I'm sourcing out my own parts to build my ballscrews and mounting hardware for a seig X3 machine.

I am looking at McMaster.com for these parts because I'm not wanting to spend a fortune on conversion kits. My questions are bolded.

5/8" BALLSCREW:
item# 5966K26
Price Per Inch: $1.27
http://www.mcmaster.com/catalog/113/gfx/small/5966kp1s.gif

What is the length in inches I need to order in ballscrew for the X3 for X,Y and Z axis? (Total length including what needs to be machined off)

I will need to machine down the ends of these Ballscrews to fit the flex coupling to the motor shaft. Also there is some material machined of of the ends and I see some sort of stopper is mounted.
What is the machined shaft diameter? Also what lengths need to be machined along the shaft endls along the axis of X,Y and Z?
What are these stoppers I see set onto the ballscrews?


5/8" BALLNUTS:
item# 5966K16
Price Each: $23.85
http://www.mcmaster.com/catalog/113/gfx/small/6641kc1s.gif


ZERO-BACKLASH FLEXIBLE SPIDER SHAFT COUPLINGS:
Plain Aluminum Item# 9845T2
Set Screw Glass Fiber Item# 9939T1
Price Each: Somewhere near $15
http://www.mcmaster.com/catalog/113/gfx/small/9845tp1s.gif
http://www.mcmaster.com/catalog/113/gfx/small/9845tc1s.gif

To connect the couplings to the NEMA 23 motor shaft, you would need a 1/4" couple on one side. For the other side I'm unclear about because I have no idea what the shaft diameter I will need to machine the Ballscrews to. 5/8" ballscrew diameter is O.D. right?
There is also an option to choose from a coupling that sets on the shafts with a set screw, or to choose the aluminum coupling with no set screw. Which one would you recommend? I think the motor shaft is ground flat on one side, but that would mean the ballscrew would also need to be machined flat on one side as well, right?
It also looks like it is appropriate to choose the proper spider to connect the couplers. I suppose choosing the proper in/lb is of greatest concern. There are also angular tolerances and hardness properties. Less hardness leades to greater cushioning I believe. Angular tolerances leads to better backlash resistance I believe.


GAS SPRING:
15.23" 90lb force
Item# 4138T55
Price Each: $19
http://www.mcmaster.com/catalog/113/gfx/small/4138tp1s.gif

What Length Spring should I order? I see the X3 has 14-7/8" of head stock travel, so there would need to be at least 15" Total Gas Spring Length (Dimension A). Also, The Compression length needs to accomidate the mill at it's lowest stroke (Dimension B). This leads me to pick Item# 4138T55. Can anybody tell me if this is correct?
I picked the 90lb Gas Spring force because of what I've read on another thread on the forums. Is this a good value to go with?



ANGULAR CONTACT BEARINGS:
(2 Per Axis Back to Back)
ITEM# 7201B 12x32x10
Cost Per Axis: $15
http://www.mygoodstuff.com/kit1084-1.jpg

As for the ballscrew mounting blocks, and motor mounts, I'm going to engineer this fine man's solidworks files to work with stepper motors.
Bob Berg's Site: http://www.rlberg.com/millprt/



I really appreciate all the help I've been getting from the members of this site. I am really getting very close to summarizing all of the parts I need and then go through with this X3 Build!

When I mentioned the ballscrew lengths, this is what I was talking about.

Perhaps this becomes more obvious when I get the machine in front of me and I remove the ACME threads.

http://i11.photobucket.com/albums/a161/ninhil/ballscrews-1.jpg


edit: perhaps the machined length of the ballscrew at the motor connection is 10mm. This will match the ID of the Bearing I listed. BUT....

But what about the coupler? I'm still confused as to what coupler size I should use. Its probably best to use the largest diameter machined down from the ballscrew I imagine.

:)

Mcmaster is a good source , SDP-SI is a real good source as well. I like OLDHAM type couplers which you can get from either source. 6ft of ballscrew should be enough. Your questions about the machining I can not answer though. If you are going to try and duplicate the conversion you linked to I would think the owner of that mill might be able to help with those specific questions. There is no set length or shaft diameters that can be used universally to determine what needs to be machined off the ballscrews. That all depends on the conversion design. I have no idea what those stopper things are. Might be a bearing or collar. I have no such animal on my mill. Steve

re things on the end of ballscrews
top one is a split ring about 3/8 inch thick that has a set screw that will make it clamp down on the 12 mm shaft so shaft is snug againt a bearing. I do not know what they are called.
The lower one looks like a spanner nut threaded on shaft(does the same thing as above but can be adjusted very precisely to squeeze bearing just the right amount. If you have room to get tool in to adjust, this is best. My z shaft has very little room on bottom to service so it has the clamping type and my x and y has threaded ends because you can get to ends easily to tighten nut up against bearing.

