You forgot to post the formula in the "back driving torque" section.
Yes I did!
Not at all!
It's all GREEK to me!
It's exactly what I was looking for!
BALL SCREWS 101
Normally, ball screws are used to convert rotary motion into linear motion. Backdriving is the result of the load pushing axially on the screw or nut to create rotary motion. All ball screws, due to their high efficiency, will backdrive. The resulting torque is known as “backdriving torque” and is the torque required to hold a load in position.
CAUTION - When using ball screws, applications should be analyzed to determine the necessity of a brake, especially when the possibility of injury may occur.
Due to the efficiency of a ball screw, a load applied to the ball nut will generate backdriving torque on the ball screw. The torque required to hold the load in position can be calculated by the following formula.
Back Driving Torque Formula:
Td = Drive Torque (pound-inches)
P = Load (lbs.)
L = Screw Lead (inches/turn)
e = Ball Bearing Screw Efficiency (90%)
Backlash (lash) is the relative axial movement between a screw and nut without rotation of the screw or nut. Lash in ball screws will remain constant during normal use.
Ball Circle Diameter:
The diameter of the circle generated by the center of the bearing balls when in contact with the screw and nut.
Bearing Ball Circuit:
The closed path that the bearing balls follow through the ball nut. Ball nuts may have one or more circuits.
A load that tends to “squeeze” the screw.
Driving torque is the amount of torque required by the ball screw to move a load. To simplify this calculation a “torque to raise one pound or one kN” value is provided in the technical data for each ball screw size.
To determine the required torque to move a load, multiply the load to be moved by the “torque to raise one pound or kN”. For more information on drive torque, see the application selection example.
The thrust load in pounds which, when applied to the ball nut and rotating screw assembly will result in a minimum life of 1,000,000 inches of travel. Metric screw designs are per ISO 3408 and show the load ratings in kilonewtons for 1 million revolutions of the screw.
The low coefficient of friction of the rolling elements of Ball Screws and Nuts results in an operating efficiency greater than 90%.
The outside diameter of the screw. This diameter is less than the ball circle diameter.
The axial distance the nut advances in one revolution of the screw. The lead is equal to the pitch times the number of starts. (Lead = Pitch X Starts)
Lead accuracy is the difference between the actual distance traveled versus the theoretical distance traveled based on lead. For example: A screw with a .5 inch lead and ±0.001 inch per foot lead accuracy rotated 24 times theoretically moves the nut 12 inches.
24 Revolutions X .500 inches per revolution = 12.000 inches of travel with a Lead accuracy of .001 per foot, actual travel could be from 11.999 to 12.001 inches.
A ball screw assembly uses rolling elements to carry a load similar to an anti-friction (ball) bearing. These elements do not wear during normal use. Therefore, ball screw life is predictable and is determined by calculating the fatigue failure of the components. Proper lubrication, regular maintenance, and operation within specified limits will allow Ball Screws to operate to the predicted life.
Load Carrying Balls:
The bearing balls in contact with ball nut and ball screw sharing the load.
A load that tends to rotate the nut around the longitudinal axis of the screw.
The axial distance between threads. Pitch is equal to the lead in a single start screw.
Preload is an internal force introduced between a ball nut and screw assembly that eliminates free axial and radial lash. Preloaded assemblies provide excellent repeatability and increased system stiffness. Preloading is achieved either by using two nuts and forcing them apart or by shifting the circuits within a single nut. Nook Industries has a variety of preload ball nut designs available.
When bearing balls circulate in a ball nut, a ball enters the ball path between the nut and screw carrying the load one or more turns around the screw. The bearing ball is then picked up and returned to the beginning of the circuit through the return guide.
The diameter of the screw measured at the bottom of the thread. This is the diameter used for column strength, critical speed calculations and end machining considerations.
The number of independent threads on the screw shaft; typically one, two or four.
The axial movement between a new SBN or SGN ball nut and screw will range from 0.003" to 0.015" depending on size. When less than standard lash is desired, SBN and SGN ball nuts can be custom-fit to a specific screw with selected bearing balls to minimize lash to 0.003" to 0.005" depending on ball size. Select fitting may result in lower life.
A load that is applied radially to the nut.
The maximum thrust load – including shock – that can be applied to the ball nut without damaging the assembly.
Although Ball Screws are manufactured from straight, cylindrical material, internal stresses may cause the material to bend or yield. When ordering random lengths or cut material without end machining, straightening is recommended. Handling or machining of screws can also cause the material to bend or yield. Before, during and after machining, additional straightening is required.
