[4]

[CNC] Another Backlash Problem

[3]
[10] Like what you see?
Click here to donate to this forum and upgrade your account!

JimDawson

Global Moderator
Staff member
H-M Platinum Supporter ($50)
Joined
Feb 8, 2014
Messages
7,155
Likes
5,401
#1
I was making a servo motor mount today and found a bit of a problem with my machine. :mad:

Cutting pressure was minimal, 0.250 end mill, 0.025 finish pass, 0.60 DOC with a spiral lead in, 0.750 mild steel. This is acceptable for this particular part, but I need to cut some bearing pockets for the same project and I'm going to have to bore those because those have to be right.
1534571985595.png

It has always interpolated circles about 0.001 out of round but this time it's really bad, about 0.008 out of round. So time to do some checking.

I noted about 0.010 backlash in the X axis so I'm starting there. First time I have noted any backlash in the X axis, the Y axis has had known 0.004 backlash for some time, but the computer was compensating for that, pretty much.

First check the end play in the lead screw. Put an indicator on the end of the ball screw to check the end play.
1534573400473.png

With the table locked, I was able to get a max of 0.005 end play. But this was by putting much more load on the ball screw than would ever be on it in normal operation. I noted about 0.0003 end play with the table unlocked.

I set up my little USB camera (the small indicator base in the picture above) so I could watch the DRO and the dial indicator at the same time. I have a camera function built into the CNC program. Here is a screen shot.
1534573332853.png

The bearing on the left side is just a support bearing and floats in the housing. The thrust bearing, back to back angular contact bearings, is on the right end of the table.
1534573678774.png

The nut in the center of the handwheel sets the preload on the thrust bearing, I tightened it a bit and got the loaded backlash down to about 0.004, better but not good enough. So tomorrow I'll go into the housing and see what I can do about setting the preload. I'm guessing a shim behind the bearing set will do it.
1534574000529.png

I put an indicator on the case to look at the loaded deflection and at max load it was only about 0.001 about 8 inches from the bearing, so that is acceptable.
1534574159134.png

But, removing the endplay in the ball screw is not going to solve the whole problem. There is still about 0.010 backlash in the ball screw. As it turns out the ball screws are single nut (I think, not totally sure) so not adjustable for wear, so I have a few choices. 1) install larger balls and hand fit the preload, you can buy balls in 0.0001 increments, 2) add a second nut, which I may do in any case, 3) sent the ball screws out for rebuild, and 4) last but not least, buy new ball screws and nuts.

I have some time to think about this, I can't tear the mill down until I get the current project out the door so it will be a couple of weeks before I can get to this. And I'll just have to bore any precision holes, the linear positioning for bolt patterns, slotting and profiling is fine and within tolerance due to the inherent backlash comp in my system.

Stay tuned and I'll post the ultimate solution to the problem when I can get to it.
 

Richard King 2

Master Machine Tool Rebuilder & Instructor
H-M Supporter - Commercial Member
Joined
Feb 1, 2018
Messages
409
Likes
401
#2
Have you adjusted the gibs on your machine?
I would guess that the Y axis has some wear in it too. Depending on the brand of the ball screw / nut some nuts have adjustments. They are a 2 piece nut with a small ratchet on it and you can loosen the ratchet stop and turn the other nut out a little.

This will help some but if the screw itself is worn in the middle it will bind on the ends. Adding new balls so it will track higher on the ball screw would help if the screw and nut are OEM and the BS it self isn't worn to much. I would suggest you clean the machine up before dismantling anything. If you have never opened up a ball nut I would suggest you do it in a cleaner room and put a white terry towel on a table. Be prepared for dozens of small ball bearings to fall out. If you don't use the towel the balls will fall on the floor and you will never find them. Some ball screws have different size balls, so be sure to watch or record the order they come out.

Today you can buy new ball screws on Ebay, so you may want to just buy the kit with both screws. Thanks for showing us the project :)
Time is money and if you have a lot of work to do you may want to replace the BS's . If your slow then go for it. Rich
 

TomS

Active User
H-M Supporter - Gold Member ($25)
Joined
May 20, 2013
Messages
1,784
Likes
731
#3
I'm going to watch this thread closely. I've been chasing backlash on my mill for quite some time and after much trial and error got it down to .001" in X and Y. My problem is I had to back off the gibs to the point where the table sags quite a bit at the extents of the X travel.
 