Both of those are shaft collars with set screws.

It's not how the ballscrews should be held in place.

There is a huge discussion on this regarding the cncfusion conversion kit and their methods.

Ideally, you'd have the screw machined to exact length needed. Then each end would be turned down to slitfit into a bearing. There's much discussion on the type of bearing.

Most of the spindles are designed with angular contact bearings. I'd guess you could continue that practice, but you'd really need some mechanical engineer to tell you why you really need one of the other.

Now that you have the ballscrew to length. And supported on each end with a bear. You need to decide how your going to contain it.

In the example picture above. It's contained by smooshing that collar against the bearing then cranking down the set screw.

The most accepted practice that I have seen involves one end of the screw being turned down over a longer length. Now you have the bearings to support the screw. That bearing fits into a pocketed plated. On the opposite side of the plate is another pocket where an additional bear will be fitted.

Now you have the screw supported with dual bearings sandwiched around the mounting plate. Now you have to keep it from moving in and out. This is done by threading the part beyond the second bearing. You can then apply a certain amount of pressure to the thread/nut and secure the screw.

Past the threaded part is where the screw is turned down to support connection to a motor. Through some type of coupler. Generally these are machined flat one one side to allow for a sets screw.

I hope that my description is good enough.

Both of those are shaft collars with set screws.

It's not how the ballscrews should be held in place.

There is a huge discussion on this regarding the cncfusion conversion kit and their methods.

Ideally, you'd have the screw machined to exact length needed. Then each end would be turned down to slitfit into a bearing. There's much discussion on the type of bearing.

Most of the spindles are designed with angular contact bearings. I'd guess you could continue that practice, but you'd really need some mechanical engineer to tell you why you really need one of the other.

Now that you have the ballscrew to length. And supported on each end with a bear. You need to decide how your going to contain it.

In the example picture above. It's contained by smooshing that collar against the bearing then cranking down the set screw.

The most accepted practice that I have seen involves one end of the screw being turned down over a longer length. Now you have the bearings to support the screw. That bearing fits into a pocketed plated. On the opposite side of the plate is another pocket where an additional bear will be fitted.

Now you have the screw supported with dual bearings sandwiched around the mounting plate. Now you have to keep it from moving in and out. This is done by threading the part beyond the second bearing. You can then apply a certain amount of pressure to the thread/nut and secure the screw.

Past the threaded part is where the screw is turned down to support connection to a motor. Through some type of coupler. Generally these are machined flat one one side to allow for a sets screw.

I hope that my description is good enough.

I understand. This relly helps to explain whats going on here. Thanks!



So from what I gather so far:

BALLSCREWS:
I'm going to need somewhere around 6' of ballscrew. It would be nice to know how much I need for each axis. I can just order the machine, and copy the length of the ACME Screws for this... Unless someone already knows. I can order the screws from McMaster cut to length. This will also save on shipping.

SHAFT COUPLERS:
1/4" is needed on the side that couples to the motor. What Diameter should I use to couple to the Screw? 1/2"?

GAS SPRING:
Is that one I spec'd out suitable?


Again, I appreciate all the help!

-Brandon

As it was posted above.

There is no "set" screw length. That depends on your setup. ((Will you be using stock location of the existing mounting plates. Or will you hollow them out to steal a little extra movement or mounting depth))

I.e. if you use a double ballnut with double angular contact bearings. Your length could be longer than someone that uses stock bearings with a single nut.

Additonally, depending on how you attach your motor, you could need a longer or shorter shaft sticking out beyond the bearing location. I.e. You have a 2-1 pulley system vs direct connect.

The couplers depend on your design. If you use bearings that have an inside diameter of 12mm, with a 10mm thread section for preload nut. Then I'd assume that your shaft portion will be somewhere between 0-10mm's.

You need to decide what size ballscrews, bearings and motor shaft your using. Then research what premade couplers exist. I.e. if they only offer 1/4 to 1/4 or 1/4 to 1/2. Your decision is already made.

As for the gas spring. I don't know. I've read of people using 75lbs. You need to know how much the head weighs, then decide how much of that you want to left. If the head is 75lbs but you use a 95lb gas spring.. You increase the work to lower the head, but make it easier to lift.

Ideally, I'd think you'd want it even, or slightly on the weaker side to allow gravity to help with chatter/kickback on the head.

The last part is only my assumption.

((Edit, cause I can't spell/type))

As it was posted above.

There is no "set" screw length. That depends on your setup. ((Will you be using stock location of the existing mounting plates. Or will you hollow them out to steal a little extra movement or mounting depth))

I.e. if you use a double ballnut with double angular contact bearings. Your length could be longer than someone that uses stock bearings with a single nut.

Additonally, depending on how you attack your motor, you could need a longer or shorter shaft sticking out beyond the bearing location. I.e. You have a 2-1 pulley system.

