Can't Machine a Simple Circle with a Probotix Fireball

[JPX2000]

Have you made any adjustments to backlash correction in the program settings?

Sent from my LGL84VL using Tapatalk

No, I haven't gone to the software settings yet. I checked for SPU adjustment, but it wasn't needed. I'm not really sure where to go or what to measure for software backlash correction.

[JPX2000]

I'm running out of ideas. There is one other possibility that I can think of that is software related, Constant Velocity (G64) vs. Exact Stop (G61) This can effect some machines. Try using the opposite value.

Failing any software related issues, I'm going to suggest a control problem, maybe a loose connection, bad solder joint, or just simply a failing board.

OK, Jim, I'll give that a try. Do I edit that in the G-code or is there a setting for it?

[Jim Dawson]

There should be a setting option when you post the code, or you could just hand edit the G code.

[GerryG]

My vote's on electrical interference. If the photo of the control box is yours, then it's clearly had a serious impact at some point - it might even just be that there's intermittent grounding somewhere in the box. I'd pull the case apart to make sure that there's nothing being shorted, and if it's got an internal metal liner to act as a faraday cage, then that's definitely been compromised by said imact.

[JPX2000]

Yeah, Gerry, that's a photo I posted to show someone the connector layout on the rear of the enclosure. Not my box, and definitely messed up (by shipping and handling, no less). I'm not 100% sure, but I think I've got a vibration problem in the spindle. Everything else has been checked out. My TIR reads less than 0.0005", but when I cut a slot for test purposes, using, say a 1/8" cutter, the width of the slot is around 0.150". Not good. But it IS something to start working on. Everything else was checking out fine.

[stealth027]

Don’t know if this will help, but when I still had my v90 I had a similar problem. Turned out to be the lead screw of the y axis. Tightened it up and it worked like a charm again.

Can’t hurt to try.

For the last two weeks I have been working against degrading operational characteristics of my good old dependable Probotix V90 Fireball. It started with non-circular rings cut out of 1/8" aircraft plywood. It has now gotten so bad that I can't cut concentric circles, let alone hold tolerances to within acceptable levels.

I'm using Vectric Cut 2D CAM and Planet CNC USB controller. The router is a Ridgid R2401. I checked for backlash, loose rods and guides, and TIR on the router. Everything seems to be fine. I have changed cutting speeds, feed rates, cutters, and even stock hold-down methods.

I'm using Haas, Hurco, and Bridgeport milling machines at work, so it's not exactly foreign territory. But this problem has me bamboozled.

Any ideas (besides getting another machine)?

[RCaffin]

Beats me. A randomly loose connector is about all I can think of.
All I can suggest it that you need a CNC expert - HW I think, to sit there for a day playing with it. Sorry.

Cheers
Roger

[JPX2000]

Yeah, I went through the whole machine checking for loose rods, drive screws, etc. and really didn't find anything. Replaced the anti-backlash nuts on the X and Y axes. No change.

Thanks for the suggestion. The V90 served me well, and was the perfect size for what I needed.

[JPX2000]

Beats me. A randomly loose connector is about all I can think of.
All I can suggest it that you need a CNC expert - HW I think, to sit there for a day playing with it. Sorry.

Cheers
Roger

Thanks, Roger. I wish I had a whole day to sit down and work this out. All of the connections look and feel good to me. There could be something I missed, though.

[RCaffin]

These things CAN be solved, although it may take some time and what seems to be some pointless exercises.
Make up a fine XY fiducial (a fine cross), bolt it down. Put fine pointer (pin?) in chuck. Align XY zeroes to cross, then cycle one axis back and forth for an hour, Does the machine retain zero?
Cycle both X & Y back and forth - return to zero?
Align X &Y, then move in an XY circle from 0,0 to 5,5 (inches) and check zero.
Cycle Z up and down for an hour: does either X or Y creep?
And so on.
Yes, I do have basic routines for doing these tests too.

Cheers
Roger

[JPX2000]

These things CAN be solved, although it may take some time and what seems to be some pointless exercises.
Make up a fine XY fiducial (a fine cross), bolt it down. Put fine pointer (pin?) in chuck. Align XY zeroes to cross, then cycle one axis back and forth for an hour, Does the machine retain zero?
Cycle both X & Y back and forth - return to zero?
Align X &Y, then move in an XY circle from 0,0 to 5,5 (inches) and check zero.
Cycle Z up and down for an hour: does either X or Y creep?
And so on.
Yes, I do have basic routines for doing these tests too.

Cheers
Roger

This sounds like a pretty good way to check the accuracy to me. However, I'm not real clear on what you mean by 'cycling' the machine. Do you mean jog it, or put a program into MDI and run it from there? Also, making a fiducial and bolting it down - this suggestion eludes me. Make it from what? Align it on an edge or in the center of the cross? Could you explain this a bit more? Sounds interesting.

