What's new
What's new

chatter when interpolating inside of circle 96 vf2

steve10778

Aluminum
Joined
Nov 22, 2010
Location
ohio
Needs some ideas here,,,,
96 vf2 having chatter problems when milling id circle. The chatter always occurs at 95 degrees to 120 and 275 degrees to 300 degrees. Milling 1.0 inch deep, .002,.005,.010 stepover ranges, no change chatter every time in the same two spots.
here is what i have done so far:

New spindle
tried different table locations
various materials, steel aluminum
5 flute carbide, 2 flute, 4 flute
speed and feed up and down, no effect
adjusted leveling foot on one side to twist bed, to see if it had any effect(squareness) no effect so put back
checked y axis bearing locknut to see if had worked loose, no.
climb milling or conventional, chatter appears in the same area. ( i would have thought it would have reversed to the other side of the quadrants, but it did not)
ball bar test shows good results
backlash according to ball bar is .00015 or less(with comps on) measured backlash(deflection with full return to indicator zero) is about .0005 to .0008
circularity is very good ( a few tenths), no quadrant marks, except chatter after the 90 and 270 quadrants.

so i started to suspect servo issues, not servo mismatch but maybe oscillation while interpolating, adjusted d gain to stiffen axis up, no change, reverted settings to normal

adjusted p gain, no help, reverted that back too.

did not try any integral, feedforward, or accel parameter tweaks.

Im thinking of just replacing the thrust bearings on both x and y, but im just not so sure.

so what does everyone else think?

thanks,

steve
 
I would look to how the part is mounted or held.

Also, could you post the code you are using. A quick picture of the setup would be great too.

P.S.: I have the exact same machine, 1996 VF-2.

Mike
 
I am assuming the Haas has linear ball ways right? I Goggled a circle with degree's, but for me to explain it a bit easier I am going to think about a clock. It appears the chatter is at the extreme ends of the Y axis in both directions at about 11:00 and say 11:50 and on the opposite side at 5:00 and 5:50. There might be something wrong with the linear guides and not the control An easy way to check this would be to get a granite square, granite straight-edge or parallel and fasten it down on the top of the table, use a .00005" indicator and tram the straightedge as straight as you can get it, fasten it down. Then remove the indicator and move the table to the the areas where the chatter happens and put a .00005" indicator back on the straightedge and jog the Y back and forth (in and out) at several different distances and when you change directions in a feed rate you use when machining and speed it up and see how much the indicator moves or shakes, I am sure you already know how to set a indicator on a surface, so I need not going to try to explain that on here. Then do it in the X direction at the same locations. If it were box ways it would be easier you could loosen the Y axis way covers and mike (mic) the vertical Y axis ways and see if they are thicker in the middle. I also have a friend who had issues with a Kitamura VMC when machining a circle, but his chatter was all the way around and they had to change the control so it took smaller moves when interpulating. Years ago when we were testing controls on retrofits our Tech would put a fine point ball point pen and put it in the spindle and tape a piece of paper on some plexiglas and draw a circle. He said it was a good way to discover problems. I am not a electronics tech, all I do is mechanical repairs, so that might be a bit to simple for your problem...
Is the machine under warranty? You don't say how old it is? Good luck.
PS: Just thought ..do you have another machine you can use to make a good part, maybe a friend ? Machine a slot or edge on X and Y so it will be easy to set up and check squareness Then use that part to check yours with the indicator in the spindle. Make sure the spindle is turned off, or the new Inter-rapid will be history. ha ha.
 
Last edited:
These areas are what I would call "transitional areas." By that I mean that an axis has come to its slowest movement area and is transitioning to the opposite direction. Depending on the size of your circle, especially a large one, it would mean that one axis is nearly stationary and other other is moving at its fastest peak. Whichever axis is stationary, it could have some minor looseness or slop in it.

If you have "0" on your circle at the top or bottom, it is the areas of transition for the "X" axis. If your circle's "0" is an the left or right, it is the transitional area for the "Y" axis.

At the chatter area, the ball screw has just begun to move in the opposite direction and is taking any slop out and would be vulnerable to movement, causing vibration.

I hope this makes some sense, but check the adjustment for the axis in transition.

Mike
 
The chatter always occurs at 95 degrees to 120 and 275 degrees to 300 degrees
checked y axis bearing locknut to see if had worked loose, no.
adjusted d gain to stiffen axis up, no change, reverted settings to normal
adjusted p gain, no help, reverted that back too.

If the chatter is happening at 3 and 9 o'clock check the X axis mechanics and parameters, not the Y axis. You can swap the X and Z axis amps to rule out the amp. Send me the parameters, settings, and program. [email protected]
 
replies to questions...

