What's new
What's new

Possible drift of x, y whilst machining

msw135morgan

Plastic
Joined
May 19, 2014
Location
San Luis Obispo
Have a Velox gantry CNC, uses 4 limit switches for homing. Gecko control box drivers, Mach3, HSD spindle.
Machine has been leveled as well as can be done (Machinists' level), used a granite master alignment
triangle to set x, y, z. No backlash to speak of. Use SolidWorks for design, VisualCam for CAM, with Win7.
Been doing this for 2 years now, certainly not an expert. We have been making negative molds for producing
carbon fiber parts. The molds are machined from HiMacs plastic, with alignment pin holes. During the past
6 months it appears as though there has developed drift of x,y coordinates whilst machining. Not sure how
to describe this. Mold pairs are machined during the same run, side by side, mirrored. When done, and the
the molds physically aligned, the midpoint of the long edges (x axis) are misaligned by 1/16 inch top and
bottom, with y ends aligned(5" long y, 45" long in x). The alignment holes are mismatched as well, as
are the mold cavities. If one of the halves is then rotated around z by 180 degrees, the mold edges, as well
as many of the alignment holes do align. It is almost as if there is drift in y, creating curved, rather than
straight lines in x. Have another Velox CNC and a Taig Micromill 2000, both of which do not have this problem.
I do not know where to begin in troubleshooting this. We routinely home before machining, and return to 0 is
accurate and precise (test this using a precision ground alignment pin at 0, with a Starrett last word). The
machine shows orthagonality with a granite triangle (longest axis 15 inches) and the path is not sinusoidal or
weird when manually running along the edge in x or y. Any suggestions or resources would be appreciated, I am
really perplexed. Thanks. Morgan
 
Hello Morgan,
A suggestion: You have a machined object of complex shape which is not aligning with another machined object of complex shape. It is probably hard to quantify exactly why, because the shapes do not lend themselves to easy metrology. The suggestion is to make some objects which DO lend themselves to detailed quantitative analysis.

For example, suppose you machine two plates to the same footprint as the molds, and mount them the same way as the mold parts. Then machine some patterns of holes or other fiducial surfaces whose positions and dimensions are easily measured accurately with the tools you have in house. For example, a grid of precise (say reamed) dowel pin holes, some edges which "should" be straight, perhaps other edges which are perpendicular to the first edges. Perhaps some rectangular or circular bosses whose dimensions can be easily measured. Maybe make the plates mirror images.

After machining, flip one plate over the other and see if the holes match up. Then measure the locations, sizes, angles, straightness, etc of the various edges, bosses, etc. This should give you some real quantitative feedback on the issue.

There is an old saying in precision work to the effect that if you can't measure the problem, you can't fix it. The best way to measure something difficult is to simplify things so you really DO understand what you are seeing. That can be very difficult to figure out how to do, but it usually brings a lot of clarity to things.

Consider also using good indicators to sweep known rectangular, straight or right angle metrology tools such as straight edges, triangles, blocks etc. You can stack up gage blocks from a known edge and see if the machine position agrees with the calculated position from the standards. Basically, try to figure this out with simple, precise measurements and tools, not by waving molds around and grumbling about the mismatch.

You may wish to check the sticky note about which machine tools are appropriate for this site, as there are a few which are generally sort of off limits. I am unfamiliar with yours, but you might check.

Best wishes for finding and fixing the problem. If you can find some simple, robust tests which are definitive and quantitative you should be able to sort out the issue.

Michael
 
Hi Michael, Thanks very much for your excellent suggestions and I shall set up a plan along the lines which you
have suggested. I shall also check the sticky note. Morgan
 
Lost motion is not unusual in a stepper motor driven machine.
.
.
stepper motors often are open loop control and loose position if overloaded. and do not auto return to position
.
servo motors with feedback, measure position change and will go back to position if overloaded. that is if machine setup that way.
.
on a prototrak mill if you program position drill X0.Y0. and you grab a handle turn hard enough and shift X position as soon as you let go it will return to position
.
not all machines are setup to measure actual position movement versus turn of ball screw which can have backlash. not all cnc hold position or if measures movement will return to position
.
servos can also oscillate not unusual to see +/-.001 movement for a fews seconds til it stops moving. sometimes never stops moving if servos need tuning or adjustment
 
Hi Morgan:
How confident are you that the X axis motion is actually linear and that it stays that way as the machine warms up?
This is your first place to look, and if it's not you have gone a long way to define the problem.

Clamp and align a straight edge on the table.
Command a G01 motion that only moves X.
Run an indicator along the straightedge.
If your motion is good, command a bunch of back and forth moves and get an infrared thermometer to see if the frame or ballscrews are heating up.
Watch the display to see if Y is moving even when not commanded to.
Stick indicators on the Y axis to see if it's moving even without a command to do so.
1/16" of mismatch is a lot...something pretty obvious is going on.
You just have to be methodical about how you look.
Cheers

Marcus
Implant Mechanix • Design & Innovation > HOME
www.vancouverwireedm.com
 








 
Back
Top