• This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn more.
  • Guest,  We want to wish You and Your Family a Healthy, Happy Thanksgiving! Click the "X" at the top right corner to remove this notice)
  • PLEASE: Read the FORUM RULES BEFORE registering!

4

[CNC] Tell Me About Ball Screws

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

cjones6108

Iron
Registered Member
Joined
Aug 20, 2017
Messages
19
Likes
2
#1
Ok, so after reading for several hours on the Hobby-Machinist forums, I have taken the plunge and ordered a PM-932M mill, the very most basic model - no power downfeed, no z-axis motor, no x-axis table feed. The idea is to begin immediately planning to convert it to CNC.

I have noticed other members doing this conversion, using 1600 oz.in steppers on the X and Y axes and 4200 oz.in for the Z-axis. These look like pretty hefty motors, and I'm in the process of sourcing the motors and controllers. I'm thinking Ethernet Smoothstepper or similar, particularly because computers haven't shipped with parallel ports or DB-anything ports for the last 10 years or more.

I realize ball screws are going to reduce friction therefore required motor torque, but it seems like with such large/hefty motors it wouldn't matter that much. So what is the reasoning behind replacing brand-new acme screws with ball screws? And do you replace the X, Y and Z screws - All of them?

caj
 

DAT510

H-M Supporter - Premium Member
H-M Supporter-Premium Member
Joined
Oct 20, 2016
Messages
219
Likes
127
#2
The biggest reason for ball screws is reduction in backlash.

Imagine cutting a circle. As the cutter follows the circular path, depending on the quadrant of the circle one screw will be turning clockwise while the other maybe turning counter clockwise. When moving into the next quadrant of the circle those screws may now have to reverse and turn in the opposite direction. Any backlash will affect the roundness of the circle being cut.

Ball screws have much lower backlash, and preloaded double nut ones have even less.

Hoped this helped.
 

JimDawson

Global Moderator
Staff member
Director
Joined
Feb 8, 2014
Messages
6,427
Likes
4,155
#3
+1 what @DAT510 said. I highly recommend ball screws. Software backlash compensation is a fallacy. It works ok for hole patterns and some other linear moves, but for accurate profiling and circular interpolation you need as close to zero backlash as you can get. Also near zero backlash is required for climb cutting, and most CNC work is climb cut. Better finish, easier on the tool, and normally more accurate. Also a ball screw has about 1/15 the rotational resistance of an acme screw under the same load conditions.

You might look at the ClearPath motors https://www.teknic.com/products/clearpath-brushless-dc-servo-motors/

I did a router retrofit for a customer using these motors and was very happy with the result. They are a drop in replacement for stepper motors, and have a nearly flat torque curve. Stepper motor torque drops off very rapidly with increasing RPM, especially the 1600 and 4200 oz/in motors. The other advantage is that the drive is built into the motor, just connect a power supply and the step & direction signals and you're off and running.
 

cjones6108

Iron
Registered Member
Joined
Aug 20, 2017
Messages
19
Likes
2
#4
OK, that makes sense. Thanks. I'll check out the Clear Path motors.

Now, about "preloading" -- Another term I'm not familiar with. Can somebody explain that?

caj
 

spumco

Active Member
Active Member
Joined
Jan 15, 2016
Messages
112
Likes
62
#6
+1 to everything above.

The Clearpath servos are very nice and having the drivers right on the servos is pretty awesome for reducing your electronics enclosure clutter. However, I'd suggest you read the "So, What's The Catch" page on Teknic's web site that describes some of the compromises they made with the Clearpath servos. You'll see that there's no encoder output. Not a show-stopper, but I eventually want to option to close the feedback loop to the controller and the Clearpath drivers are not capable of that.

The one killer for me is that the cable connections at the servo are not waterproof - not mentioned on that particular page. I run fairly high pressure flood coolant and didn't want to have to cobble up some sort of coolant-proof shield for them.

Otherwise, they're awesome. I saw them demonstrated (and met some very nice Teknics guys) at the NYCCNC open house this year. I simply can't say how professional and helpful the Teknics reps were. And you can't beat the 3-month money-back return policy.

Another option to consider for axis motors are DMM Technology servos. Looks like they're roughly competitive with Clearpath on price for a given motor power/size/speed, but the DMM drives have significantly more features than does the Clearpath. The DMM are more in line with Teknic's Hudson servos and Eclipse drives. I've heard/read nothing but good stuff about them.

My situation is not your situation, of course. You may be doing this on the cheap, so you'll be sticking with steppers - they'll work fine at low speeds. If you consider your PM-932 as basically a Tormach 1100 then you can figure out what size stepper motors the Tormach uses. Your spindle head weighs quite a bit more because of the gearing, of course, but you'll be ditching the gears sooner or later and moving to a 1 or 2 speed belt drive so the weight will be back to (about) the same.