Couplers depend on your design. If you use bearings that have an inside diameter of 12mm, with a 10mm thread section for preload nut. Then I'd assume that your shaft portion will be somewhere between 0-10mm's.

You need to decide what size ballscrews/bears and motor shaft your using. Then research what premade couplers exist. I.e. if they only offer 1/4 to 1/4 or 1/4 to 1/2. Your decision is already made.


ahh i see now. so 6' it is!
@1.27 per inch, this aounts to $91.44 of material.
shipping may be higher on 6', but at least i wont have any lengths cut too short.


As for the gas spring. I don't know. I've read of people using 75lbs. You need to know how much the head weighs, then decide how much of that you want to left. If the head is 75lbs but you use a 95lb gas spring.. You increase the work to lower the head, but make it easier to lift.

Ideally, I'd think you'd want it even, or slightly on the weaker side to allow gravity to help with chatter/kickback on the head.

The last part is only my assumption.

((Edit, cause I can't spell/type))

I'm also going to assume the weight should be even or slightly more than head weight. This is to keep the ball nuts held against the screws, am I right?

CNC Fusion says it is used for making raising and lowering the spindle head easier. SO... How much of a difference does this really make?

Unless you preload the ball screw support bearings, you're going to have direction change slop issues to contend with that backlash comp will NEVER get rid of. You can preload them by either shimming them or sending them to KAF Manufacturing in Stamford CT and have them preload them DB for you.

I"d go with about 150 lbs of TOTAL preload for the each set of 2 bearings (distributed 75lbs per bearing). This is NOT that much as a true ball screw bearing can be set with as much as 500 lbs of preload. The 150 is pretty much a "standard", accepted value for a true, milling machine grade mills.

To determine how much to shim is not doable by calculation - you have to dead load the bearings axially and measure the individual raceway offsets. KAF can do this for you, individually grind them, and then they etch the chevron markings on OD to denote proper positioning.

I don't recall what the cost is - its been a while since I had any done - but it is reasonable all things considered. Since you're already sparing no expense for the bearings (true ball screw bearings go up to $800/set), you should be able to afford precision/professional preloading services.

All things considered (as in time wasted, hurt parts, screwing around), having a proper preload ground into the bearings is simply good money well spent. YOU WILL NOT REGRET IT.

Unless you preload the ball screw support bearings, you're going to have direction change slop issues to contend with that backlash comp will NEVER get rid of. You can preload them by either shimming them or sending them to KAF Manufacturing in Stamford CT and have them preload them DB for you.

I"d go with about 150 lbs of TOTAL preload for the each set of 2 bearings (distributed 75lbs per bearing). This is NOT that much as a true ball screw bearing can be set with as much as 500 lbs of preload. The 150 is pretty much a "standard", accepted value for a true, milling machine grade mills.

To determine how much to shim is not doable by calculation - you have to dead load the bearings axially and measure the individual raceway offsets. KAF can do this for you, individually grind them, and then they etch the chevron markings on OD to denote proper positioning.

I don't recall what the cost is - its been a while since I had any done - but it is reasonable all things considered. Since you're already sparing no expense for the bearings (true ball screw bearings go up to $800/set), you should be able to afford precision/professional preloading services.

All things considered (as in time wasted, hurt parts, screwing around), having a proper preload ground into the bearings is simply good money well spent. YOU WILL NOT REGRET IT.

I've read about threading the ballscrew shaft, placing a nut and washers on the screw, and inserting a spring in between the washers. This presses against the bearings.

I probably will be doing this myself, and not sending these parts to a shop for their analysis. But thanks for the info.

Hello,
I'm a mechanical engineering major and for my senior project I have been assigned the task of designing a gantry style table top cnc mill. I've never designed a mill, so I am trying to get some more information about the technical details in designing a mill.

Both of those are shaft collars with set screws.

It's not how the ballscrews should be held in place.

There is a huge discussion on this regarding the cncfusion conversion kit and their methods.

Ideally, you'd have the screw machined to exact length needed. Then each end would be turned down to slitfit into a bearing. There's much discussion on the type of bearing.

Now that you have the ballscrew to length. And supported on each end with a bear. You need to decide how your going to contain it.


The most accepted practice that I have seen involves one end of the screw being turned down over a longer length. Now you have the bearings to support the screw. That bearing fits into a pocketed plated. On the opposite side of the plate is another pocket where an additional bear will be fitted.

Now you have the screw supported with dual bearings sandwiched around the mounting plate. Now you have to keep it from moving in and out. This is done by threading the part beyond the second bearing. You can then apply a certain amount of pressure to the thread/nut and secure the screw.

Past the threaded part is where the screw is turned down to support connection to a motor. Through some type of coupler. Generally these are machined flat one one side to allow for a sets screw.