JimDawson

Global Moderator
Staff member
H-M Platinum Supporter ($50)
Joined
Feb 8, 2014
Messages
7,155
Likes
5,401
#4
Richard- I have not adjusted the gibs for some time. I'll check into the table play today. The Y axis BS has always been a bit loose (about 0.004) but this X axis thing is new. Yes, the machine needs a bath. :) And it will get cleaned up before I tear it down, I need to pull it out anyway to get the X BS out.

I'll profile the ball screws before I make a final decision on the best way to proceed. If I see quite a bit of wear in the center I'll either send them out for regrind & rebuild or just buy new ones. I've repacked a few ball screws and have lost a few bearings. :) The good news is that ball bearings are cheap. If I buy new I will definitely go with double nuts.

I can't tell if these are double nuts, but the housing doesn't seem long enough to suggest that they are double. I can't find any documentation on this machine so I won't really know until I get into it deeper. The machine is a 4V manufactured by Topwell, but it was retrofit by Eagle in the USA, and Eagle is no longer in business. Eagle imported and did a number of Anilam CNC conversions in the mid 80's. I think this machine is an 89, not sure if Topwell supplied the machine with ball screws. The ball screws are Hiwin, and I have a part number from the nut, but that number doesn't cross with any current number.

Tom- Chasing that last 0.001 will drive you nuts. But having the gibs that loose doesn't seem right. If you've tightened up the thrust bearings, the problem may be ''spring'' in the system. Or you just have some slack in the ball nut and maybe oversize bearings would be the answer. https://bearingballstore.com/index.php?main_page=index&cPath=391
 

JimDawson

Global Moderator
Staff member
H-M Platinum Supporter ($50)
Joined
Feb 8, 2014
Messages
7,155
Likes
5,401
#5
A little update, I added a 0.005 shim to the X axis thrust bearing and got rid of most of the end play, now down to about 0.002 max under locked table conditions. Almost not readable under normal conditions and I can feel a little preload on it now. It improved the overall backlash a bit. So I'm going to say that is taken care of. But it still doesn't fix the backlash in the ball screw.

There is almost no detectable end play in the Y axis.
 

TomS

Active User
H-M Supporter - Gold Member ($25)
Joined
May 20, 2013
Messages
1,784
Likes
731
#6
Tom- Chasing that last 0.001 will drive you nuts. But having the gibs that loose doesn't seem right. If you've tightened up the thrust bearings, the problem may be ''spring'' in the system. Or you just have some slack in the ball nut and maybe oversize bearings would be the answer.
I've checked my ball screw bearings mounts and there is no detectable movement. My ball screws and double nuts are asian imports. I suspect they are the source of my backlash. Snugging up the gibs increases load on the ball screw/ball nut and my backlash increases. Not complaining as I was fully aware when I purchased them that they were not top quality. And thanks for the link for oversize balls. That may be an option.
 

JimDawson

Global Moderator
Staff member
H-M Platinum Supporter ($50)
Joined
Feb 8, 2014
Messages
7,155
Likes
5,401
#7
My ball screws and double nuts are asian imports. I suspect they are the source of my backlash.
If you have double nuts than all you have to do is adjust them until you get the proper amount of preload to eliminate any backlash in that area.
 

TomS

Active User
H-M Supporter - Gold Member ($25)
Joined
May 20, 2013
Messages
1,784
Likes
731
#8
If you have double nuts than all you have to do is adjust them until you get the proper amount of preload to eliminate any backlash in that area.
True. They have a shim between the two nuts. All I need to do is figure out how to disassemble them so I can adjust them.
 

JimDawson

Global Moderator
Staff member
H-M Platinum Supporter ($50)
Joined
Feb 8, 2014
Messages
7,155
Likes
5,401
#9
The next time you decide to tear into your machine take a picture of the ball nuts and post it. There are a few different types so the way you attack the problem depends on what you have. On Alloy's Shizuoka we worked a little magic on the X ball nuts to make them easily adjustable, they were the shim type also and nothing we did would adjust them correctly. But we removed the shims and added 4 grub screws to the securing flange, this allowed us to adjust to the correct preload.