Pay attention to what Jim wrote above - ball screws have much less rotational friction than acme screws, and the weight of the head can back-drive the axis screw. Figure on adding some gas springs or a counter weight to the head because the stepper (or servo) may not be able to hold the head up when they're not powered once you have a ball screw on the Z.

What you don't have is Turcite on the dovetail ways like the Tormach. The friction on your ways will be higher than the Tormach, so maybe add 10-15% of stepper (or servo) motor 'umph' to overcome this deficit. Otherwise I think the Tormach motors should give you a good starting point for choosing a motor.

I'd just caution you that trying to save a few bucks here or there has it's place, but eliminating backlash is not the place to compromise. Ballscrews and pre-loaded ball nuts are your friend.

While you're at it, you might as well fantasize about substituting one of those DMM Tech 1.8kW servos for the spindle motor and gears you currently have. A 2.5:1 belt drive and you're good from 0-7500RPM. Plus rigid tapping, 'cause it's a servo and you don't have to add an encoder to the spindle motor. Hoo-ya.

Keep us in the loop how it's progressing, and when you get to certain decisions or options on the electronics. There's more out there than just Mach3 and an ESS...

-Spumco
 

cjones6108

Iron
Registered Member
Joined
Aug 20, 2017
Messages
19
Likes
2
#7
Thanks for those insights. I'll check out these other things you mentioned. Gas springs on the z-axis is an interesting idea, and fairly inexpensive I would think.

You've pretty much sold me on the ball screws, and in the brief check I did on those, they don't seem terribly expensive either. I'm not trying to do this on a shoe-string budget, but I would like to get a quality job that's not too wildly out of proportion to the machine I decided on. My original purchase of a Bolton lathe/mill has been an education, primarily in what NOT to do.
 

cjones6108

Iron
Registered Member
Joined
Aug 20, 2017
Messages
19
Likes
2
#9
Next question: Can someone tell me the exact dimensions for the ball screws and nuts I need for the 932M? My machine won't be delivered until later in the week.

I did find the motors (steppers) on EBay and got them on order, so I'm committing to this project for better or worse.
 

spumco

Active Member
Active Member
Joined
Jan 15, 2016
Messages
112
Likes
62
#11
Yea, what he said. Don't reinvent the wheel. Copy someone else's research.

And don't spend huge money on high-dollar ground ball screws. Get some rolled ones and pre-loaded nuts - you'll never notice the difference in positional accuracy between ground ($$$) and rolled ($) with the mill you're converting. Most controllers also have ball screw mapping features which let you calibrate the controller so it adjusts the steps per inch depending on where the axis is in the travel. All you're really looking for here is zero backlash, low friction, and decent accuracy.

On the other hand, check Fleabay for used ground screws; you might score a useful set fairly cheap. Just remember that ball screws are hard as rocks - you can probably turn/machine the ends, but cutting one in half and putting a new end on it takes some serious machinery.
 

TomS

Active User
H-M Supporter-Premium Member
Joined
May 20, 2013
Messages
1,469
Likes
459
#12
I've attached my original ball screw and ball nut specification for your reference. I sent this to Linear Motion Bearing during my build for quoting purposes reference. I'm not sure if you are direct driving your ball screws or using timing belts and pulleys. My ball screws are direct drive so please take that into consideration when determining final length. I also specified single ball nuts but since have installed Linear Motion Bearing double ball nuts. Another consideration when calculating ball screw length is the thickness of your shaft support/motor mount material.

Hope this helps.
 

Attachments

cjones6108

Iron
Registered Member
Joined
Aug 20, 2017
Messages
19
Likes
2
#14
Yea, what he said. Don't reinvent the wheel. Copy someone else's research.

And don't spend huge money on high-dollar ground ball screws. Get some rolled ones and pre-loaded nuts - you'll never notice the difference in positional accuracy between ground ($$$) and rolled ($) with the mill you're converting. Most controllers also have ball screw mapping features which let you calibrate the controller so it adjusts the steps per inch depending on where the axis is in the travel. All you're really looking for here is zero backlash, low friction, and decent accuracy.

On the other hand, check Fleabay for used ground screws; you might score a useful set fairly cheap. Just remember that ball screws are hard as rocks - you can probably turn/machine the ends, but cutting one in half and putting a new end on it takes some serious machinery.

