manufacturing power in the hands of ordinary people, but it hasn’t come through yet. It’s at the stage where computers were when people made them from kits, using soldering irons. If you want to be a CNC hobbyist, giving up your other hobbies is a really good idea.

CNC is having a big impact on people like me, but people like me aren’t that common. It won’t really explode until an average guy can take a prefab, user-friendly, plug and play setup and start making parts in a couple of weeks. That’s probably five years off. It can be done. If the Mach3 people had as many software engineers working on it as Bill Gates has working on the next bad version of Word, it would have happened already.

It will be a while before printing your own household items will be as common as printing your own annoying family newsletters. But it will happen.

Here is the part I made today. You are probably wondering what it is. Me, too. But I made it, and I had very few problems. I wonder why I didn’t try machining acetal sooner. It’s very useful, and the finish is incredible. I used a pointy Chinese carbide tool with, essentially, no radius, and this thing is so smooth it looks like it came out of a mold.

I’m still wondering what’s up with the steps/inch settings. My x screw is supposed to do 1/25″ per turn, and the motor makes 3200 steps/revolution, so you would expect 80000 steps/inch. Measuring the lathe’s movements, I came up with 79760. Explain that. Maybe the screw is actually metric, and it only approximates an imperial pitch. Hmm…what if it’s 1mm per turn? That would give me 78740. Who knows?

The lead screw is not metric. Definitely. Probably. It’s American. It moves 1/2″ per turn, which comes out to 6400 steps/inch. But I have to use 6321. I haven’t figured that out.

I had a horrible time getting the screws right. I kept measuring the movement of the lathe, doing math, and adjusting the motors. I finally realized the lathe motors jumped every time I turned it on. This jump wasn’t recognized by the computer, so it threw everything off. The guy who makes the boards told me, “Oh, yeah, the boards do LEAP ON THEIR OWN every time they come on, and it happens all the time because I put code in the program that makes them time out and go dead.” I am paraphrasing. I had to look through the C initialization program and try to figure out which part of it meant “time out and stop,” and then I had to figure out how to rewrite it so the motors would always stay on. I succeeded at that, and things started to fall into place.

I also had issues with the backlash stuff. You can’t just tell a machine, “Add 0.005″ of backlash.” Backlash always has to be taken off the same side of the screw. You apply a correction when the lathe moves in one direction, but not the other. You have to be consistent. So I had to root through the manual and find out where to put the correction. It turns out that if you have a KFlop board, the backlash correction is only added when you move in the positive direction. Good to know.

Keeping the motors on all the time makes it impossible to locate the tool manually, turning the motor dials. I had to learn how to use the “jog” controls to move the tool into position. That was a nightmare. The z jog was like a rocket launch. I found it unusable. I wondered why anyone would ever try to use it.

I got my screws accurate to within 1 part in 500, and then I tried to run a simple program, and the lathe jammed. I had to turn it off. The tool just rammed the work, like it was mad at it. Finally, I found out the acceleration figures for the motors were too high. If you set your acceleration too high, the motors just ignore commands. Like, “Are you kidding? I’m not doing that.” So if five lines of code, which the motor likes, say, “Go left,” and five other lines it hates say, “Go right,” it just keeps going left.

It turned out I had the Mach3 acceleration figures so high, they were roughly comparable to the acceleration of a rifle bullet leaving the chamber. This is why I had jog problems. And guess what? You have to put acceleration figures in Mach3 and also in the software that comes with the board. So pitfalls abound.

Anyway, I made this white thing, and I’m really happy with it.

Now that I know the lathe will function, I can pay for a Mach3 license and get the full version of the software, and I can think about ordering a z axis ball screw. I also have to get to work on the spindle encoder thing. I’ll need to buy hook spanner wrenches. You can never have enough tools.

CNC is great, but if you want to get into it at your house, you will still have to play Robert Goddard, even though CNC is over 50 years old. If you decide to do it, go with a mill. Don’t even think about a lathe, because you will be ALONE. Even a router will be easier. A mill is the best CNC tool, though, so that’s what you should get.

I’ll bet no one read this far. I don’t care. I love my shiny new part.