Trouble with round pockets not cutting round

[Dick]

I'm running a PRS Alpha with 2.2 hp spindle, 1 year old.
Operating with shopbot link.
Up until now my hinge pockets have not cut accurately, so I am trying to analyze the problem.
I have run some tests outside of shopbot link and discovered the same problem when cutting simple holes through Partworks.
The holes are not round and have flat/straight portions along the x axis.
My first thought was a loose Y carriage, but having tighten that up, still the same result.
Second thought was bit deflection.
I have tried slowing down the cut and only cutting an 1/8" deep, but still have the problem.
I have tried raster and offset.
I could set a separate tool path for the final size for the trials in Partworks, but I don't think that's possible in shopbot link.
1/4" compression bit, 10000rpm, 10ipm
I have found other threads on the forum that discuss similar issues, but none that end with an answer that I haven't already tried.
Photos of trials - http://www.talkshopbot.com/forum/attachment.php?attachmentid=16668&stc=1&d=1353089414

Thanks

[Rick W]

Collet?????

Rik

[khaos]

Shot in the dark.

You could put a short pencil in the collet after some tape or trim to get the right size. Then tape the collet so it doesn't move run your circle cut again without turning on the motor of course. If you still have the issue its not the collet.

BTW, putting some soft foam under the panel you are 'drawing' on is a cheap way to get spring tension effect at the tool.

[tomwillis]

Rich - this probably doesn't help much since my machine is a 2005 PRT Alpha - but I had very similar issues. Turns out - it was multiple issues. First, in the software (which is different than yours) - I had to check a box - called Corner Cleanup which allowed the bit at the end of the run to make a full 360 circle. Additionally - it also depended upon the ramp-in procedure - whether it was a helix, linear, or none. And lastly - the post was incorrect - because I was getting abnormally large arc errors....where I and J in the G-code were not within tolerance.

[Brady Watson]

There appears to be a deviation at 6 & 12 'O Clock. I'd put my money on a pinion gear being loose or not fully engaged in the rack. Drop the motors (X & Y only) and check that the grub screws are cranked on tight enough to bend an 1/8" Allen wrench.

Then, re-seat the motors into the rack pushing up pretty hard while tightening. Then with the control box on, wiggle each side of the gantry and the YZ car to see if you can feel (not hear) any slop in the system. If you feel slop, re-seat that motor into the rack again until it is tight.

If you get deviations from the programmed path in different software, and your move speeds are reasonable, then something is going on mechanically.

-B

[Dick]

Thanks for all of the quick responses.
It looks like Brady nailed it.
The Y motor gear had a lot of slop. One of the grub screws was not even set, let alone tight. There was probably .030 play (outer edge) of the gear.
After tightening there is still some slight slop in the gear/shaft, maybe .010 or less.
I also tightened the x motors.
The cut improved but not totally.
I will contact SB about the remaining slop in the Y motor shaft.http://www.talkshopbot.com/forum/attachment.php?attachmentid=16671&stc=1&d=1353159804
I am hoping that this is also the reason that some of the tenons (pocket cuts) that are cut during SB link files have a slight offsets, resulting in sloppy joints.