[nlancaster]

I have a professionally built CNC router, 4x8 foot, in the x-axis travel direction we are seeing discrepancies of as much as .015 inches in feature positions.

At this moment I don't want to discuss the manufactures name to avoid backlash because I am actively working with them on the same issue and not really getting anywhere.

It uses a multi start ballscrew with AC servos with encoders mounted on the servo itself.

These are operated in what I assume is an "open" loop style by WINCNC control PC.

The servos and encoders are connected to servo drivers and that is the only feedback loop I am aware of.

The driver is told to move 16.25 inches and applies what it believes to be the correct number of rotations to the screw to accomplish this.

But when we cut or drill features they are short only in the x-axis by upto .015.

Some features are actually over, see the below example.

I am drilling 3 holes in a pattern that will be used on a part.

Those holes are 7.375 inches apart.

From X 0 to X +14.75

Pair 1 to 2 +.002 to +.004

Pair 2 to 3 -.010 to -.015

Pair 1 to 3 -.010 to -.015

TO the best of my ability to measure the steps per unit in WINCNC appear to be set correctly.

There is no excessive movement in any x-axis or z-axis component when tested with a .00005 DTI.

Used a precision tape measure and vbit, over 58 inches of travel, no discrepency large enough to be visible.

Any tips, or things to check the braintrust here can think of?

[CNCMAN172]

I have a professionally built CNC router, 4x8 foot, in the x-axis travel direction we are seeing discrepancies of as much as .015 inches in feature positions.

At this moment I don't want to discuss the manufactures name to avoid backlash because I am actively working with them on the same issue and not really getting anywhere.

It uses a multi start ballscrew with AC servos with encoders mounted on the servo itself.

These are operated in what I assume is an "open" loop style by WINCNC control PC.

The servos and encoders are connected to servo drivers and that is the only feedback loop I am aware of.

The driver is told to move 16.25 inches and applies what it believes to be the correct number of rotations to the screw to accomplish this.

But when we cut or drill features they are short only in the x-axis by upto .015.

Some features are actually over, see the below example.

I am drilling 3 holes in a pattern that will be used on a part.

Those holes are 7.375 inches apart.

From X 0 to X +14.75

Pair 1 to 2 +.002 to +.004

Pair 2 to 3 -.010 to -.015

Pair 1 to 3 -.010 to -.015

TO the best of my ability to measure the steps per unit in WINCNC appear to be set correctly.

There is no excessive movement in any x-axis or z-axis component when tested with a .0005 DTI.

Any tips, or things to check the braintrust here can think of?

You need to check a few things. First do a test and check steps per unit and make sure they are correct. Second check with an accurate dial indicator to see if you have backlash on that axis.

Good Luck

[nlancaster]

I used a .000050 DTI and no apparent backlash. If there is any it is less than .0005 on the DTI.

I setup the DTI at the drill position of the last hole that is off by .015 and ran the program several times, the DTI was always at or below .0005 discrepency.

I checked the Steps per unit with a precision tape measure and vbit. Moved 58 inches and no visible discrepency from that movement.

We are measuring using pins in the drilled holes with 2 different calipers, both calipers passed calibration less than 6 months ago, both get the same readings within .001.

[CNCMAN172]

What did you use to create the gcode program. You could use MDI commands to simulate the code you might have the wrong post processor selected. Using a precision rule or digital scale to verify 15 inch moves seems to suggest the hardware is good

[nlancaster]

We use mastercam for the programs.

This post has been used many times to make parts that are in the +/-.003 range on this very machine.

This issue in x-axis travel is new.

For the manual moves I was using MDI.

[nlancaster]

In a plot twist....

I had been measuring hole to hole.

Did a different test where I cut a datum surface and measured from that datum surface to each hole, and they are within =/-.008 of nominal position, but the hole to hole dimensions are still out by .015 on one set of holes.....

- - - Updated - - -

In a plot twist....

I had been measuring hole to hole.

Did a different test where I cut a datum surface and measured from that datum surface to each hole, and they are within =/-.008 of nominal position, but the hole to hole dimensions are still out by .015 on one set of holes.....

[CNCMAN172]

Have you measured your cutter diameter to see if it is undersized. Also check cutter compensation it might be active and causing the size differences

[nlancaster]

Drilling holes, so cutter comp wouldn't be involved at all.

And doing further checking yesterday we found the Z-axis to be out .035 in 10 inches. Going to work on getting that back to perpendicular today.

[CNCMAN172]

If you are using a drill bit you are correct but you could have been cutting holes with endmills and in that case cutter comp could come into play

[CNCMAN172]

Are you using stepper motors or servo motors with encoders. If you have steppers u good be losing steps

[nlancaster]

servos with encoders, run in an open loop mode from wincnc.

Meaning wincnc tells it how many steps to move and that is the last wincnc knows about the situation.

[nlancaster]

and late last week we found the z-axis was out of tram.... by .033 in the x-axis.

Retrammed the z-axis itself.

While checking the spindle foudn it was out of tram by .045 in 12 inches..... in the same direction as the z-axis.

retrammed the spindle.

Drilled some more test holes and found we were withing .006 of nominal.

Drilled all 17 holes in the customer part and they are all .003 or better to nominal.

issue may have been the tram the entire time....

[CNCMAN172]

That actually makes sense