A load that tends to “stretch” the screw.
A load parallel to and concentric with the axis of the screw.
Last edited by widgitmaster; 05-21-2006 at 10:06 PM.
It's not what you take away, it's what you are left with that counts!
You forgot to post the formula in the "back driving torque" section.
Somewhere, I have seen a thread on how to "repair" a ballscrew if you let the balls out by removing the nut. That information would also be a useful adjunct, as would a description of how to mount a ball screw with angular contact bearings to minimize the axial play. Lastly, since every boy and his dog is constantly wanting to know whether his latest eBay find will "work", noting down the formulas to convert speed in inches per minute and resolution based on steps or encoder resolution would be useful. If you had all of that, it would be a full 2-semester ball screw introduction.
The idea of a "fits all" spreadsheet is cool BUT it has its limitations.
With regard to the support bearing design: The screws and end bearings are designed to be "complimentary". Thus, the package is optimized with regard to the intended operating environment as well as the load capabilities of the components.
THus in some cases, you'll need a duplex end bearing configuration and in others a quad. Although a generic preload is usually adequate, specific applications need more or less. Even when you know how to preload bearings, it is not a thing that is adviseable for DIY involvement.
Technically, the proper/correct ball screw support bearing is going to be MUCH, MUCH more expensive than a more popular A/C bearing or a kluged up ABEC 1 set of deep groove ball bearing. Hence, the "can I use this instead of that to save money" situation inevitably develops, especially with the DIY crowd. Or, "...it takes these, I have/can get these from Free-bay, can I use them or modify them to make 'em work???" Probably not but I tend to steer clear of getting involved in such discussions.
Think the price difference isn't substantial?? Price out some 20TAC47's ball screws bearings from NSK then price out some 7204CtYDUHP4's then price out some 2 @ 7204BYG (which will have to be preload mofied after purchase). Performance in the application is inversely proportional to price and the difference is inevitable due to the production quantity issue.
There are currently bearing handbooks that show the machine designer how to design for and select bearings for just about any application. Links to them are here:
Sadly, a lot of bearings are used pursuant to Free-bay availability (price) and/or happenstance as opposed to true numbers based applications engineering. Being as price sensitive as the next guy, I realize that you can't buy what you want, especially for a hobby. Yet for a business, can you afford NOT to buy the good stuff???
Sometimes you have to look at it like this: If you wake up in the middle of the night with the signs of having a heart attack, you DON'T go shopping on Free-bay for a doctor or hospital. You go to the nearest one and trust that they'll give you the best in treatment and care - you'll figure out how to pay the bill later.....
When it comes to the hobby, you can usually cut corners and adjust your expectations accordingly. Your game, your rules, your choices. But, when it comes to a business, buy the best and you won't be disappointed.
Should someone wish to create a spreadsheet, I'd be inclined to get the NSK E1102 catalog for starters as it has a bunch of applications data provided.
NC Cams, you lost me on the last post.
Are you talking about the bearing blocks that the machined ends of the ballscrews fit into? What is a "quad" end bearing configuration?
Also, I'm a bit confused as to why you put particular emphasis on the end bearings. Aren't there a lot of other potential sources of error that are more significant?
NC, calm down man! You're scaring the kids...
I recognize that what you say is absolutely true, you get what you pay for, you have to know what you are doing, yada, yada, yada. HOWEVER, and this is a BIG however: we are not talking about refitting a Haas VMC with skate bearings on the spindle and then running it at 20,000 rpm.
Rather, we're talking about trying to improve some Asian mill bought at Harbor Fright or perhaps how to do better than an ACME screw with no backlash compensation whatsoever. We're improving on rack and pinion, chain, or belt drives. In short, we are doing hobby/junkyard CNC. The guy reworking the Haas VMC isn't going to be reading this Ballscrew 101 thread anyway!
That Haas guy doesn't have time to screw around (pardon the pun) with eBay ballscrews. He has most likely already purchased a complete set of goodies at a premium price, slapped 'em back on his mill, and is happily making chips and $$$ again. If not, there is always the special Ballscrew Grad School thread which can be started when we finish Ballscrew 101.
So, we need to do what one always does in a situation like this. We need some rules of thumb and some simple examples. We do not need to spec out $1000 worth of bearings so some guy can make his Asian round column mill drill eek out 0.0005" accuracy on one-axis (he can't afford to upgrade the other axes after that anyway, and his one axis has so much way slop it doesn't approach the theoretical ballscrew performance).
