X axis not constant: Any ideas?

[becikeja]

I’m setting up a new machine. I am using Mach 3 software and need some direction.
As you can see in the picture when I cut the number 16 everything works perfect. The right depth, shape etc.. Everything seems to work exactly like I want it to. Now look at the shape above the 16. This is supposed to be a 1.5” square with a raised circle in it. The raised circle seems fine, the Y axis seems fine, but what is happening to the X axis? It is consistently off creating a slanted affect. It is being cut from a planar toolpath generated by Bob Cad. The Mach screen shows the cut perfectly, and watching the X axis reading on Mach seems to indicate I have no issues. I first thought I had a sticking problem with the ball screw and backlash nuts, but how could it be so consistently off if that’s the issue? I have no clue what to even look for next to fix this.

Your help in resolving this issue is most appreciated.

[microcarve]

I'd look for the coupling slipping. Or if it's a flexible type
with a lot of excess "springiness" to it.

The screw could also be shifting end to end on dir change.

Nut may not be tightened good enough

Looks like a coupling is slipping at first glance to me...


John

[becikeja]

Thanks for the note microcarve,
I doubled check my couplings and made sure they were tight. I also double checked to make sure there was no movement on the screw by trying to move it, is there a more accurate way to check this?? After I tightened everything up, I ran the test cut again. I get the same slant. It’s consistent, so I guess that’s a start.

The couplings are the flex type, is there something I should do to these, or will I need to replace?

[microcarve]

Thanks for the note microcarve,
I doubled check my couplings and made sure they were tight. I also double checked to make sure there was no movement on the screw by trying to move it, is there a more accurate way to check this?? After I tightened everything up, I ran the test cut again. I get the same slant. It’s consistent, so I guess that’s a start.

The couplings are the flex type, is there something I should do to these, or will I need to replace?

Looks like something is slipping mechanically.


The spring coupling should be consistently wobbly if it were that, but
you're losing the actual position, so that's probably not that.

The nut for the axis could be shifting.

It's something simple. It's an additive type of loss of position so
it's something just not tight enough.

It can't be much....

I guess you did a run without the router running so you'd hear missing
step sounds/glitches...?

Got a picture of the setup?


John

[microcarve]

It sure does look like a coupling or nut slipping the more I look at it.

Couplings can seem tight but the shaft in them can still twist.
Especially if there's any oil or grease on either shaft.

If material or the router itself was shifting, that'd be very noticeable.

It looks like it 'tries' to get back on track in a few places, so
lost steps wouldn't be regained.

It looks exactly like a coupling problem I had a while back.



John

[cyclestart]

Alignment marks on the motor shaft, coupler, and screw would tattle on a loose coupler. Easier said than done on some machines.

I'm wondering if missed steps from over aggressive machine settings could cause the same effect ? Try replacing the G00's with G01's using your favorite text editor. Think of it as a fishing expedition

I had a similar problem caused by driver timings but that's not a common cause. I'm with microcarve, the simplest answer is usually the correct one.

[becikeja]

I went through the entire machine to insure everything was tight. I slowed the machine down to 25 ipm and then to 10 ipm with the same results. I tried cutting at a depth of .05 inches to make sue that was not causing unnecessary drag. Same issue. I then redirected my tool path to move from top to bottom instead of from left to right. Similar results just much more random. I am coming to the conclusion that it has to do with either the bearings or the anti-backlash nut. When I turn the spindle off and listen it seems to still make a lot of noise when it moves. This does not seem right. I think the bearings may be shot and are causing unnecessary friction leading the motors to misstep. But what is puzzling is the repeatability of the misstep if it is indeed being caused by the bearings. Anyone seen this before??? Any other ideas??

[microcarve]

Welll....if it's not the usual simple mechanical problem, then
there's the possibility of a few things.

Backdriving. If motors are too weak. If it's a high lead screw.
Anti-backlash Nut grabbing on direction change can cause that to happen.

I'd turn off power to unlock the motors. Then see if I can move the
axis by pushing it. Force it to backdrive. If it's easy to do, you
could need motors with a little more holding power.

You didn't mention what sort of screw or motors you have....?

I guess a really bad bearing could cause it to be harder to move
in one direction than another. Haven't seen that before though.

John

[ger21]

What may sound like bad bearings could be resonance.

Try cutting your acceleration in half, though.

[tommylight]

Just my 2 cents, but i would bet my be-hind that you have so type of backlash compensation set in MACH that is causing that to happen, since it does fine number 16 that has very litle movement but when it has to change direction often it keeps over compensating in one direction.
Feel free to take a torch on me if i am wrong, but i would like a nice cold beer if i am right !
Tom

[louieatienza]

Are you using ACME rod or allthread? How do you have your leadscrew fixed? If you made it too tight you might have overloaded the bearings if they're not angular contact bearings? Also do you have a flat on your drive shaft for the coupling, or is it a fairloc or other type of locking collar?

Sometimes the linear bearing might stick in one direction but not the other? I would disconect the leadscrew, and check that the axis moves freely.,, Make sure your screw is in line and not bent. If you suspect it's software, you can disconnect the screw from the stepper, and put a flag on the shaft, and run the program. If the stepper's getting the right signals, then the flag will stop at two positions exactly the same each time....

[cyclestart]

Anyone seen this before???

Yes, and I solved it in software parameters, specifically the step and direction timings. If the lost position is always on the same axis this isn't likely to be the solution.

Have you tried swapping the x&y motors and drives ? Process of elimination backyard mechanic style

If you suspect it's software, you can disconnect the screw from the stepper, and put a flag on the shaft, and run the program. If the stepper's getting the right signals, then the flag will stop at two positions exactly the same each time....

That will tell you if the stepper is following commands. Good idea.
If the stepper is dual shaft follow up with this: A flag on the screw and another on the back of the stepper will tell if things are rotating together. On a machine with open access laser pointers could be attached to the rotating bits and aligned to marks on a wall, now we're talking real accuracy . ( I'm half kidding, having some fun with the concept) We really haven't been told much about this machine such as how much that lost position means in terms of screw rotation.