>>>>loner ind: the feed does not slow down. program is arcs, using ijk, not r (had problems with rounding errors in the past with r values, so i dont use them)

>>>machineit: the current test piece is a piece of fremax 15 8" round, 2" thick, clamped flat to the table with standard clamps, hole is 2" diameter, 1" deep. I am adjusting wear comp to keep running my test. i.e. -.005 write/enter to wear comp, run program. this is all for testing purposes. so no picture needed. the code is just rapid in to depth, lead in tangential and sweep 90' then once around, lead out rapid up. i have changed the start points, but chatter is in the same spots. let me see about posting the code later this evening. Thank you.

>>>rbcmetalwork: it is a new spindle, but i do have a clamprite i can access and i will check it. i believe cat 40 is supposed to be around 2500psi . it does chatter regardless of the tool diameter. it is better with less flutes, and obviously boring (single point) is fine.

>>>Richard King: the chatter would be at the clock position you say, if you are envisioning minutes as divisors of the hour segments :) your idea of checking tram (axis straightness at different areas) is ok, but the ballbar does that as well. neither test is pointing to a gross problem, probably because of the lack of cutting forces. 96vf2

>>>machineit: i know about the transitional areas, the goofy thing is that changing the cut direction does not change the chatter areas (mirror across the axis) it very strange. checked for loose locknuts but they check out ok, and measured backlash is as mentioned above. no quadrant marks in cut.

>>>ken foulks: as mentioned, the chatter is NOT at 3 and 9, but centered at 11 and 5, milling ccw AND when milling cw. it does not flip flop. I like the idea of swapping the amps, also this is a brush machine so i was thinking it could be dithering around the setpoint while interpolating. (i.e. my true circle path has oscillations and they occur when x axis is at near max velocity, or when y is near min velocity) the F factor though is that the position of chatter does not change with all other tests/changes. i will see about sending you a parameter file, settings and program later this evening. i will make a fresh back up and send it to you. are you sure this isnt a warranty issue? haha.

>richard King: his mazak fix i dont think applies, but i did check the tightness of all the bolts on the ballnut, and linear trucks, and screw supports.


i ordered new ballscrew support bearings for x and y, i have to make the spanner sockets and locking tool to set bearing preload, maybe the parts will show up for the weekend. i hate throwing parts at a problem. Ken Foulks, what is the factory spec for backlash? .000050 to .0001? maybe my .000097 and the spongy deflection of a few tenths added together from x and y is my problem, seems sooooo unlikely though.
 
>>>Richard King: the chatter would be at the clock position you say, if you are envisioning minutes as divisors of the hour segments :) your idea of checking tram (axis straightness at different areas) is ok, but the ballbar does that as well. neither test is pointing to a gross problem, probably because of the lack of cutting forces. 96vf2

I think what Richard is describing is what I call "shake." That is movement in one axis when another changes direction. Certainly worth exploring given the symptoms, and not something that shows up on the ball-bar like a blinking red light.

Side note on the ball-bar (not a criticism of the OP, he's trying to isolate the issue): it's an aggregate measuring tool, and will never replace a granite square for isolating certain geometric error. NEVER adjust based on ball-bar indications. It just tells you where to start looking.
 
found the problem, i think

well after much prodding and prying, i found what i believe to be the problem:
if i have an indicator mounted on the bottom of the spindle casting, reading flat against the table( tram on y axis direction) if i push and pull at the spindle nose, i dont see too much happen, but if i go up toward the top of the head, pushing on the sheetmetal higher up on z, i get about a thou of movement. i also get some movement pushing and pulling the spindle along x direction..0008" . what i believe is that the linear rails have wear, allowing the cantilevered spindle to sag ( go out of tram) and be loose. this allows the cutter to oscillate or yaw up and down in cut. It doesnt explain why the chatter happens in the same spot, but i decided to not do anything to fix the problem, because i think changing the rails would be a huge undertaking of both time, tediousness and money. I think to do the job i would have to completely remove the head, remove and replace the rails, align the rails, align the ballscrew, replace the head, etc etc. i dont have a crane. and i sure dont know the cost of the rails and trucks.

comments anyone?

and thanks for the help!

Steve
 
march update:

Ordered a new set of linear rails for the z axis, installed them yesterday after a call to haas about the "apparent" preload. My old rails and trucks were nsk's, the new ones are INA. My old trucks were very stiff to get moving (still have preload and no noticeable movement) new trucks on ina rails slide down almost from there own weight. Haas Factory says its ok, different manufacture, different preload. so i go ahead with the install.

Installing the rails was not bad at all, with a helper about 4 hours. Did one side at time, cleaning and stoning after removal, stone the backs of the new rails, clean clean clean. slid new rail and trucks in between spindle casting and column, lightly tighten mounting bolts, fully tighten wedge screws,
(raise and lower the head to gain access to top half of bolts and lower half of bolts) Then final torque on mounting bolts from bottom to top.

