Error Propagation and CNC Drilling

This forum is dedicated to those Hobbyists Interested in CNC machining & 3D Printing in their home shops. (Digital Read Outs are also topical, as is CAD/CAM as it relates to CNC)

Moderator: Harold_V

Post Reply
User avatar
SteveHGraham
Posts: 7788
Joined: Sat Jan 17, 2009 7:55 pm
Location: Florida

Error Propagation and CNC Drilling

Post by SteveHGraham »

As long as people are busy not answering my questions, I'll throw in some new ones.

I got the mini-lathe working pretty well. My Chinese dial indicator says it's accurate to within a couple of thousandths on the z-axis, over 0.500". Now I want to make it better.

I'm trying to understand propagation of error. It seems to me that messing up the steps/inch number would be worse than screwing up the backlash correction, because the backlash correction is applied and removed over and over, when the carriage goes back and forth. Messing up the step number would give you an error that always increases as the lengths of movements increase. If you're off 0.010" over half an inch, you'll be off 0.020" over an inch.

Is this more or less right?

I'm supposed to have 6400 steps/inch on the z axis, but that doesn't work. It's actually more like 6321. I don't know how this happened, since it's a real imperial screw, not a metric approxmation.

I now feel cocky enough to order a real z ball screw. I found a Chinese job that will pretty much drop into the lathe, except that I may have to machine a mount for the nut.

Another question: how should I handle drilling? Putting drill bits in a tool post holder would be a nightmare for obvious reasons. I'm thinking I should rig the cross slide up so I can plop a permanent chuck mount on it whenever I want. Another clever answer would be to put a third stepper on the tailstock and move it up and down the ways with a drill bit in the chuck; that just occurred to me. The cross slide mount thing would be a lot easier, and I could use it for boring. I can get a Taiwan keyless chuck and stick it on there.

I could rig up a chuck mount, get it aligned, and screw two little chock-type deals onto the cross slide to index it against. Then it would be easy to attach and remove with good repeatability.

I've seen mini-lathes with tool changers. I think that's really cool, but it seems like it's over the top, because you can always pause a program and switch tool holders.

I have to figure out how to position tools. It seems like it doesn't matter whether I use machine or tool position for the x axis, because the work's x axis will always be the machine's x axis. I think using machine coordinates for z would be smart, because I would be able to move the work around in the chuck; I wouldn't be limited to some fixed position.
Every hard-fried egg began life sunny-side up.
User avatar
Dave_C
Posts: 960
Joined: Mon Apr 21, 2008 10:34 am
Location: Springfield. MO.

Re: Error Propagation and CNC Drilling

Post by Dave_C »

Steve,

There is no such thing as a "perfect" ball screw. However, the more you pay up front the better chances are you'll have less error over a given length.

Ground ball screws will be pretty consistent in the error per foot while a rolled ball screw may actually be a drunk thread.

Having said that, the way high end CNC controllers deal with this issue is through ball screw mapping. It is usually done using laser measuring equipment and creating a table to correct for the errors.

Most of us don't need that expense or that degree of accuracy. Being off .002" though is quite a bit and is not what I would consider accurate. But, on small cheap machines designed for fun this can be an acceptable error.

So if this machine is just for fun, then don't spend more than its worth to make it accurate. On the other hand, if you plan on doing accurate bores for bushings or bearings you gonna have to do better than .002" error.

If I understand Mach 3's approach to backlash I think if you enter your error (.010") in the backlash entry then the software actually know to go past the desired coordinate and then back up .010" to be spot on. This takes the "slack" out of the setup and should cut as needed.

Not sure it can do be-directional cuts when backlash is enabled. (I might be wrong as it has been sometime since I read the manual)

On a lathe setup, cutting toward the headstock all the time is not a bad idea anyway. A little slower on some parts but should be fine on our hobby machines!

Hopes this helps,

Dave C.
I learn something new every day! Problem is I forget two.
User avatar
SteveHGraham
Posts: 7788
Joined: Sat Jan 17, 2009 7:55 pm
Location: Florida

Re: Error Propagation and CNC Drilling

Post by SteveHGraham »

I agree that 0.002" is not a great accuracy figure, but there are a lot of things I can make successfully now. And as long as I know where the machine is actually cutting, I can lie to it and make it move to where it should be. Depending on how complicated the job is.

The controller I have will not let Mach3 fix the backlash! It insists on doing it, itself. This particular controller always makes the correction on the + trip.

Ball screw mapping sounds like a royal pain! I think I can live without it.

I'm thinking I should just go ahead and get a decent Taiwan or China lead screw. The x axis can wait, since it's considerably better than z.

What do you think about drilling? The more I think about it, the more I think I should make a dedicated chuck holder I can mount a couple of inches in front of the QCTP. I can use the CNC lathe to bore a taper into a block of aluminum, and that would hold the chuck. You must have found a solution to the problem, since your lathe is already working.

I've been trying to get tool location figured out, and I'm starting to think I'm better off homing the tool manually than stressing myself with switches and sensors.
Every hard-fried egg began life sunny-side up.
User avatar
SteveHGraham
Posts: 7788
Joined: Sat Jan 17, 2009 7:55 pm
Location: Florida

Re: Error Propagation and CNC Drilling

Post by SteveHGraham »

I keep learning disturbing things.

Evidently, you have to have at least a C5-grade screw in order to have less than 0.001" of error over a foot of travel. Most of the screws that aren't outrageously expensive are C7.

Arrghh.
Every hard-fried egg began life sunny-side up.
User avatar
Dave_C
Posts: 960
Joined: Mon Apr 21, 2008 10:34 am
Location: Springfield. MO.

Re: Error Propagation and CNC Drilling

Post by Dave_C »

Steve,

My drilling setup is real simple since I have a QCTP. I use my MT-3 Chuck that I once used in my tailstock but I mount it to the QCTP with an adapter that I bought from grizzy.

Dave C.
I learn something new every day! Problem is I forget two.
User avatar
SteveHGraham
Posts: 7788
Joined: Sat Jan 17, 2009 7:55 pm
Location: Florida

Re: Error Propagation and CNC Drilling

Post by SteveHGraham »

That kind of sucks the fun out of it, but it makes sense. It looks like Tormach makes an MT2 tool holder in 0XA size.
Every hard-fried egg began life sunny-side up.
Magicniner
Posts: 532
Joined: Thu May 30, 2013 4:40 am

Re: Error Propagation and CNC Drilling

Post by Magicniner »

You can get 1" parallel OD MT2 sleeves which are remarkably useful for putting MT2 shank tooling in unexpected places,

- Nick
WJH
Posts: 1417
Joined: Sat Jan 16, 2010 9:29 pm
Location: Florida

Re: Error Propagation and CNC Drilling

Post by WJH »

Well, I would simply buy a Chinese glass scale and use it for positional feedback into your Kflop. This would create a closed loop stepper system. The scales are around 100$ and are simple quadrature output. Won't need a super nice screw, you only need to worry about backlash. You're welcome...
User avatar
SteveHGraham
Posts: 7788
Joined: Sat Jan 17, 2009 7:55 pm
Location: Florida

Re: Error Propagation and CNC Drilling

Post by SteveHGraham »

That would be nice. I don't look forward to finding room for it on a mini-lathe.
Every hard-fried egg began life sunny-side up.
Post Reply