Zero backlash (or close enough?)

D

DMS

Active User
Registered
Joined
Aug 5, 2011
Messages
1,567
So, I have been working on a X-Y ballscrew conversion for my knee mill. After spending the last several months designing and machining parts I finally assembled everything earlier this week, only to find way more backlash than I would like (around 4-5 thousands on the X axis). Walked back into the house, my head down, dejected and tired. I guess I was just expecting it to be perfect the first time round (silly me ;)

So after sleeping on it, I came up with a plan. Today I increased the preload on the Y axis ballnuts, and added a shim between the bearings in the bearing block to decrease radial play. Backlash on Y is now down to just under 0.001", which I can live with , but I'm not sure if I'm happy with. So, what's the deal, what do you guys consider a "reasonable" amount of backlash in a CNC machine? For what it's worth, I think the remainder is from spring in the machine components, not sure how much better I can do.

I still have to rip the table off to get at the X axis assembly, but things are improving. At the very least I shouldn't have any more issues with climb milling, and my rapids are up to 120inches per minute (haven't really tweaked them any more to see if I can go higher).
 
It's not uncommon to see that much. In fact, unless the bearings are made for a high preload, when things start warming up it may be too tight. I would want to put the machine in a move loop for an hour and check it again.

Unless you are interpolating bores or closely sized radii, I would be content with 0.001. You'd only see that as small steps on the quadrant points in the circle anyway. Tool pushoff could easily be more than that anyway.
 
Thanks Tony. I've been doing a bit of reading elsewhere, and 0.001-0.002" inches seems to be "good" for anything but a high end machine. I may play with backlash comp once I get the X axis sorted. Overall it's running smooth; I think I just had some unrealistic "expectations". In my mind I was thinking "I would be happy with 5 tenths". Hehe, what a schmuck ;)
 
I'd be very happy with a half....but not really disappointed in 0.001. I have seen and made quite a few parts that location is critical to the point that positioning was more critical than that, but easy enough to work around just by making sure the toolpath approach was the same on all related features. Unless you are doing work that doesn't allow for that, you may have to just work with comp to get what you need. A conversion is likely not going to be quite as accurate as a dedicated design CNC machine anyway, so I'd be content as is. After you run for a while, it would need to be rechecked anyway. Things will change.
 
Hi DMS
If you plan to run Mach3 or 4 Mach has a backlash comp on in it, and it works good. You can almost get 5-6 thou working with it, and .001 is no problem. A thou is'n bad on an Asian mill. You can also repact the ballnuts with larger balls that will realy get the backlash down quick. I repacted mine with balls I got off fleebay cheap. Went up by .0002 each try and the backlash just got better untill it was down to less than .001 but the machine want hold lesss than a thou any way.:allgood:
Congrat's on getting your mill going.
CH
 
Thanks for the kind words.

I am running a dual ballnut setup with belleville springs. One of the things I did to improve the backlash was to increase the preload to 210 lbf (up from 105 lbf) made a huge difference. Also, preloading the bearing blocks brought the play in the bearing blocks down below 2 tenths unless I really crank on things. X and Y are both at around 1 thousands now.

As far as backlash comp, I may play with it. I have been running with the stock acme screws for about a year now, and they had about 12.5 thousands backlash in them ;). I am using LinuxCNC, and it was able to get pretty decent results. I was also thinking of using leadscrew mapping because I used low grade (0.004" per foot) screws. We'll see how that works. I got things tuned tonight. Tomorrow I will try cutting some circles, and see how round they are. I was able to run some pretty nice circles just air cutting; 1 inch diameter at 120ipm and it sounds smooth.
 
I would also but a bit of weight on it to duplicate weight on the table, set up a loop and let it run for a long time.. Do the run in on it, then readjust...
 
I have about .002 in X an Y on my zx45 and it's not an issue for what I build. mach3 takes good care of compensation and if certain areas of a part require real precision like bearing seat locations and sizes, I machine that area larger than needed and true with a boring head to spec. The bulk of most parts is the shape and only a few areas of a part require real precision. I have the cnc build the item then go manual for exacting precision where needed. I think of it like working with castings. If you price real quality ball screws at thousands of dollars then look at what we are using in our home grown machines a compromise will keep you from beating your brains out and quality parts will still be made. it's a matter of knowing your machine.
steve
 
I posted some pictures of the conversion here [thread]18812[/thread].

After some tweaking I am getting about 0.001" on both X and Y, and am satisfied. Everything is smoother, tighter, and rapids are about double what I was getting before. I took a gamble on these cheap screws; they were a bit of work to get working, but ultimately they work quite well. I still have issue of accuracy (these are rated at 0.004" per foot), but I rarely make anything over 6 inches, and at that size I think I can manage a couple thousandths until I map the screws.
 
You must log in or register to reply here.
Back
Top