Very true, I have no intention of doing much in the way of inventing here, as this path should be pretty well beaten already. I appreciate all the accumulated knowledge.
 

cjones6108

Iron
Registered Member
Joined
Aug 20, 2017
Messages
19
Likes
2
#15
I've attached my original ball screw and ball nut specification for your reference. I sent this to Linear Motion Bearing during my build for quoting purposes reference. I'm not sure if you are direct driving your ball screws or using timing belts and pulleys. My ball screws are direct drive so please take that into consideration when determining final length. I also specified single ball nuts but since have installed Linear Motion Bearing double ball nuts. Another consideration when calculating ball screw length is the thickness of your shaft support/motor mount material.

Hope this helps.
Thanks for this. Other thoughts come to mind on this, so I think I will post them in a new thread.

(an hour later) So, after some additional reading I'm thinking to go ahead with direct drive for this project, although there can definitely be advantages to going with the timing belt/pulley option.

caj
 
Last edited:

cs900

maker of chips
Active Member
Joined
Mar 21, 2016
Messages
179
Likes
192
#16
On the other hand, check Fleabay for used ground screws; you might score a useful set fairly cheap. Just remember that ball screws are hard as rocks - you can probably turn/machine the ends, but cutting one in half and putting a new end on it takes some serious machinery.
Very good point. I got some really nice thomson (made in the USA!) ground ballscrews for dirt cheap off ebay. I couldn't be happier with them. After mapping my ballscrews I'm holding +/-.0005 pretty consistently with steppers.

And also on this topic, machining new ends it's as terrible as most people make it out to be. The screws are only case hardened which means only the outer most layer is super hard. I've found it you take a hand torch to the ends to be machined you can anneal the case very quickly and it cuts much easier after that.
 

spumco

Active Member
Active Member
Joined
Jan 15, 2016
Messages
112
Likes
62
#17
Very good point. I got some really nice thomson (made in the USA!) ground ballscrews for dirt cheap off ebay. I couldn't be happier with them. After mapping my ballscrews I'm holding +/-.0005 pretty consistently with steppers.

And also on this topic, machining new ends it's as terrible as most people make it out to be. The screws are only case hardened which means only the outer most layer is super hard. I've found it you take a hand torch to the ends to be machined you can anneal the case very quickly and it cuts much easier after that.

Very, very good to know. Opens up some otherwise unusable ebay finds. MAPP gas work ok, or oxy-fuel?
 

jbolt

Active User
H-M Supporter-Premium Member
Joined
Dec 3, 2013
Messages
1,217
Likes
995
#19
On my PM-932 conversion I went with pulleys on the x,y axis to keep the machine envelope as small as possible due to space limitations. If I had more shop space I probably would have gone with direct drive for simplicity. That being said I like the compact format and am glad I went the way I did.

I have a 4200 stepper on the z and it works great. With the large stepper the z does not move when off like a servo will.

Sent from my SM-G955U using Tapatalk
 

cjones6108

Iron
Registered Member
Joined
Aug 20, 2017
Messages
19
Likes
2
#20
Very good point. I got some really nice thomson (made in the USA!) ground ballscrews for dirt cheap off ebay. I couldn't be happier with them. After mapping my ballscrews I'm holding +/-.0005 pretty consistently with steppers.

And also on this topic, machining new ends it's as terrible as most people make it out to be. The screws are only case hardened which means only the outer most layer is super hard. I've found it you take a hand torch to the ends to be machined you can anneal the case very quickly and it cuts much easier after that.

After you anneal and machine it, do you then need to re-harden it? I recently made a small custom carving gouge (for wood) where I had to soften, shape and then re-harden. For the re-hardening, I just heated it again and quenched it in a little hydraulic oil. Does that sound about right or is there a better method?

caj
 

cs900

maker of chips
Active Member
Joined
Mar 21, 2016
Messages
179
Likes
192
#21
After you anneal and machine it, do you then need to re-harden it? I recently made a small custom carving gouge (for wood) where I had to soften, shape and then re-harden. For the re-hardening, I just heated it again and quenched it in a little hydraulic oil. Does that sound about right or is there a better method?

caj
not needed at all. The areas you machine are at the very ends of the screw where the ballnuts will never reach, so there is no need to re-harden them.
 

Wreck™Wreck

Active User
H-M Supporter-Premium Member
Joined
Sep 29, 2014
Messages
1,919
Likes
1,463
#22
No need to anneal, use a hard turning insert to get through the case then turn, thread and key as normal, CBN works well

This is a drop from a Nook 1" 4 TPI ballscrew with 4 starts (1.000" Lead), I part them to 14" in length from 6' stock screws, turn a bearing diameter on both ends and a drive pulley diameter with keyway on one end without annealing.
100 or more per year. I part them with a 6MM wide plain old carbide insert tool which is easier then bandsawing them.
The interrupted cut is a bit rough on tooling but is preferable to questionable annealing.

 
[6]
5 [7]