[lazyman8065]

Hello,

After a lot of reading on the cnczone forums, I bought an X2 clone from Harborfreight and have outfitted it with a ballscrew CNC kit from CNC Fusion. I decided against buying the pre-loaded ballnuts.

Initially, I just used the CNC'd machine as a glorified drill press for drilling holes in PC boards. Now I am trying to actually do some CNC milling.

In Mach3, I have tried the "Cut Circle" wizard. Using a 0.25" diameter end mill, I tried cutting a 1" diameter circle... and ended up with an oval that is 0.975" wide and 0.940" tall.

From what I have read, my problem is almost definitely backlash. I set up my dial indicator on the X axis and found backlash of less than 0.001". I then tried pushing the X axis back and forth. There's the problem! It can move side to side by about 0.012". However, under no load, this slop does not appear to cause any backlash.

Does anyone know what my problem is? Is it backlash? And how do I fix it? Do I need to go to pre-loaded ballnuts?

Thanks,
Kyle

[jalessi]

Kyle,

Have you checked your machine with a dial indicator first?

It is possible your "steps per" settings in motor tuning may be off by a small percentage.

Jeff...

[lazyman8065]

Hi Jeff,

Thanks for the reply. I check the machine with a dial indicator, and that was not the problem. After a bit of checking, I've found out that problem is almost definitely backlash. I'm not sure what I was doing wrong before, but now I'm measuring X axis backlash of 0.012" and Y axis backlash of 0.027".

I've stripped apart my machine to look for the problem. I haven't found anything on the X axis, but there was a loose bolt on the Y axis that might be causing some of the problem.

Does anyone know if 0.012" is a reasonable amount of backlash for a ballscrew kit from CNC Fusion?

Will backlash compensation in Mach3 be able to get rid of this? If not, do I have to get oversized balls for the nut?

Thanks,
Kyle

[Seb]

I set up my dial indicator on the X axis and found backlash of less than 0.001". I then tried pushing the X axis back and forth. There's the problem! It can move side to side by about 0.012". However, under no load, this slop does not appear to cause any backlash.

Are your gibs adjusted? If they're not, that might give your table some wiggle.

[jalessi]

Kyle,

Those numbers are gigantic for a new ball screw.

Are you sure its not coming from the bearings in the block supports?

Jeff...

[Tboh]

Hello,

From what I have read, my problem is almost definitely backlash. I set up my dial indicator on the X axis and found backlash of less than 0.001". I then tried pushing the X axis back and forth. There's the problem! It can move side to side by about 0.012". However, under no load, this slop does not appear to cause any backlash.


Thanks,
Kyle

Interesting problem,
Can you explain the movement a little clearer? Is the movement causing this seperate from the nut - screw (backlash)? Are the gibs adjusted well? Is it consistent?
Tboh

[jalessi]

Kyle,

If the problem you are having was backlash you would have flat spots at 12 o'clock, 3 o'clock, 6 o'clock and 9 o'clock inside the pocket.

Jeff...

[SCzEngrgGroup]

Hi Jeff,

Thanks for the reply. I check the machine with a dial indicator, and that was not the problem. After a bit of checking, I've found out that problem is almost definitely backlash. I'm not sure what I was doing wrong before, but now I'm measuring X axis backlash of 0.012" and Y axis backlash of 0.027".

I've stripped apart my machine to look for the problem. I haven't found anything on the X axis, but there was a loose bolt on the Y axis that might be causing some of the problem.

Does anyone know if 0.012" is a reasonable amount of backlash for a ballscrew kit from CNC Fusion?

Will backlash compensation in Mach3 be able to get rid of this? If not, do I have to get oversized balls for the nut?

Thanks,
Kyle

0.012" is a completely unreasonable amount of backlash for even the worst ballscrew conversion. If they're single nuts, expect backlash to be 0.001-0.005". Sources of backlash can be any or all of:

1) Backlash between the leadscrew/ballscrew and nut. This should generally be no more than about 0.005" for a single nut ballscrew, less than 0.001" for a double nut ballscrew, and perhaps 0.010" for a leadscrew with adjustable nut.