You can just see 3 of the four grub screws a the 3, 6, and 9 o'clock positions. Sorry, this is the only picture that I have of the ball screw, we didn't get any when it was on the bench.
1534644481999.png
 

TomS

Active User
H-M Supporter - Gold Member ($25)
Joined
May 20, 2013
Messages
1,784
Likes
731
#10
The next time you decide to tear into your machine take a picture of the ball nuts and post it. There are a few different types so the way you attack the problem depends on what you have. On Alloy's Shizuoka we worked a little magic on the X ball nuts to make them easily adjustable, they were the shim type also and nothing we did would adjust them correctly. But we removed the shims and added 4 grub screws to the securing flange, this allowed us to adjust to the correct preload.

You can just see 3 of the four grub screws a the 3, 6, and 9 o'clock positions. Sorry, this is the only picture that I have of the ball screw, we didn't get any when it was on the bench.
View attachment 274169
My apologies for hijacking your thread. Say the word and I'll start my own thread. Here is a picture of my X and Y double ball nut. The shim is between the two nuts and underneath the tie bar. It looks like the two nuts can be separated by removing the tie bar then unscrewing one of the nuts.

20170207_114005_resized.jpg
 

JimDawson

Global Moderator
Staff member
H-M Platinum Supporter ($50)
Joined
Feb 8, 2014
Messages
7,155
Likes
5,401
#11
You're not hijacking the thread, this is all good information for others down the road. :)

Yup, just take out the tie bar and unscrew the nuts. Add a shim and screw it back together. That is the same configuration as the Shizuoka has, and we tried to add an aluminum foil (0.002) shim. That was not successful, it just crinkled up. (It was a Sunday afternoon and that's what we had) :) That's when I came up with the idea of the grub screws because of the way the nut mounted in the housing. A hardened SS shim or something a bit tougher than aluminum foil would have worked I think. Available from McMaster.

I have always liked the idea of a screw adjustment, that way you can set it to what you want, and is adjustable in the future to compensate for wear. In your case, that might require a modification to the housing. That's what I will do with mine if needed.
 

Richard King 2

Master Machine Tool Rebuilder & Instructor
H-M Supporter - Commercial Member
Joined
Feb 1, 2018
Messages
409
Likes
401
#12
You could cut a shim in 1/2 ( ) and try sliding it in there. The Grub screw is brilliant :) Van Norman use to do that on OEM headstock spindle bearings for their Crank Grinders. Was real handy when you adjusted the angular contact thrust bearings. Great thread :) Rich
 

JimDawson

Global Moderator
Staff member
H-M Platinum Supporter ($50)
Joined
Feb 8, 2014
Messages
7,155
Likes
5,401
#13
Well I MAY have found the problem. I did a bunch of comparative measurements this morning and found that it was also giving me oversize parts on linear profiles, but only on the X axis. And always 0.010 oversize no matter the size, so it's not proportional to length as would be with the scale factor incorrect. It was adding 0.005 to each side of the part, not just adding 0.010 on one side as you might expect from backlash. I checked the G code and compared the tool path to the drawing and the part, the G code and drawing agreed. Then I checked the table travel against an indicator and found I was getting some strange readings, the DRO and the indicator didn't agree. Hmmmmm..... Backlash is about 0.002 so taking that into account it still didn't explain the readings I was getting.

Keep in mind that this problem developed rather suddenly, it just reared its ugly head the other day.

So I pulled the mag scale read head, cleaned it and reinstalled with 0.005 clearance, I had been running it at 0.015 clearance. I checked the table travel with the indicator again and it seemed to agree with the DRO now. Then did a test cut on the bearing pocket I'm working on. The test cut measurements: X axis = 2.250, Y axis = 2.250 :encourage:

I need to do a bit more testing before I'm going to say it's fixed, but it looks promising.
 
Last edited:

JimDawson

Global Moderator
Staff member
H-M Platinum Supporter ($50)
Joined
Feb 8, 2014
Messages
7,155
Likes
5,401
#14
I think it's fixed, at least as far as I can go for now. I made a number of 0.010 test cuts using a few different cutting scenarios and the pocket is running about 0.0004 out of round at the 4-10 o'clock position but otherwise on the programmed size and round. The out of roundness at the 4-10 position has always been there, it's better now than it was. This also eliminated software rounding error as a possibility, there is a lot of math going on internally to cut circles and rounding errors start adding up.

I can live with it for now. When I complete the current project I'm going to tear into the machine and fix the ball screws.
 
Last edited:
[6]
[5] [7]
Top