Being driven insane by an offset problem!

[Cold Fusion]

*Sigh*

Here we go again. Today I achieved .0005 precision (really excited about that) but the offset was still there. Attached is an example of what's happening.

So far I've ruled out router mount deflection, bit deflection, and the g code. In a few minutes I'll rig up a ball point pen and trace the outline of this part to see if I'm getting the error there too. If I do then the problem must be in Mach2.

[Cold Fusion]

This is crazy. 30 minutes ago I went out to test it on pen and paper and the profile was perfect. I thought that the problem must lie in the hardware since I was getting perfect results with a no load cycle. So, I used a piece of 1/4 Lexan to make a test cut of the above file. Cut it fast and in one pass to put as much stress as possible on the system. When the part popped off the sheet I was expecting the same old problem. To my amazement it wasn't there. Of course eyes can deceive so the calipers were used. It only varied by plus or minus .001 on the X axis.

Don't ask me what fixed it because I'm really not sure. The only thing that might be causing it was the fact that I did a single pass profile instead of a multipass.

[JRoque]

Crazy question to a crazy problem: could it be a ground loop/noise issue? You don't see the problem with a no load run, a pen or cutting plastics... Maybe you can try it by running a wire from the frame of the spindle to ground; just a thought.

Later,
Julio

[Cold Fusion]

Considered it. There is one thing that doesn't go with that theory though. I've machined plenty of plastic (delrin) pieces of this design and I got the problem.