Now, we are going to be dealing with light duty applications. We're probably going to do good to hit 60 ipm rapids, and if we go past that, we are on our own. We want the widest range of possibilites so as not to limit our ability to sample from the freeBay smorgasbord of possibilities. We therefore need to keep some very simple principles in mind:
- We must use angular contact bearings. Here are 4-6 categories of those bearings and the pros and cons of each. We need the bearing selection equivalent of 5Bears (the Swede) linear rail selection guide--e.g. he calls out SHS, HSR, etc. in order of quality best to worst with a brief explanation of why. A little discussion of how to map these categories to different bearing manufacturers would be slick. Links to suppliers with good prices would also be welcome.
- We must mount them in a certain way. A drawing makes this understandable.
- We must fit them to the mounting block according to some certain tolerances. BTW, here is how we can accomplish this in our pitiful small home shop ("I am attempting to build a mnemonic device with stone knives and bearskins," says Mr Spock).
- We need to machine the end of the ballscrew thus in order to fit the angular contact bearings. There's a shoulder, a shaft fitted with x tolerance to the bearings, and a threaded end.
- We want to preload those bearings to x spec.
- Some idea of how to measure the performance of our result and measure/diagnose where we may have gone wrong would also be very helpful.
Is all of that stuff hard to figure out, approximate, and not all that optimal if applied literally? Absolutely! Can it help a guy who has got some cheap ballscrews off eBay build a nice CNC conversion? Sure. Does everyone win? Yup.
The Swede has done a pretty good job on this on his 5Bears site. Folks should take a look. Swedes notes go further than anything seen elsewhere, and he also did a very nice note on CNCZone (the original ballscrew 101) which may be found here:
Here is also a nice little thread on how to turn down the ballscrew to fit your bearings:
How do you turn-down the ends on a ball-screw?
It would be great if someone wants to take it to the next level of a "ballscrew cookbook."
Ball screw support bearings are mounted at the ends of the ball screws - the "blocks" carry the bearings in whatever configuration is contained therein. The thrust absorbing ends are duplex, triplex or quad in configuration:
Duplex = two bearings mounted in tandem, at one end (light duty) or both ends (HD)
Triplex = three bearings mounted in tandem, usually for higher thrust than double duplex.
Quad = four bearings mounted in tandem (killer thrust or very heavy, fast tables)
As you "tandem" mount bearings, you gain axial and/or radial capacity because more sets of rolling elements carry the applied radial/axial loads.
When it comes to ball screws, you REALLY want to use the "TAC" style bearings as the are A/C type with a contact angle particularly suited for absorbing axial thrust with low rolling resistance and high stiffness (IE: 20TAC47 for a Bridgeport mill).
A 7204A5TYDUHP4 will "fit" but not be as stiff AXIALLY as the 20TAC47. These should be cheaper. Then you get to a 7204CTYDUHP4. These are same size but even less axial stiffness. For lowest cost, a pair of 7204BYG's is the call BUT they won't be stiff because of NO preload.
To learn, find what the differences are, I'd suggest checking out the literature from the NSK site. No sense to repost the written word, especially when it is so clearly written and well explained by bearing experts.
Having done bearing applications work, I've found that people like to oversimplify the loading because it is a PITA to figure it out so as to compute the bearing designs needed. When it comes to over simplifying ball screw bearings, it is easy to make something that "fits" but doesn't work...
However, one has to really figure out/understand the loading to do PROPER bearing applications work. Otherwise, you're merely guessing which tends to be a lot easier albeit much more risky when it comes to assuring performance potential. Guesstimates can be trusted for neither accuracy nor durability.
I've published a numer of treatises about my Extrak and the work we went thru trying to tune up the ball screws and/or the support bearings several times on this site. Not the least of which was shimming and/or exchanging bearings until I was blue in the face and on and on.
Ultimately, ball screw bearings (20TAC47's) solved ALL the problems. I was lucky enough to find some surplus and got them for a pittance. However, in trying to make do, I learned that "make do" stuff works OK but no where's near as good as the RIGHT stuff.
I, too still do my share of economizing. However, I've learned that ABEC 1 bearings, no matter how you massage then, won't give you what a good set of ABEC 7 (ISO P4's) will give - PERIOD - when it comes to true machine tool work.
Error from the ball screw SYSTEM (important clarifier) is cumulative product of the screw and the bearings. As you move in any direction without change, the lead error and any axial runout error from the bearings affects you most of all.
