Surface finish issues on CNC conversion

[JLeatherman]

I'm just getting going on my CNC-converted South Bend "Fourteen" 14x40 lathe. It's was factory converted to ballscrews and servomotors in the late 80's or early 90's, and recently the control was swapped out for a PMDx board and gecko drivers and I'm running Mach3. I'm having some strange issues with surface finish. It's not that the the finish isn't smooth, but there are regular waves in it. Basically the diameter of a finished part wanders in and out .001"-.002" every 1/4" of length or so. The strange thing is that changing the feed rate doesn't alter the location of these waves. You would think that a slower feed rate would make them closer together and a faster feed rate would make them farther apart, but that doesn't seem to be the case. It's most exaggerated on a high-rake cutter (like for aluminum). I would also note that it seems more inclined to do it when running a g-code program than when I'm just using the jog function. It seems like the x-axis servo wanders in and out about a thousandth depending on the z-axis position. I'm wondering if this is something that can be tuned out? I've included a picture of a pair of aluminum parts I turned (about 1" OD) with a nice CCGT high-rake aluminum insert. It left a smooth finish, but you can see the color striations which correspond to a diameter variance. The dark bands are lower diameter, about .001"-.002" smaller (diametral) than the light bands. Even another light pass yielded the same pattern. Any thoughts? I only recently acquired this machine and most of the converting and setup work was already done.

 

[RManley]

I have this on my denford cnc machine. The leadscrew was found to be bent therefore one rotation of the leadscrew creates a wave. The centrally worn bed doesn't help either as its tight by the outer end of the bed but free by the chuck. My machine needs a bed regrind and a new leadscrew, two things I wont be able to afford for a few years yet . I hope yours isn't as serious.

Rob. :-\

 

[GailInNM]

IF the spacing between the bands is the same as the z axis lead screw pitch then it can be caused by a bad lead screw bearing, a bent lead screw or a problem with the coupler between the servo and the z axis lead screw. That's one place to start. There are others.
Gail in NM

 

[JLeatherman]

Rob,

That's an interesting theory. I hope it isn't the case, but I'll have to check it. The bed is fine, that much I can see for certain. It's a hardened bed, no ridges and no measurable wear, and very little use. The screw could be bent, however. It has covers over the ball leadscrew to prevent debris from getting in, so it will take some disassembly to check the condition of the screw.

I have achieved an acceptable and near-perfect finish with a boring bar on these same parts, however. Granted this was after multiple spring passes (small boring bars flex like a slinky) but the resulting bore is within several tenths over the entire 1" length. The coloring in the pictures below is graphite residue from the test piston. The one on the right shows no, or atleast un-discernable, diameter variations and the pistons fits as though there are none. The boring bar has nearly no rake, of course, although a cemented-carbide OD cutter on another operation did unfortunately yield some diameter variances.

 

[JLeatherman]

IF the spacing between the bands is the same as the z axis lead screw pitch then it can be caused by a bad lead screw bearing, a bent lead screw or a problem with the coupler between the servo and the z axis lead screw. That's one place to start. There are others.
Gail in NM

I'll keep my fingers crossed for a bearing or coupler problem. Thanks for the tips! I suspect the spacing of the striations is the same as the leadscrew pitch, although with the covers in place I can't actually compare them. Certainly makes sense.

 

[MuellerNick]

I had a similar problem during *very* light cuts with my manual Chinese-crap-lathe.
The reason was hard to find, but it was found: It is the handwheel of the saddle. When the handle goes over TDC, it feeds just a tad more towards the head stock (consuming play).
As a ball screw is supposed to be without play, the axial bearing of the lead screw might be the culprit.

But first check wether the lead screw's pitch and the distance of the ridges match.

