[cgallery]

I have a small K2 CNC desktop machine I use very occasionally.

I use Mach3 to drive a Xylotex driver, I create my files w/ Cut2D these days. I'm using a USB stepper into the parallel port of the Xylotex.

I use a circle in a piece of ply that is 9.0625". The ply is .5", I took .125" bites (climbing) and did a .001" final pass (reversing).

In my X axis, I think I'm very close to 9.0625" (maybe .003" over or so which for my applications is plenty close).

In my Y axis, I'm a little under, maybe .01". Still usable for my speaker-building stuff, just wondering if it could be better.

But the circle is slightly elliptical and I'm wondering if I could get it better.

This machine has acme screws and I guess my question is, is it possible that one acme screw could be .01" off over 9"? Or is that pretty unheard of.

Any suggestions on where to look for a problem like this. I'd love to find a FAQ on making final adjustments/dialing-in a machine, if there is one.

I have a dial indicator that resolves to .001" and has about 1" of travel. I also have digital calipers in 6" and 12" lengths.

[ger21]

The first thing I'd do is cut the same circle, with the toolpath in the opposite direction. If the Y axis size changes, then it could be deflection. If it's very close to the same size, then I'd adjust the steps/unit to get it to match the X axis.
But, Before you do that, make some straight moves to verufy that the distance it's moving is off the same amount as your circle.

[cgallery]

I can do that.

I also occurs to me I have some test squares I cut to check for square, but I never measured if X and Y were equal.

Are acme screws ever off by .01" over 9"? I guess that would be .001" per inch.

[ger21]

If they are cheap single start screws, maybe.

[cgallery]

I'll bet your deflection suspicion is the culprit. I believe that that Y axis is the most flexible on my machine. I was hoping a reversed final pass was the answer but I may not be so lucky.

I appreciate all your help.

[cgallery]

So I dug-out the test pieces I used for checking for square. Just square blocks of plywood. They are square, but they aren't squares. The largest is about .01" shorter in the Y dimension.

So I got my digital calipers out and mounted a 1/4" dowel pin in the router and pushed and pulled the calipers and compared my results to the Mach3 DRO. Not the best test because I'm applying hand pressure to hold the calipers, I'll admit to that. BUT, the variance found, between my five tests (dropping the two outliers), was .0011 per inch off, possibly explaining my .01" error on my 9.0625" circle.

I duplicated the same test on my X and was off .000375 per inch, or about .003" on the same 9.0625" circle.

So I think my Y axis screw is a problem. I had figured during my tests that maybe the motor connection was a problem but I checked that and also made sure I didn't zero Mach3 and my calipers after pushing in the travel direction, trying to remove any backlash from the equation.

So I guess I'll change the Y setup in Mach3 and do some more testing.

Should I just change "steps per" in motor tuning? I think if I change from 2000 to 2002 "steps per," I'll reduce my error substantially. Maybe not perfect but better than I'm getting.

[ger21]

2002 should get you close. You can also use decimal numbers if you need to.

[cgallery]

2002 should get you close. You can also use decimal numbers if you need to.

Just wanted to report back that on my actual machine's copy of Mach3, my "steps per" for Y was 16000 (when I said 2000, I was looking at a copy of Mach3 I keep on my working PC so I can refer to it when needed).

So I upped it to 16016 and tested with my calipers again. Now the Mach3 DRO was matching my calipers within .003" or .004" through the full travel.

So I cut that same circle again and it is very nicely round, within about .003 or so (plywood is a challenge when it comes to precise measurements).

Needless to say, I'm pretty thrilled at the improvement.

Thanks for the help and encouragement!