I was confused about some of the thing brought up in this post. What is a bear?
I can't visualize the "dual bearings sandwiched around the mounting plate".
or the "threaded part beyond the second bearing"
Any link to where I can see a good example of this?

I was particularly interested in this post:

Unless you preload the ball screw support bearings, you're going to have direction change slop issues to contend with that backlash comp will NEVER get rid of. You can preload them by either shimming them or sending them to KAF Manufacturing in Stamford CT and have them preload them DB for you.

I"d go with about 150 lbs of TOTAL preload for the each set of 2 bearings (distributed 75lbs per bearing). This is NOT that much as a true ball screw bearing can be set with as much as 500 lbs of preload. The 150 is pretty much a "standard", accepted value for a true, milling machine grade mills.

To determine how much to shim is not doable by calculation - you have to dead load the bearings axially and measure the individual raceway offsets. KAF can do this for you, individually grind them, and then they etch the chevron markings on OD to denote proper positioning.

I don't recall what the cost is - its been a while since I had any done - but it is reasonable all things considered. Since you're already sparing no expense for the bearings (true ball screw bearings go up to $800/set), you should be able to afford precision/professional preloading services.

All things considered (as in time wasted, hurt parts, screwing around), having a proper preload ground into the bearings is simply good money well spent. YOU WILL NOT REGRET IT.

What exactly is meant by preloading the ball screw support?
Preloading them how? press fitting the bearings in place?

"you have to dead load the bearings axially and measure the individual raceway offsets."
Huh?

It's kind of a big project and all the little details are starting to get a bit overwhelming.
Some good issues were brought up in this tread and I'm just trying to get a little more info on these issues.

There are indeed a lot of details.

Check google for documentation about preloading bearings.
I don't know what the cost is to get regular bearings preloaded. BUt it may be more economical to simply buy a duplex already preloaded from New Hampshire Ball Bearing. Plus, your original bearings will probably be of higher quality than generics. All of Danaher Motion's duplex bearings are NHBB.
That said, I successfully preloaded 3 pairs of 10$ VXB bearings, and the results are fantastic.

I came up with a nice link
http://www.eminebea.com/content/html/en/engineering/bearings/preload.shtml


Preload and Stiffness
There are two basic methods of Preloading: Solid Preload and Spring Preload. Solid Preload can be obtained by mechanically locking all of the rings in postion while under an axial load. The advantages of this type of design are that the components remain simple and the stiffness is high. The disadvantage is high variation in Preload under temperature variation, and that the Preload can reduce with wear. Spring Preload (or Constant Pressure Preload) can be applied using a coil spring or a spring wave washer, etc. An advantage of Spring Preload is that it maintains consistent Preload with temperature variation.The disadvantages are that the designs are more complex and normally have lower stiffnesses.

Solid Preload
Solid Preload
http://www.eminebea.com/content/html/en/engineering/bearings/images/pic8-2.jpg
Spring Preload
Spring Preload
http://www.eminebea.com/content/html/en/engineering/bearings/images/pic8-3.jpg

Hello,
I'm a mechanical engineering major and for my senior project I have been assigned the task of designing a gantry style table top cnc mill. I've never designed a mill, so I am trying to get some more information about the technical details in designing a mill.



I was confused about some of the thing brought up in this post. What is a bear?
I can't visualize the "dual bearings sandwiched around the mounting plate".
or the "threaded part beyond the second bearing"
Any link to where I can see a good example of this?

I was particularly interested in this post:



What exactly is meant by preloading the ball screw support?
Preloading them how? press fitting the bearings in place?

"you have to dead load the bearings axially and measure the individual raceway offsets."
Huh?

It's kind of a big project and all the little details are starting to get a bit overwhelming.
Some good issues were brought up in this tread and I'm just trying to get a little more info on these issues.



Bear = Bearing. I just type faster than I think sometimes.

I've attached a picture to help explain.

orange = mounting plate.
blue = bearings
green = ballscrew turned down to fit through and threaded
purple = nut to hold ballscrew.

The orange plate would then be mounted to whatever you attaching it to. I.E. X-axis table etc.

Graphic is primative, but should help.

Bear = Bearing. I just type faster than I think sometimes.

I've attached a picture to help explain.

orange = mounting plate.
blue = bearings
green = ballscrew turned down to fit through and threaded
purple = nut to hold ballscrew.

The orange plate would then be mounted to whatever you attaching it to. I.E. X-axis table etc.

Graphic is primative, but should help.


thats a great graphic..

I'm going to assume the ballscrew is only threaded to accept a mate with the nut. for that section only.

I am going to try a system wherein there is a strong spring in between the nut and the bearings.. similar to the picture I posted from that website I found earlier.
From what I've read, its better to install a system such as this one to keep the tension upon the bearings as constant as possible. A straight nut to bearing assembly may have a reduction in tensions over time.

Here is a shot that details the way I do it with my X2 yoke set-up. It shows detail of the ball screw ends.

