Test cuts to check for accuracy, squareness

[analogman]

What are some good test cuts to check for accuracy, repeatability and squareness. I suppose some I could measure with a tape or small enough to measure with the calipers.

thanks

[CarveOne]

If you have 6" calipers cut a 6" circle and 6" square. The circle should be the same diameter at any opposing points around the circle. Even if the measurement is not exactly 6" the circle should measure equal distances anywhere you measure it.

All sides of the square should be equidistant and 6". Check the square with a known 90 degree angle tool.

Cut a channel in MDF that is as long as you can cut and measure the length end to end and subtract the cutter diameter. Is this number exactly the distance that the gcode file told it to cut?

If the gantry is not square with your X axis you will see obround circles and parallelograms, not squares.

CarveOne

[Dean448]

I have built a gantry style machine that is working great but I don't expact a great deal of accuracy. The X-Axis gantry may not be perfectly straight and the rails guiding the Y axis also may be out. There are other accuracy issues from the way the spindle is mounted or how square it is to the table through X travel.
Other then keeping things tight the only thing I can do is square the machine. In this machine that amounts to adjusting the A (slaved to Y) axis limit switch. This limit switch effects where the right side of the machine is.. resulting in squareness.
This assumes that you don't lose steps from high loads or that Mach doesn't keep the steppers syncronized fior some other reason.
Its also probably a good idea to mill the table flat.
I wouldn't even worry about repeatability when cutting components out of wood (.001" to .005" repeatability). It should repeat every time.

[analogman]

Hey thanks CarveOne, that helps a lot.

[CarveOne]

On my first machine I cut a 18" circle and square because I have a 18" caliper. The larger the test cut is, the more a small error will be detectable.

As long as the error is consistently repeatable then you can correct it in the "Steps per" settings of Mach3 or EMC2 on an individual axis basis.

CarveOne