Test cuts: 1018 steel, 2" diameter hole, 1/2" deep, 3/4" flute endmill chatter still there at 11 and 5 o clock. it doesn't matter if i take .02, .050 or .001 step over, the tool screams in these areas. Ran conventional and climb position of chatter does not change.
So i swapped x and y amps, set friction comps to 0, no change. (put these back where they were)

So now i want to prove its not a control problem or something related to x and y working together to make a circle., so i mounted a .500 double cut carbide burr in a er collet holder and ran that around the test hole, 8000 rpm, .0005 stepover, feed was somethin or other... hole was round, and no chatter no quadrant marks, and no noise. surface finish looked nice all the way around, so my thoughts were if i had a jittery y servo i would see it with this test.

My next move today is to disrupt all of the mounting feet and machine level, then re level using the outer 4 feet to adjust to a precision level, then adjust table tram with the two middle feet and do another test cut. Im just thinking maybe i have a soft foot or something. The chatter marks line up on a 130 degree angle which is usually related to squareness, and because changing direction , feed, speed, engagement, tools, place on the machine or height of milling in z, none of which mitigates the problem. So I am going to re qualify the machine level and mounting and see if that gets me anywhere. Im really stumped.

Any other ideas? Please read my first post and replies to see what i have done so far.


thanks

Steve
 
Wow, you spent some serious coin on this issue!

I did not solve my issue. Since I do not see any chatter marks on the holes that I interpolate (when getting down to the finishing), I've not been forced to search for the answer.

Did you have to shim the spindle cartridge when you installed it, in order to get the table tram to zero out? How many shims did you insert, and where?

Did you test the spindle taper for tool contact, even though it is new? Did you rebuild and check the drawbar tension? Do you get tool shank fretting when taking some heavy cuts?
 
there was one .0005 shim, i put it back where i found it, tram was good.

after changing the rails, as i said i did them one side at a time, tram was still good.

i will post tomorrow the results after doing a complete breakdown and re level of the machine.

Drawbar/tool retention is new/checked out. It doesnt chatter all the way around the circle, only in those two vectors do i get chatter, direction of cut doesnt change a thing.

the only time i ever got fretting was plunge milling with a big cutter sticking waaaaay out, and i expected to see it. no fretting with common tools and common jobs.

Thanks for getting back to me, curious what your status was.

steve
 
Find someone in the balancing business.

Sorry, couldn't finish what I was typing :D

Vibration analysis can help pinpoint the machine component that has the resonant frequency. You might even discover a sub-audible beat frequency that is due to some persistent vibration, and that becomes stable in certain cutting conditions.

For example, I had a vibration in one of my lathes, which I suspected was some sort of main motor balance problem. But the issue only arose in particular speeds. Through vibration analysis, we tweaked the balance of the rotor a little bit, and the sweet spots for vibration all disappeared.
 
ok,
-releveled machine dead nuts, with calibrated precision levels, after disrupting everything.
**found a broken bolt on the front y axis ballscrew support. replaced all bolts with new bolts and torqued to spec. Because this is just a support bearing, and not a thrust bearing, this did not fix my problem neither did releveling the machine.

I am starting to lean back towards y axis ballscrew/sleeve bearing assembly. I do have a .0003 lost motion on y, but any is indicative of lost preload. i dont see any axial movement on screw, just lost motion between the screw and the nut. It doesnt explain why the chatter only happens in the same two places, regardless of direction of cut. I do know that i cant be closed minded as to what the problem is though. Y is waving its flag at me again.

I think i want to do some oscillogram test cuts as mentioned by "gar" using a 1 flute cutter in different feed rates and directions, it is a sound principle and good visual test. I guess depending on what i see there on an x cut verse a y cut, verse an x/y 45 degree might show some stuff. we shall see.

I am going to recheck my lost motion on y, if i have no axial play on the screw, and all of the .0003" is between the nut and the screw, is that enough to make it scream like a banshee? I dont have the experience to know for sure. This machine will be fixed this time though. i have it all ripped apart and i cant take it anymore....

any thoughts?
 
Update:
Replaced the Y axis ball screw and sleeve bearing assembly. Once the screw was out it was evident that it had no preload. It wasnt too terrible of a job, it just took some time. After breaking away for a bit to make up two spanner sockets to tighten the nuts, i got the new ball screw in and went back to my indicator checks.

*no Backlash on screw end
*no backlash between screw and nut

I put indicator on table, indicating on front of spindle, with servos turned on pushing and pulling on the table in the y axis direction =.0003" lost motion. A test cut with my goofy 5 flute endmill caused chatter in same suspect spots, althought the length around the circle was smaller. a 1/2" 4flute endmill cut ok, leaving a small quadrant mark. Ended up putting 7 in the backlash comps for Y (7 * .0000293) quadrant marks went away. also with the servos on, indicator from table to spindle, i can turn the ballscrew by hand cw and ccw into the torque of the servo, it will push back but i still show the same amount of lost motion. Soooooooo........

I am ordering a new servo coupling today.

Its getting better!
 








 
Back
Top