Ken

Yea it's only threaded enough to allow the nut to get a good bite and secure the screw within the bearing sandwich.

Also to the left of the nut, the screw would be turned down to match your motor coupling.

Usually one side of the turned down section is milled flat to allow for a set screw.

Here is a shot that details the way I do it with my X2 yoke set-up. It shows detail of the ball screw ends.

Ken


that is a really good picture. thanks!

"Spring" based preloading is fine - as long as you NEVER overcome the spring preload.

Being an engineer, you should understand the F=MA equation which, depending on the A and the M, overcome the F generated by your proverbial spring.

When you position preload via mechanical means (offset, nut torque, shims, whatever), the spring rate of the bearings (which virtually infinite) becomes the "variable". Since it can't/won't vary, your preload stays relatively constant regardless of your F, M or A.

At one time NSK had some decent fotos in their catalogs that explained what/how raceway offset did/does to preload in bearings. I haven't be abot to load their website of late. The raceway offset method is what true ball screws used to achieve very rigid, non-deflecting, low friction ball screw thrust/radial support bearings.

If you look at true machine tool ball screws, they usually affix the support bearings to one end and tighten them with a nut that clamps the inner rings together to set any preload built into the bearings.

This is machine tool grade stuff. Hobby grade stuff can perhaps be spring, shim or whatever preloaded IF you are prepared to live with any compromise that you choose to accept/incorporate. Your project, your call, you're responsible for the peformance, good bad or indifferent.

"Spring" based preloading is fine - as long as you NEVER overcome the spring preload.

Being an engineer, you should understand the F=MA equation which, depending on the A and the M, overcome the F generated by your proverbial spring.

When you position preload via mechanical means (offset, nut torque, shims, whatever), the spring rate of the bearings (which virtually infinite) becomes the "variable". Since it can't/won't vary, your preload stays relatively constant regardless of your F, M or A.

At one time NSK had some decent fotos in their catalogs that explained what/how raceway offset did/does to preload in bearings. I haven't be abot to load their website of late. The raceway offset method is what true ball screws used to achieve very rigid, non-deflecting, low friction ball screw thrust/radial support bearings.

If you look at true machine tool ball screws, they usually affix the support bearings to one end and tighten them with a nut that clamps the inner rings together to set any preload built into the bearings.

This is machine tool grade stuff. Hobby grade stuff can perhaps be spring, shim or whatever preloaded IF you are prepared to live with any compromise that you choose to accept/incorporate. Your project, your call, you're responsible for the peformance, good bad or indifferent.


so you're saying that there is something lacking in a spring-preload, as compared to a straight nut-to-bearing preload?

Spring versus position preload: Yes, spring preloads are NOT going to give you the same stiffness potential of a properly offset ground set of A/C bearings. They simply CAN'T.

It takes a MASSIVE spring to come up with the equivalient instantaneous spring rate of a preloaded A/C. Consequently, you'll never be able to package a spring that can/will do an adequate job.

It is one thing if you're simply trying to take the slop out of a bearing (what a wave washer spring does in an electric motor). Such springs only need to apply about 1% of the rated radial capacity axially to the the bearing.

When you preload ball screws, you need WAY more stiffness than that. Example: a 20TAC47 true 20MM ball screw bearing (typical of what would fit in a Bridgeport mill) has nearly 500 lbs of axial preload. Result: very stiff with amazingly low rotating friction

The factory 20mm deep groove ball bearings (a psuedo ball screw equivalent via modified 6204's) at best, has 150lbs of preload - relatively low friction but not as stiff as a 20TAC47 due to the internal geometry differences 'tween 20TAC47 and the 6204's.

A wave washer would be hard pressed (no pun intende0 to generate 20lbs of load. Moreover and more importatnt, this relatively lame preload could EASILY be overcome when accelerating hard or if you got a tool to start chattering real hard while cutting.

Yes, i'm contending and know from experience that a spring preload bearing (although convenient to DIY) is NOT as robustly rigid as a properly applied position preloaded bearing. BTDT and now know better....

Spring versus position preload: Yes, spring preloads are NOT going to give you the same stiffness potential of a properly offset ground set of A/C bearings. They simply CAN'T.

It takes a MASSIVE spring to come up with the equivalient instantaneous spring rate of a preloaded A/C. Consequently, you'll never be able to package a spring that can/will do an adequate job.

It is one thing if you're simply trying to take the slop out of a bearing (what a wave washer spring does in an electric motor). Such springs only need to apply about 1% of the rated radial capacity axially to the the bearing.

When you preload ball screws, you need WAY more stiffness than that. Example: a 20TAC47 true 20MM ball screw bearing (typical of what would fit in a Bridgeport mill) has nearly 500 lbs of axial preload. Result: very stiff with amazingly low rotating friction

The factory 20mm deep groove ball bearings (a psuedo ball screw equivalent via modified 6204's) at best, has 150lbs of preload - relatively low friction but not as stiff as a 20TAC47 due to the internal geometry differences 'tween 20TAC47 and the 6204's.

