X and Y axis Calibration?

[onlyjus]

So you can probably guess form the tittle that my X and Y axes are not moving exactly the same distances....

Some background: I made a machine from skateboards bearings and aluminum extrusions like the one on instructables. My X and Y axes should be exactly the same: 1.8deg/step, 1/4-20 thread rod with a DIY anti-backlash nut. I use EMC as the controller, and a cheap stepper driver off ebay and HeeksCNC to generated gcode from step files I create in Solidworks.



So, I tested my machine by drawing several squares at 1", 2", 3", 4", 5" and 6". I then measured (with calipers) the squares in both the X and Y directions.





Above you can see what I measure. As you can see, my X axis cuts slightly too large by 0.0181" and my Y axis cuts slightly smaller by 0.0134". I thought that I could adjust the threads per inch in EMC to compensate for the difference, however if this was the case, the error should grow linearly with the larger the square. As seen above, the error seems fairly consistent.

Any thoughts/suggestions/ideas would be much appreciated! Thanks in advance!

[PaulRowntree]

A constant error would suggest backlash problems. You can measure this using a dial indicator and simple one-axis translations (no cutting required). Measure the distance from an arbitrary zero position to near one end of the translation axis in two ways.
a) With the slide moved to '0' from positive direction before your translation, and
b) With the slide moved to '0' from the negative direction before the translation.

The difference between (a) and (b) is the backlash. Mach3 has ways to compensate for backlash in software for some motions (not sure about emc2) , but the best fix will always be to tighten the backlash out of your mechanics.

[RomanLini]

I devised a simple "test piece" shape you can cut that measures your backlash.

It is discussed in this thread;
http://www.cnczone.com/forums/mechan...ash_gurus.html



The way the piece is cut means it adds a +XY backlash to the upper rectangle and a -YX backlash to the lower rectangle, and leaves the 2 rectangles joined as one test piece.

This makes it very easy to measure the backlash with a micrometer as the error is doubled and twice as easy to measure.

You can cut the piece fairly small in size and out of acrylic which is then small enough and hard enough to put a micrometer on the piece and get very exact backlash measurements.

[onlyjus]

Thanks for the responses! I'll have to try your suggestions and see if I can nail down the backlash.