View Full Version : CNC accuracy problems

03-28-2006, 07:48 PM
I have a cnc router using gecko drivers, amtek dc servo motors, and
a 65 volt dc power supply. My software for design is a combination of
MasterCam Mill V.9 and AutoCad 2006 and my post processor is Mach 2.

The problem I'm having is getting straight cuts. In straight lines the
cutter arcs the cut slightly instead of a straight line.
In other instances the cut of a rectangle will be skewed so that the
the lower right corner and the upper left corner will be slightly off center.

I have checked the gantry and table and all is properly set up and square.

Any ideas will be helpfull.

03-28-2006, 07:57 PM
I would bolt a machine square to table indicate to square then do a program move to see if you have either a drift under power or maybe some play in your lead screw this does sound like it a mechanical issue and not a controller issue unless it a backlash comp issue

03-28-2006, 08:38 PM
Does it do it in straight X only and Y only moves?

How long are the straight moves?

How fast is the PC?

03-28-2006, 09:02 PM
X axis only. One cut is approx 3 inches long and the arc is about .0625 at the center
of the arc.
On the skewing, the cut is only about 3/4" long or less but I get about 1/8" of
skew at top left and lower right .
The machine is a Pentium 4 2ghz running on Windows 2000

03-28-2006, 09:07 PM
Do you get the same skewing at different places on the machine?

03-28-2006, 09:47 PM
I guess the answer would be yes.
I am cutting two pockets. They are easiest described as
rectangular shapes but actually they are routs for
Humbucker pickups so they are mainly rectaingular with smaller "ears"
cut on each side. It is in these ear cuts that I get the skewing.
The upper right and lower left corners are correct but the lower right
and upper left corners drift out about 1/8" each. The rest of the cut
is fine.
I do two of these cuts about 2 1/2" to 3" apart and both cuts have
the same problem.

03-28-2006, 10:11 PM
Do you have backlash comp on and are you running constant velocity?

03-28-2006, 10:40 PM
No backlash compensation has been set as far as I know.
I'm not in my shop right now but I have a copy of Mach2
on the computer I do my design work on and I have not
set the backlash comp on either computer.
My router uses ball screws and not lead screws if that makes a

03-29-2006, 04:58 AM
Is this the sort of thing you mean ?
If this is the case, have you tried exact stop mode?

03-30-2006, 12:32 AM
Actally, it lokks more like this.
The skewing is mostly on the "ears" at
each side.

03-30-2006, 01:09 AM
That is not a perspective drawing? that is a view from directly above?
Post the cutting file in this thread and a few of use can test it out on our versions of Mach.
This will help us detemin if its a CAM problem or a Mach problem or a Machine problem.
EDIT: Also post the dxf too

03-31-2006, 12:33 AM
No, this is as viewed from above.

03-31-2006, 01:54 AM
LOL, I meant post it in a txt file and upload that. :)
You might want to edit your post.

03-31-2006, 08:52 AM
I've run the code in a couple of simulators I have and it looks straight and valid. I would say you should re-check the motor tuning to make sure the pulses per unit are correct. Are you using steppers or servos? A mechanical problem can cause loss steps in x or y fairly consistently.

03-31-2006, 01:40 PM
I'm using servos with gecko 320 drivers.

03-31-2006, 05:46 PM
Sorry, I see you said that before. Do you use double screws on one axis? Can you run a simple square the same size?

04-01-2006, 12:19 AM
Both the X and Y axis use a single ball screw.
I'll try running a square and rectangle without the ears and see
what the cut comes out like.

04-01-2006, 09:00 PM
Okay, I just ran a plain square (2"x2") and rectangle (1.5"x3"). Nothing
fancy here but I still get a skewed cut. bottom left slants slightly in,
bottom right slightly out. In addition I used a zig-zag cut pattern and about
every third cut it jumped over one row (leaving a rough island). Also, the
cuts in one direction were slightly deeper than the cuts coming back the
opposite way.
I ran a straight line down a piece of scrap wood approx. 30" long and then
ran the cutter next to the line in the direction of the skews (the line on the
scrap was aligned with the line cut in my waste top that is square with
the gantry's travel).

The cut was square to the line I had drawn on the wood when using the jog
to transverse left to right.

04-02-2006, 05:27 PM
Have you tried increasing the pulse width?

04-03-2006, 07:52 PM
Okay... first I've found the problem with the rough cuts on my square and
rectangle. I programed the cuts using a 1/2" bit but, since all my other
stuff has used a 1/4" bit, I forgot to switch bits. Duh.
Anyway, I checked the pulse width. It was set at 1. I reset it to 4. The cuts are
far cleaner but I don't really know if the pulse width change had anything to do with
that. I assume using the proper bit is what accounts for that.
There is still a slight skewing to the cuts but now it is the opposite of what it
was before.

03-24-2009, 07:47 AM
Hoping someone can help, not sure this is even the right place to post this...

I have a problem with gaining steps, I have marked on the work crosshairs and this is my start point. Using Mach3, a short program (2 minutes) goes out, engraves its thing and then when it comes back to what Mach3 says is X,Y =0 it is out by just under 1 mm in both axis in the positive direction.

It is not a sudden thing, this occurs over the course of the program. My guess is that noise is being generated from somewhere, affecting both the X and the Y but sheilding all the wires seems to have no effect.

Hoping someone can please help, out of ideas and hair :)