2) Poor mounting or inadequate preload in the leadscrew/ballscrew thrust bearings. This is often adjustable by tightening the nut on the end of the screw. Getting this to zero requires GOOD bearings, well-designed mounts, and proper adjustment.

3) Poorly designed bearing supports. The bearing supports must be very stiff, and must be VERY securely mounted. Cast aluminum mounts can have an amazing amount of flex. The X axis on my knee mill had 0.006" of backlash due entirely to flex in the cast aluminum bearing support. Replacing it with a home-made one made from 1" thiuck 1018 steel completely cured that problem.

4) Overly loose gibs. This allows the table to tilt as it moves, rather than sliding as it should.

5) Excessive friction due to poorly fitted sliding surfaces, poor or overly tight gibs. Generally, you want to tighten the gibs until apparent backlash increases, then back it off a bit. The apparent backlash will increase when the gibs are tightened as the added friction will cause other things (bearings, supports, leadscrew, etc.) to flex more.

Tracking down the source of backlash is time-consuming, and requires going through each piece of the system one at a time, measuring how much it's moving, and eliminating any undesirable movement.

Contrary to (apparently) common belief, software *CANNOT* correct for backlash. The best it can do is partially mitigate the effects, but it is incapable of eliminating the side-effects. Even with software backlash correction, you will still end up with "divets" where the tool stops, or changes directions in a cut, or anyplace the load on the work changes direction. The cutting forces will push the work one way or the other within the range of the backlash, leaving that much uncertainty in the actual tool position. You will see this most clearly by cutting a circular pocket. You will find you have "divets" in the circular profile at 90, 180, 270 and 360 degrees, where the cutter cuts slightly outside the desired circular perimeter. If you want to get truly good results, you MUST fix the hardware to eliminate the backlash.

Regards,
Ray L.

[philbur]

You forgot 6) sloppy motor to ballscrew coupling.

Phil

0.012" is a completely unreasonable amount of backlash for even the worst ballscrew conversion. If they're single nuts, expect backlash to be 0.001-0.005". Sources of backlash can be any or all of:

1) Backlash between the leadscrew/ballscrew and nut. This should generally be no more than about 0.005" for a single nut ballscrew, less than 0.001" for a double nut ballscrew, and perhaps 0.010" for a leadscrew with adjustable nut.

2) Poor mounting or inadequate preload in the leadscrew/ballscrew thrust bearings. This is often adjustable by tightening the nut on the end of the screw. Getting this to zero requires GOOD bearings, well-designed mounts, and proper adjustment.

3) Poorly designed bearing supports. The bearing supports must be very stiff, and must be VERY securely mounted. Cast aluminum mounts can have an amazing amount of flex. The X axis on my knee mill had 0.006" of backlash due entirely to flex in the cast aluminum bearing support. Replacing it with a home-made one made from 1" thiuck 1018 steel completely cured that problem.

4) Overly loose gibs. This allows the table to tilt as it moves, rather than sliding as it should.

5) Excessive friction due to poorly fitted sliding surfaces, poor or overly tight gibs. Generally, you want to tighten the gibs until apparent backlash increases, then back it off a bit. The apparent backlash will increase when the gibs are tightened as the added friction will cause other things (bearings, supports, leadscrew, etc.) to flex more.

Tracking down the source of backlash is time-consuming, and requires going through each piece of the system one at a time, measuring how much it's moving, and eliminating any undesirable movement.

Contrary to (apparently) common belief, software *CANNOT* correct for backlash. The best it can do is partially mitigate the effects, but it is incapable of eliminating the side-effects. Even with software backlash correction, you will still end up with "divets" where the tool stops, or changes directions in a cut, or anyplace the load on the work changes direction. The cutting forces will push the work one way or the other within the range of the backlash, leaving that much uncertainty in the actual tool position. You will see this most clearly by cutting a circular pocket. You will find you have "divets" in the circular profile at 90, 180, 270 and 360 degrees, where the cutter cuts slightly outside the desired circular perimeter. If you want to get truly good results, you MUST fix the hardware to eliminate the backlash.