Nick

 

[JLeatherman]

Ok, so it does indeed seem to be something to do with the leadscrew, leadscrew bearing, or leadscrew ballnut. The striations turn out to be the same pitch as the leadscrew. I did some disassembly and investigation last night with interesting results. It turns out that if I loosen up all the bolts that hold the headstock-end leadscrew support bearing in place the striations go away. This leads to other problems, however, so I can't run it that way. I cut a few inches of test bar while tightening and loosening those bearing screws and you could actually see the problem start and stop. Tightening the upper screws on the bearing support make it worst. I believe that every revolution the leadscrew is applying slight upward pressure on the saddle. That's why it makes no difference what the feedrate is. It's a mechanical issue. I've temporarily reached a happy medium with the bearing support screws just snugged (and one of them removed entirely) and the lathe is cutting a mostly un-noticeable pattern that I can polish out.

It seems to me that there are a couple possible causes for this. I can't really tear the lathe down until I'm done with the current project (a Christmas present) but for now I'm bouncing ideas around. I thought that one of the three bearings in the leadscrew system had to float? You can't build it aligned well enough to run 3 fixed bearings, right? Does anyone know if it's the left support bearing that is typically supposed to float, or is it the ball nut that's supposed to float? It could be that whichever bearing is supposed to float has stopped floating; possibly gotten a big chip lodged in it or a little surface rust. I thought that the ballnut was supposed to be the one that floated axially to isolate the saddle from the leadscrew system? Unfortunately the more likely culprit is a bent leadscrew, probably during moving the lathe.

I can't quite figure out how the leadscrew comes apart on this lathe. I looked at it last night and it has me a bit stumped. The servo motor comes off readily (I had it off last night) but I can't figure out how the left side bearing carrier comes apart. It looks like I have to drop the leadscrew, apron, and both bearing carriers off in a lump and then take it apart after that. A job that big will have to wait until after Christmas for sure. The leadscrew is driven by a nice poly coupler, BTW, and I didn't see anything wrong with it so that doesn't appear to be the problem.

 

[JLeatherman]

Also, upon further reflection (and working from my fuzzy memory) I think the holes that mounted the servo/thrust bearing assembly to the lathe were slotted. Not the ones I loosened last night, those are the screws that affix the thrust bearing to the servo, but the ones that mount the whole drive unit to the headstock. I think they were slotted for vertical adjustment. Maybe the adjustment got bumped? Of course, if the whole drive unit were bumped up then it shouldn't produce a pattern and rather just a constant upward pressure on the leadscrew?

 

[Noitoen]

You could disconnect the x axis and block it with the jib adjustment and then retry to see if the result changes.

 

[bret4]

I remember having a problem something like this on a Bridgeport EZ-path lathe when facing. We would get a pattern across the face when facing. We adjusted the gibs and the problem went away. Haven't had it come back in over 12 years now.

 

[JLeatherman]

Took another look at the lathe last night. I removed the servo again and tried to adjust the headstock end of the leadscrew. While there were in fact adjustments built in (see pics) I discovered that after proper adjustment was obtained the mount was drilled and roll-pinned in place. It could not have shifted during transport and can not be adjusted now.

I also cut a 16" test bar. The pattern was present the entire length. In the middle of the test bar the pattern becomes very faint, but I think this is due to the bar flexing. If I had larger diameter material I believe that the pattern would be present and unchanged for the entire length. It seems too coincidental that it became faint at exactly the center of my test bar and returned as the carriage approached the spindle.

I forced back the leadscrew covers and moved the carriage to both ends. I don't have a long machinist's straightedge, just a long woodworking straightedge, but there was no detectible bend along the length of the leadscrew. I would think that if this were a bent leadscrew that it would have to be bent severely enough to be detectible with even a reasonable straightedge, unless it is bent right at the drive end where it enters the thrust bearing assemble (entirely possible, but I have to remove it to check that). The carriage gibs appear adjusted, but that is tricky to tell on a CNC. I need to remove the servo and turn it by hand, and I may need to remove the enclosure to reach the gibs well enough for a fine adjustment.

 

[bret4]

Are you sure it's not just a gib adjustment on the cross slide like I mentioned? Make it tighter and do a test cut.

 

[JLeatherman]

I've checked the gibs, but I'm not sure how tight they can be on a machine that uses servos and not handwheels. They're not abnormally loose, but I'll admit that doesn't mean much.

On a side note, when I pried back the leadscrew covers I doscovered that this machine has a 1 1/8" leadscrew. The force required to bend it would be immense, and I would expect that whatever caused it would have to have damaged the apron and/or ballscrew covers (which are thin sheetmetal) as well. I can find no damage at all to the lathe, apron (which also is only sheetmetal), or covers. I'd bet that nothng short of lifting the lathe by the leadscrew, or tipping it over onto the apron, could bend it.

The other thing I'm struggling with is the fact that the previous owner never experienced this. I've been corresponding with him and he is as stumped as I am. It would seem that the most likely explanations, a poorly-machined or bent leadscrew for instance, or a gib adjustment problem, should have been a consistent problem before and after the move. As well done as the conversion is, and the fact that the previous owner had no similar issues with it, I have to believe that it's not a problem with the leadscrew alignment, apron alignment, etc. If it's a mechanical problem then it has to be something I jarred loose during the move, or a part that has worn out/failed.

Of course I've been wrong before

 

[Noitoen]

I don't understand how the leadscrew can be blamed. It only transmits movement along the axis. Isn't there any vibration?

 

[petertha]

FWIW. I experienced finish stripes on my workpiece quite similar to yours when my lathe was essentially new. The service guys noted the pitch on the workpice was similar pitch as the leadscrew. He removed the leadscrew & rolled it on piece of glass I had on my workbench. We could see a visible eccentric wobble. A new leadscrew was ordered & they changed out the end bushings same time too I believe. After that, no more striping problem. He figured either the moving guys or original shippper had a strap around the leadscrew or carriage & somehow got bent. Or maybe it was bent at the factory & installed that way. Not sure if this is your issue, but another datapoint hopefully.

 

[bret4]

When you mentioned that the previous owner didn't have this problem that made me think. What I think is that maybe the machine isn't level. It may be possible that this would cause the bed to be twisted just enough to let the carriage not sit quite flat. If it rocks just a little bit that could cause this problem. I would get a precision level and make sure it is sitting level. Even if you already did this when first installed it is still a good idea to recheck level after a while just in case it settled some. With larger lathes like this you have to level them with a precision level, not a level made for carpentry if you want them perfect.

 

[ninefinger]

A bent leadscrew could cause this problem if the gibs are too loose. The stripes are because the tool tip is moving relative to the work centerline (in and out or up and down and therefore cutting less/more material). The only way the leadscrew could cause this is if its moving the saddle - hence the loose gibs theory. Tighten them up and give it another try - seems to be an easy troubleshooting step.

Mike

 

[goldstar31]

I'm merely guessing but why not physically lock ALL the gibs solid and turn using just the top slide advance.
It might be the headstock bearings which are loose or worn.

 

[MuellerNick]

It might be the headstock bearings which are loose or worn.

Then the pattern would not have the pitch of the leadscrew.
Also, on a CNC-lathe, there is no top slide. Normally.

An other reason might be the axial bearings of the leadscrew. I have seen bad Chinese ones that wobbled axially by a few hundredths of a mm. Or there is some dirt under the axial bearing on the leadscrew's side. Would have the same effect.

If you have access to the right end of the leadscrew:
There is a center bore. Place a ball in there (with grease) and put a DI axially to the leadscrew on top of the ball. Rotate the leadscrew one turn and observe the DI. Also check, wether the play (if any) changes over one rotation.


Nick

 

[goldstar31]

Humm Nick, you have something there. My next( daft) question is- was the lathe recondtioned?
In other words, is the leadscrew being pulled up as it moves the saddle?

Norman