A wave washer would be hard pressed (no pun intende0 to generate 20lbs of load. Moreover and more importatnt, this relatively lame preload could EASILY be overcome when accelerating hard or if you got a tool to start chattering real hard while cutting.

Yes, i'm contending and know from experience that a spring preload bearing (although convenient to DIY) is NOT as robustly rigid as a properly applied position preloaded bearing. BTDT and now know better....


thanks for the info. You've just convinced me to go without a spring pre-load. Not only for reasons being that its better than spring, but there are less parts involved and looks to be even simpler.

I suppose I'll go with a two nut design.. One to press against the bearings, and one other to lock the assembly into place.


Thank You

Deviant's picture above sets the stage for half of the backlash discussion. It is in fact a great way to look at how one secures the lead screw at one of the ends.

By the way, in the picture that showed the shaft clamp on what is the long ball screw and the nut on the shorter ball screw....that picuture is from the cncfusion kit. I have one of the kits for my X3 (in the process of installing now!). I think the assertions as to what exactly each element is, have been in dispute. Since I have what is in the picture, I thought it would be good to share as well as offer the idea that mounting, backlash adjustments and loading should be divided into the two elements that contribute most to backlash.

1. Backlash evident in the mounting arrangement for the "screw"

2. Backlash evident in the mating of the screw to the nut (acme, ballnut, whatever) where the nut is typically attached to the movable axis element.

Both should be attended to. This by the way ignores the linearity of the nut to screw mating along the entire length of the screw and host of other factors including backlash that might be in the motor to screw mounting (discussed elsewhere). I mention this as a backdrop to the notion that if backlash is "mostly" predictable/constant, it can be compensated for in the machine controller. Most will argue that is never good enough. I won't argue, but I have a relatively inexpensive mill of offshore manufacture so I know ahead of time it has some limitations......

Back to the elements in the picture early in the thread.

The longer screw (for Z) has a shaft collar with a single screw clamping action (not a setscrew - it simply makes the ID of the collar slightly smaller) - it is used to eliminate as much backlash in the lead screw relative to its lower end mount - the Z in the case of the X3 cncfusion kit. It has no influence on the backlash in the ballnut to ballscrew action. The nut on the threaded portion of the end of the other ball screw (for the X on the x3) is a spanner nut with a nylon compression insert used to "lock" it into position. That is why two nuts are not used (never mind if it is good or bad).

The very small washers shown are the transition between the machined down portion of the screw shaft and the actual remaining ball screw portion - so that bearings are not rubbing on the machined transition.

On the X3, the stock arrangement for all of the lead screws has a pocket machined on each side of the end mount closest to the standard hand wheels for X & Y and near the lower end of the Z. In each case, the mount has a smaller diameter through hole between the pockets to accomodate the lead screw. Deviant's picture above shows the arrangement exactly. The standard bearings used are normal (cheap) 3 part roller thrust bearings in each pocket where one is expected to lightly preload using the spanner nut or shaft collar. This forms an inexpensive mechanism for very low backlash for the screw only. As properly pointed out, opposing bearing pair preloads represents a "better" method to minimize backlash than springs. The use of the standard roller thrust bearings means one cannot pre-load the bearings to the same extent one could with AC bearings. By the way, my search for AC beardings of the same OD/ID/Width has thus far proven elusive. One would have to machine the pockets to accomodate what seem to be standard AC bearing sizes (at least the high precision ones).

Now the second element of backlash mentioned above.

To accomplish ballscrew backlash minimization, one can go with a variety of methods or combinations. One way is pre-loaded ball nuts (normally accomplished with replacement balls slightly larger than standard to minimize axial movement i.e. tighter nut). Another way is via double ball-nuts - tightened against each other to pre-load. Then there are variations with ground vs. rolled ball screws doing much of the same as above along with strange combinations of AC bearings mated with the ballnuts.

For those staying with the acme nuts and screws - high precision nuts and screws and various double nut and/or spring combinations are about all that is available. Or you can stay with the standard split brass acme mounting and use screws to close the gap - a variation on the double nut approach.

Any change to the standard X3 costs money. You want to go high on the curve of better specs, expect to spend a lot more money.

I went with the cncfusion single ballnut per axis ballscrew upgrade kit. Adding additional ballnuts generally causes a penalty in travel but various extended/modified mounting structures can compensate. Reloading the ballnuts with larger balls is also an option to get closer and can be done by the individual. Super high tolerance pre-loaded ballnuts and ground ballscrews are too expensive for me. So, from where I started, there is room for me to self improve without spending gobs more money.

Regards,
George

just for clarification...

