What's Wrong With This Picture?

[adt2]

I need some help quick. I am trying to carve some names in 1/2" MDF for a function my wife is hosting tomorrow. I'm using VCarvePro with a single-line font for the name text. I've just recently gotten my machine up and running, and while it moves in all the appropriate directions, a quick look at the results in the picture below tells you something is wrong.

I've marked the X- and Y-axis directions on the stock. I hope you can see the letters in the picture; they should all be the same size and the same height, but they're kind of all over the place. Any suggestions would be greatly appreciated.

[Al_The_Man]

At first glance I would say Backlash exists in the Y?
Al.

[ger21]

If it was backlash, the S would be messed up. I'd say you're losing steps on the Y axis. decrease the accel and velocity on the Y by 25%-50% and see if it helps.

[adt2]

Well, that seems...bad. During assembly and initial testing, I found a couple of small (dime-sized) O-rings on the machine, and I couldn't imagine where they'd come from. I had to disassemble part of the gantry to make an adjustment at one point, and all this shredded O-ring material came off/out of the Delrin 5-start nuts on both the Z- and Y-axes. After some light poking around here and a couple of emails back and forth with Ahren at cncrouterparts.com, we determined the O-rings were from the OD of the Delrin nuts but that they probably weren't mission-critical. So, I am currently running with no O-rings on the two Delrin nuts.

Sounds like maybe I need to replace them. In the meantime, is there something I can do to minimize/eliminate the backlash? At least long enough to run this one project? If not, I'm gonna be in trouble with the wife.

[adt2]

If it was backlash, the S would be messed up. I'd say you're losing steps on the Y axis. decrease the accel and velocity on the Y by 25%-50% and see if it helps.

This was actually the first thing I tried. Decreased Y- and Z-axis velocity and acceleration (velocity by half, acceleraction by 2/3). No change in the outcome. I can post another picture if you like, but it looks basically identical.

[ger21]

What are they currently set to?

[adt2]

Started at 160 and 3. Totally pulled those numbers from thin air. Round two had them at 80 and 1. No change.

FWIW, on this particular job, speed I irrelevant. I have all day if setting the speed to one inch per minute will solve the problem.

[adt2]

I am using rack and pinion hardware from cncrouterparts.com. Per the math on Ahren's site, my X- and A-axes are set to 1909.861 steps per inch. When I chuck a pencil in the router and tell the X-axis to move 1", it moves 1" (or as close as I can measure using this uber-scientific measurement system).

My Y- and Z-axes both use 1/2"-10 TPI 5-start Acme rod. Per the math in the Mach user guide, I have those set to 4000 steps per inch. I have not tried the scientific "pencil-in-the-router-spindle" test on Y/Z. Should I give that a whirl?

[ger21]

The accel should be OK at 15 or 20.

Also, try swapping the X and Y drives. Just swap plugs at the G540 and test. See if the problem moves to the other axis.

Actually, I'd check for a loose coupling on the Y axis.


Can you post the g-code?

[adt2]

Okay, I'll head out and try your two suggestions. In the meantime, here's the GCode for the names.

EDIT: The Y-axis coupling appears to be fine - good and tight. On the switching motors...how do I do that without a bunch of rework on the machine? Seems like that's going to involve more than just swapping X- and Y-axis motor cables, isn't it? What happens to A-axis? Seems like if I just swap X- and Y- cables, then the A- is now slaved to the Y instead of the X, and the origin is in the wrong place on the GCode, etc.

I may need some more guidance on this deal.

[ger21]

Seems like that's going to involve more than just swapping X- and Y-axis motor cables, isn't it? What happens to A-axis? Seems like if I just swap X- and Y- cables, then the A- is now slaved to the Y instead of the X, and the origin is in the wrong place on the GCode, etc.

Yeah, don't do that.

Try swapping the Y and Z. Just change the pins in Ports and Pins after swapping the plugs.

[adt2]

Okay....swapped Y and Z. Changed ports and pins. Y seems to operate smoother (by eye/ear), but I could be imagining that. Z is definitely choppy now; that's still eye/ear analysis but it's definitely there.

Ran the program and got results I wasn't expecting. Letters are still different heights (in the Y); but now the letters get deeper (in the Z) as they go. My uneducated analysis leads me to believe that the Z axis is losing steps in the + direction (i.e. after each pass, it's not raising as far up as the computer tells it to; hence subsequent passes are deeper than they're supposed to be).

Still no idea what the deal is with the height of the letters, though. Or why it doesn't complete the "A" at the bottom right, for example. The bit "ramps up" out of the material at the bottom right corner of the "A" without completing the cut.