HOWEVER, as soon as you "dwell" or change direction, the lack of stiffness in the "system" and/or backlash in the ball nut and/or end bearings (whatever you want to call them) comes into play. NOTE: Hysterisis is NOT the same as back lash.
Keep in mind that ANY slop (gibbs, ways, belt drives, software error, etc) will show up in the part and it amplifies and/or adds to screw system error. Easiest way to check the system is to mill a circle.
IF it ain't round and if there are flat spots at the 12, 3, 6 and 9 o'clock positions, you have an error that is accumulating that has to be eliminated. When you do get all the error contributors minimized/eliminated, the part will be "rounder".
After several months of screwing around and several thousand $$$'s of expenditures, we got a lowly Bridgeport Eztrak to darn near do what a Haas VMC could do for a small fraction of the outlay.
As in roundness in near single digit tenths and direction change dwells of 2-4 tenths. Far cry from thousandths out of round and ten times what we originally saw for flats at direction change. It (error elimination) can be done only it takes time, $$$ and attention to details, bit by bit.
I"d love to find a way to provide generic info that would help folks understand bearings and ball screw bearings better. Sadly, ball screw design and tuning is more of a tailor made reality that people try to solve with "off the rack" solutions.....
Some things are not that easy to oversimplify.
Last edited by NC Cams; 05-23-2006 at 07:41 PM. Reason: fixed omissions
Good info. NC Cams, one more question. Err.. two rather.
When installing a ballscrew, is it recommended to thread one of the machined ends so that slight tension can be put on the screw?
Also, are angular contact bearings designed to take axial loads in both directions or just one?
Ball screw bearings are typically mounted DF to allow for better self aligning capability under preload. Thus, if you have a duplex set, you're threaded end will merely tighten/preload the bearings against each other. On a mill, they typically mount the preloaded bearing end solid and let the other end slip/float (slip fit no slop) in the housing after they mound the bearing axially solid with a threaded nut.
To "stretch" the screw, I'd be more inclined to shim between outer race and screw side of housing to obtain a position preload in the screw.
You really need to be careful when you start stretching screws, especially when you're trying to "straighten" them. Better to get straight ones or straighten them than to yank on them at assembly.
If you do this (stretch em), you'd better use duplexed pairs on both ends of the screw to better control ball screw position preload.
Shim type position preload at one end relies TOO much on table and screw not growing/shrinking with temp. Preload and hysterisis won't/can't stay constant...
The printed NSK catalog I have shows a number of pictures of acceptable mounting methods.... I hope/suspect the digital ones will do so likewise as they tended to be pretty thorough with their tech info.
Look around here for a machine tool bearing catalog here:
A/C bearings are designed to absorb unilateral thrust such that you try to compress the contact angle. Only when you mount 2 in DF or DB configuration can you absorb bilateral thrust. Again, check the manuals/catalogs for details.
Last edited by NC Cams; 05-23-2006 at 08:43 PM. Reason: forgot to cover A/C's
nice thread and lots a clear and usefull info thx for this guys lot is clear to me now BUT...
i'ld love a drawing of a ballscrew assembly plz Please PLEASE(cuz i'm sure i'm not the only one whomes english is SO bad i just cant understand whats meant by all the tradelanguage)
and what does :Ball screw bearings are typically mounted ........DF......to allow for better self aligning capability under preload say ?
one more question bout the back driving torque on ballscrews isn't this the reason that its so important to have sterpper/servo's with enough torque to withstand this torque wich is expressed in Newton meters over here in europe and i thaght american used oz. please someone correct/enlighten me
thx in advance guys
Here is a link to the Rockford Ball Screw Manual, it has a fair amount of information, as well as some nice illistrations!
Last edited by widgitmaster; 06-01-2006 at 06:11 PM. Reason: typo
It's not what you take away, it's what you are left with that counts!
I one does a Google search for "bearing preloading DF" it turns out that the VERY FIRST search result that came up was:
It can't be explained any more simply nor anymore concisely.
Theoretically, a ball screw an be back driven - BUT depending on the ball pitch (as in turns per inch), it may or may not be easy to do so.
Torque is expressed in newton/meters whereas torque in USA is usually expressed in oz-in or ft-lbs. The use of simply math should enable you to convert newtons to oz or lb force and meters/mm to feet or inches.
A 20 turn per inch screw will be harder to back drive than a 10 turn/inch which will be harder than a 5 and so on. Factor in the gear reduction back to the motor and the ease changes even more. A 1:2 motor/screw drive will be easier to back drive than a 1:1 which will be easier to back drive thatn a 2:1