When using a pair of AC bearings separated by a block...
One end of the screw is threaded so that a nut can squeeze the bearings together and preload the bearings and screw. So one side is pressed against the nut, but what about the other side?
Does the lip in the screw press against the other side? Is there a washer? how much of a lip should there be? How much distance should there be between the two bearing (ie. block thickness)? Does it matter?

Also what is the advantage of using an AC bearing pair instead of a single thrust bearing? Is it just that the pair can take more preload?

A lot of your questions are simple packaging issues. Shoulder heights for housings are found in bearing design handbooks. Some of your other questions can be readily self answered by drawing up the bearing and shafts and shoulders to scale and applying a bit if deductive reasoning on what's moving and why and how.

The packaging of A/C bearings is generally done without shims or spacers because you don't have to machine the housings - you take advantage of the built in accuracy that the bearing maker provides via the use of precision ground bearings.

HOWEVER, should the packaging require spacing the bearings farther apart to create better overturning moment resistance, then you make precision ground spacers and/or use precision machine housings.

In a ball screw, for the inner rings, you have ball screw shoulder'd shaft, bearing 1, bearing 2, clamp nut that clamps innner rings together on shaft. The outer rings press directly against each other at the mating surface. between 1 and 2 when the inner rings are clamped together.

The outer rings sit in a c'bored cavity mouted to the machine bed that has a clamping plate to retain the outer rings from moving axially.

The screw nut is mounted solidly to the yoke and this yoke moves the table when you turn the ball screw shaft.

Why not a single bearing? What is it going to resist axial motion against? Sort of like clapping with only 1 hand.

You can try to use a single thrust washer to absorb load BUT generating a consistant preload WITHOUT creating gobs of friction is hard to do with a single thrust washer. THis all has to do with the internal geometeries involved and, without a long illustrated explanation, it is simply too hard to try to explain via text message.

Suffice it to say that a heavily preloaded, high contact angle, rolling element ball bearing is going to give you the best compromise of high stiffness/low friction.

Why a ball bearing? Simple, ball bearings have point coniact at the roller contact point, whereas cylindrical needle or roller bearings have LINE contact. Since there is always going to be a radius difference between the ID and OD of an axial thrust bearing roller, there is always going to be sliding friction to contend with. Not so with a ball bearing.

Do some research into "ball screw support bearings" if you want to more about the internal geometry and why they act the way they do.

Thank you for taking the time to write such a detailed response
I can certainly see why one would use a duplex bearing setup at each end of the screw. But this arrangment is alot more expensive than having a single bearing at each end. Also, for my particular application, I don't think I need the tightest tolerances. (Mostly engraving, plastic machining, maybe some light duty metal work (AL & Cu)).

So if a duplex setup isn't possible, what are some alterntives?
Because I'm dealing with relatively light duty machining, are AC bearings the best choice? Compared to maybe deep-groove bearings, I think they cost about the same, so why use one over the other?

Also, don't some bearings come preloaded?

Couldn't you machine the screw lenght so that the ends of the table push on the outer bearing races, and the screw itself by getting slightly compressed, pushes on the inner races. I realize this could cause buckling issues, but with a 5/8" screw @ 130 oz-in steppers, I think I'm ok.

A/C's have better stiffness than simple ball bearings - it has to do with the fact that a deep groove ball bearing, loaded axially, is loading the raceways in an area where it is not technically designed to carry load.

Deep groove ball bearings are designed to carry RADIAL loads, not axial loads. By going to A/C's, the raceway is canted 15, 30, or whatever contact angle is designed into the raceway. For true ball screws, you're looking at a 60 degree contact angle. By some simple trigonometry, you can figure hout how much axial load is converted to a radial and axial vector.

Why A/C's? simple they perform better than a deep groove loaded, essentially improperly. SOme regular deep groove ball bearings are/were used to make low buck ball screw support bearings - Bridgeport did this on their classic mills. The problem with this method is that you'll run into axial deflection and backlash issues. By adding preload, you'll improve it BUT you'll eventually get to the point where the ball screw is very hard to turn and you'll still have backlash.

Ultimately, there is a cost/benefit/performance factor to contend with with ANY machine component. If you can live with less than perfect backlash, nearly anything will suffice. When your trying to machine stuff without dwell issues at direction reversals or ROUND circular mille pockets or whatevers, you'll eventually run out of patience with "good enough" screw support bearings.

Do a site search for "Extrak and backlash" or "Extrak and ball screw"and you'll see a my postings on what we did to address the very same issues that you or anyone who starts messing and tuning of mill ball screws will encounter. You get what you're willing to pay and work for.

We tried all sorts of tricks to get deep groove ball bearings to work as ball screw support bearings. Ultimately, for our needs (cam lobe master milling) only the best proved to be adequate. The kluged up stuff was better but not good enough. Interestingly, every person who's used our mill cringed at the price we paid to get the ball screw bearings we ended up with.

