Re: [CAD_CAM_EDM_DRO] testing for squareness

cadcamcenter@... wrote:

> Hi,
>
> Without wasting bandwidth, I'll like to post these question to the
> list:
>
> How does one test and compare CNC machines, particularly CNC mills?

> How does one measure repeatibility, precision, orthogonality
> (squareness)?

The professional methods use ball-bar indicators or precision
squares. If you had a very accurate square block, you could wipe
a dial test indicator along 3 adjoining sides of it. Any change
in indicator reading (once it was aligned with the axes) would
show ortogonality error on that axis. this is how it used to be
done.

The current standard is the ball-bar tester. These have an extendable
bar with ball sockets at each end, and a precise measuring device across
the extendable joint. There are end blocks that have a ball mounted on
top of it. The fancy ones have a shaft encoder in one of the blocks. This
records progress as the machine is programmed to orbit in a circular
arc move. the bar length measurment is recorded with the rotation
measurment, and then reconstructs an exaggerated graph showing
how far off an ideal circle the move was. By noting the position and
shape of the error graph, you can detect backlash, linearity errors,
orthogonality errors, stick-slip friction, servo jitter and a host of
other ills.

There are now also laser-based, non-contact measuring systems that
are even more expensive than the ball-bar testers.

>
> I was told one way to test for squareness is to machine a circle. If
> the x and y axes are not 90 degrees to each other, the circle will
> not be a perfect square. Has anyone ever done a test like this, or
> any other tests? How does one measure the deviation from squareness?

I assume you mean "not be a perfect circle." Yes, this will definitely
work. You would get an ellipse. Using an inside mike or a regular
mike and a telescoping gage, you find the largest and smallest
diameters, and that will tell you which way to move the top axis on
the stack. If there is much backlash, then you will have additional
problems interpreting the results.

>
> For the z-axis, tramming? was described in some earlier posts. Can
> this be used as a measure of the orthoganality of the z-axis to the
> xy plane, perhaps a figure expressed as deviation from horizontal per
> inch or per meter?

That is exactly what it does. But, just swinging an indicator around the
table will assure the spindle is orthogonal to the TABLE, but not
necessarily
to the X-Y plane of motion. That is why I face off the top of a piece of
scrap
with a small end mill (on manual machines) or cut a round groove in the
top of the scrap on a CNC machine. This surface is parallel to the actual
plane of X-Y motion. Then, you sweep a dial test indicator over the surface

(or through the center of the round groove) and adjust the head until there
is no change in the indicator.

>
> | |
> | | collet
> \ | /
> \ /
> | wire bent into Z shape and rotated
> |_______
> tip |
> touching | tip distance d above surface
> .................... flat surface

I use a fairly rigid linkage called an "In-di-col" which attaches a
set of adjustable links to the spindle nose. A dial test indicator
is attached to the far end of the links. I usually sweep a 6 - 8"
diameter path and align the machine spindle to less than .001"
indicator deviation across the entire circle. If you have local
wear on the table or ways, you will get deviations that are not
linear across the X or Y directions, indicating the surface you
are sweeping has a saddle shape. This is an indication the bottom
of the table may have a reverse saddle shape worn into it.

Jon