[RCaffin]

Take a bit of stif card and draw a cross on it with a fine point biro. Tape that down to one side of the mill table.
Put a thin pointy thing in the spindle. If all else fails, try a 1 mm drill bit or a fat darning needle. Make sure it is centred when you spin the spindle.
Align the pointer to the cross and zero the X & Y DROs.
Then run a tiny program like this (inches assumed):

m98 p10 L20
m30
%------
o10
g0 x5
g0 x0
m99

This will cycle from 0,0 to 5,0 and back 20 times. See if it ends up at the cross.

Cheers
Roger

[JPX2000]

Take a bit of stif card and draw a cross on it with a fine point biro. Tape that down to one side of the mill table.
Put a thin pointy thing in the spindle. If all else fails, try a 1 mm drill bit or a fat darning needle. Make sure it is centred when you spin the spindle.
Align the pointer to the cross and zero the X & Y DROs.
Then run a tiny program like this (inches assumed):

m98 p10 L20
m30
%------
o10
g0 x5
g0 x0
m99

This will cycle from 0,0 to 5,0 and back 20 times. See if it ends up at the cross.

Cheers
Roger

Thanks for the tip, Roger. I'm trying to get the unit to accept a sub-program, so I have to go back and look at the capabilities of the software.

[RCaffin]

Oops. That code will run on any NIST-compatible machine, but I do not know about other machines.
So the principle is there, but the implementation may be different.

Cheers

[JPX2000]

Oops. That code will run on any NIST-compatible machine, but I do not know about other machines.
So the principle is there, but the implementation may be different.

Cheers

OK, back at it. I ran various programs in different areas of the table. The gantry was actually stalling and then jerking during the simulated cut of a circle. The Y axis definitely has bigger problems than the X. I attached a photo to show a ring that's 1" ID X 2" OD. You can see that the ID is pretty close to the programmed dimension. The 2" OD, however, didn't even start at the programmed position, but at least it was consistent in sucking at being non-concentric through the 10-cycle run.

Backlash was checked, over tightening was checked, loose motor connectors, etc. I can't find binding anywhere. Oh, and since there was no actual cutting going on (had a dried up plotter pen in the spindle which, ha ha, I didn't realize was out of ink), I certainly can't blame this bad run on a sloppy collet or a flexing spindle mount. The quest continues.

[he1957]

Looks like one of the drive shafts is moving/sliding. If the Y is top/bottom then look at Y. Your backlash test may not be exerting enough pressure to show it. Try using RCaffins RTZ (ReturnToZero) loop to test it in a repeating fashion and at a moderate velocity. Put a pause between direction changes to allow time to see the DTI. The idea is to move quickly and stop suddenly then change direction.

[JPX2000]

Looks like one of the drive shafts is moving/sliding. If the Y is top/bottom then look at Y. Your backlash test may not be exerting enough pressure to show it. Try using RCaffins RTZ (ReturnToZero) loop to test it in a repeating fashion and at a moderate velocity. Put a pause between direction changes to allow time to see the DTI. The idea is to move quickly and stop suddenly then change direction.

Thanks, I'll try that. It's still weird how the non-concentricity of the ID and OD is consistent, though. Maybe a setting somewhere?

[he1957]

It looks mechanically off center (in one axis by the look of the image). Whatever is moving (that shouldn't) doesn't always move but does. How many times were those circles drawn?

Draw a third and fourth circle with proportionally larger diameters using the same centre and see if the offset(ness) are proportional.

Try drawing a square repeatedly (say) 5 times over itself, then move it by 5mm in X and Y, repeating to cover 4 quadrants - maybe that might help identify but guessing Y shaft.

If the machine has Acme Lead Screws, check the one on the right, if it has a split brass nut it might be the split section is fractured.

[RCaffin]

Been off the air for a few days. The HUGE tree on the phone line has now been removed.

Cheers
Roger

[JPX2000]

UPDATE: Now that I've had some relief from 12-hour workdays, I have had time to do some research on the possible causes of my problems with machining using the Fireball with the Planet CNC Mk.1 controller.

I reset the controller and reinstalled the Probotix settings file. Tested by cutting circles, squares, and zig-zag patterns, but really no change. Checked connections again to make sure that the stepping motors were getting power. Looked good there. Took a program written in Vectric Cut 2D, used the G code and put it through simulation on a HURCO at work. No problems.

So, I've taken the frame work completely apart and inspected everything for cracks, warpage, bending, etc., and put it back together. Cleaned all electrical contacts, ran system checkup on my dedicated PC, and fired it up again. Same problem. It started out not as bad, but has progressed to where X and Y axes are 'drifting' further than before. I believe unshielded cables, and having too many devices connected to the same power strip must be part of some kind problem, but unlikely the root cause.

With Christmas just a couple of days away, I'm going to look at getting either a new USB connectable controller, or splurge and get a more sturdy build altogether. This was NOT meant to be another hobby unto itself, but as a means to support the hobbies which I already have. I'm moving my workspace around to a different spot in the basement, so that's the entire holiday vacation right there.

My sincere thanks to the folks replying to this thread. Merry Christmas, Happy New Year.

Joe