Yet, several have called and asked "how to do it". Once you feel the smoothness and experience backlash free table travel, it is rather addictive. The thing is DEADLY accurate when it cuts and you simply don't screw up or screw around when milling. And you DON'T make mistakes by OOPS'ing and not comping for backlash slop.

Your mill and your choice. You don't have to justify your decision(s) to me. I've already made mine and spent the time and $$'s to do it in a primo fashion. We don't regret it either even though it hurt like the dickens when we spent the time and $$$'s to do it originally. It has paid back time and tima again countless times since.

I'm using a slightly different bearing setup. It's probably not as good as a set of good angular contacts though.
I'm using the twin ball thrusts that came with the mill along with a pair of deep grooves on each table axis in the bearing blocks. So it is 3 piece thrust, 2 deep grooves and another 3 piece thrust in the bearing housings.
On the X axis I also have a deep groove on the end of the screw. The Y screw has an unsupported end.

To keep everything compact my X axis oldham coupler is threaded and acts as the preload nut and a center setscrew locks the preload. On the Y axis I do the same but with the driven pulley.

It seems to work real well and my mill will hold .0005" backlash on the x and Y axis with dual preloaded ball nuts. I would say it is a very good hobby mill setup.
When you are jogging in increments of only .001" it is real rewarding to see movement on a direction change!
I spent a lot of time learning how to tweak everything to obtain the low backlash. It's real easy to get down to .002" or so backlash. Getting it to .0005" and have it HOLD that tolerance for many hours of run time takes some work.
But it is well worth it as NC Cams points out, even at the hobby level.
I see a lot of people thinking they can use software to account for backlash.
Well if using Mach3 and steppers the backlash comp only works with exact stop mode.It won't work properly in constant velocity mode. Plus the cutter forces do not like backlash and will play havoc. Climb milling for example. If a cnc machine can't climb mill well it is a real huge handicap.
I never use backlash comp on my machine and it will cut some impressive circle interpolations.

Steve

In our situation, we need to hold cam profile shapes to within 0.0001" or better. With work, tuning and both ball screw and ball screw support bearing work, we're well within that range of accuracy.

It's to the point that the masters we cut for race cam grinding via the use of the tuned up system grind as good a cam as those ground on a ultra-trick full Landis 3L CNC ground master. Spintron and dyno testing of the engines that use the cam prooved the accuracy/capabilities of the process.

ALthough the work we did on the ball screws was for this purpose, everything else we machine is merely icing on the cake as a result of the extreme accuracy that can be achieved.

Perhaps if Bridgeport had used a similar tact instead of trying to make cheap bearings work to save money, they'd still be in business. THey might also have been able to retain the level of business that Haas and others took from them instead of being merely relegated to now being merely a "brand name" as opposed to an industry icon that they used to be.

NHBB will sell you true duplex bearings, all grinding done, for a pretty good price. 10mm ID duplex was $108/pair last time I checked. But this is already ground, you just clamp the inner and outer races and it's properly preloaded. Not to mention, NHBB makes all their stuff in house. All high end motion control products like those from Danaher Motion's upper end use NHBB. Figures since NHBB is very close to a Danaher office here.
Compared to a 2 dollar set of radial bearings or thrust bearings, 108 is a lot. But when you look at what people spend on getting their mills going sometimes, that is not even the cost of a Gecko drive.

That said, I was cheap and went for 10mm angular contact bearings from VXB.com, for 10 bucks a pair :D. But I used NC_Cams advice and measured the axial offset of each bearing, and devised a spacer scheme. I could not feel any backlash once the bearings were installed and preloaded. Maybe today I will check it with an indicator.

NC_Cams, would you recommend the same scheme if the motor was offset and pulley-attached, rather than directly inline with the screw?

Thanks

Dodson

I am glad to see that this thread is helping people source their own parts.


I happened to actually order the CNCFusion kit. I figured I would save a lot of time over making this for the first go.

I have had ballscrews in operation for about 3 years now. I purchased them from homeshopcnc, .com, and are machined like this:

Motor end is annealed and turned down to 0.500", to fit in a bearing block, and to facilitate ballnut installation. After the smooth shaft, it is threaded 7/16-18, for the nut that will squish the angular contact bearings together, beyond that thread, it is turned down to 0.375" with a flat on one side for a setscrew within the coupler.

I hope this isn't too late.

I have had ballscrews in operation for about 3 years now. I purchased them from homeshopcnc, .com, and are machined like this:

Motor end is annealed and turned down to 0.500", to fit in a bearing block, and to facilitate ballnut installation. After the smooth shaft, it is threaded 7/16-18, for the nut that will squish the angular contact bearings together, beyond that thread, it is turned down to 0.375" with a flat on one side for a setscrew within the coupler.

I hope this isn't too late.


useful information! This is exactly what i was looking for