Regards,
Ray L.

[SCzEngrgGroup]

You forgot 6) sloppy motor to ballscrew coupling.

Phil

You're right! :-)

Regards,
Ray L.

[Crevice Reamer]

You forgot 6) sloppy motor to ballscrew coupling.

Phil

Yes! a loose setscrew get's my vote. Also, make sure that if you move each axis in a straight line an indicated 4 inches, the axis actually MOVES exactly 4 inches.

CR.

[lazyman8065]

Yes! a loose setscrew get's my vote. Also, make sure that if you move each axis in a straight line an indicated 4 inches, the axis actually MOVES exactly 4 inches.

CR.

Thanks for all of your ideas!

I decided to start with the Y axis, since there's no point in fixing up the X axis and then removing it to fix the Y axis.

Of the six items mentioned, I found:

1. Don't know how to measure this.
2. There was about 0.015" play in the mounting of the ballscrew to the base of the mill. Tightening the nut eliminated it.
3. This might be a problem. The kit I bought uses milled aluminum mountings. Also, the Y axis ballscrew is only secured at the front (the rear end is floating).
4 and 5. I adjusted the gibs tighter and looser. This eliminated about 0.002" of backlash once I found a happy medium.
6. There doesn't appear to be any play in the ballscrew to motor coupling.

Using all the great advice you guys gave me, I got down to about 0.010" of backlash in the Y axis.

To try to get the backlash figure lower, I decided that I would dismantle the Y axis and clean the ways. As I was moving the saddle, I heard a "plink" as a ball fell out of the ballscrew.

Now I've got two questions: how do I re-insert the balls into the ballscrew? Also, since I'm going to be having to do it anyway, is it worth buying oversized balls now?

Thanks,
Kyle

[Dougal]

I've recently fitted up my CNCFusion kit to X and Y and have small amounts of backlash in each. I haven't yet meausured the amount but it's in the screw to nut interface.
You can see the screw turning and moving inside the nut.

I haven't got my machine powered up yet, still in the build phase.

[Crevice Reamer]

Some good reading here:

http://www.precisionballs.com/Ball_Screw_Repair.html

CR.

[lazyman8065]

Thanks for the information. I've decided to buy oversized balls. I'll post an update once I've installed the ballscrews.

Kyle

[Dougal]

Thanks for the information. I've decided to buy oversized balls. I'll post an update once I've installed the ballscrews.

Kyle

Will you be buying just the balls and reloading them yourself or new nuts with oversize balls?

I've just noticed CNC fusion do have a "preloaded" ballnut option. Would have taken it if I knew.

[mkenney]

Any update on your progree? I'm working on the same machine and have it down to about .005-.007. Sounds like people are shooting for .001-.002 region? I still have a lot of learnning on what you have to do to get the backlash out!!

Mark [mkenney]

[Dougal]

Any update on your progree? I'm working on the same machine and have it down to about .005-.007. Sounds like people are shooting for .001-.002 region? Still have a lot of learnning on what you have to do to get the backlash out!!

Mark [mkenney]

CNCFusion are going to sell me some slightly bigger balls. It'll be a fun day when I try to install those.

[SCzEngrgGroup]

CNCFusion are going to sell me some slightly bigger balls. It'll be a fun day when I try to install those.

Wow! I'm impressed! Not many men are secure enough to admit, in a public forum, that they need bigger balls.... :-)

Regards,
Ray L.

[Dougal]

Wow! I'm impressed! Not many men are secure enough to admit, in a public forum, that they need bigger balls.... :-)

Regards,
Ray L.

Never thought I'd need soo many too.:rainfro: