[HackMax]

It's not a mistake. Single nuts are simply not designed to have zero backlash. Ballnuts are for low-friction drives. If you want zero backlash (there are many applications where backlash doesn't matter), you use a double nut. It is *critical* that the second nut be able to "float" a bit, or you WILL get binding in some positions and/or backlash in others, as the screw pitch is not absolutely constant. Rigidly locking the two nuts together is NOT a good idea, and will generally lead to rapid wear.

Regards,
Ray L.

Yeah Ray I understand fully what your saying I just thought the specs would be smaller than .010 inch. I guess I shouldn't worry about .0035 inch but I'd still be happier with .001 or less. Even the spring washers would let some backlash appear with a heavy enough cut but thats another reason to take a small final cut.

Rick

[caleb105]

So,

Do any of you have PICTURES of a double-nut setup using these spring/bellville washers?

I'd like to see what other people have done.

Thanks!

-Caleb105-

[SCzEngrgGroup]

Here's how I did mine:

http://cid-f31d82e92e218e66.skydrive...n/P1040521.JPG

The Belleville washers are not shown in the photos, as I didn't have them yet when I took these. They simpy go between the two nuts, held in place by the aluminum "sleeve" between the nuts. Pre-load is easily set by adding/removing/shimming the Belleville washers. These are 1" screws, set to about 250# preload.

BTW - MSN SUCKS!!!! It seems they've "enhanced the experience", and I can no longer order the photos as I want. You upload them, and they go in whatever order the Gods of Redmond deem appropriate, and there is no way to change it! What a bunch of morons....

Regards,
Ray L.

[caleb105]

Here's how I did mine:

http://cid-f31d82e92e218e66.skydrive...n/P1040521.JPG

The Belleville washers are not shown in the photos, as I didn't have them yet when I took these. They simpy go between the two nuts, held in place by the aluminum "sleeve" between the nuts. Pre-load is easily set by adding/removing/shimming the Belleville washers. These are 1" screws, set to about 250# preload.

BTW - MSN SUCKS!!!! It seems they've "enhanced the experience", and I can no longer order the photos as I want. You upload them, and they go in whatever order the Gods of Redmond deem appropriate, and there is no way to change it! What a bunch of morons....

Regards,
Ray L.

Ray,

Thanks for the picture. A couple more questions:

The two ballscrews are being held together b/c the 2nd ballnut can't rotate (b/c of the pin), and therefore can't spin away from the 1st ballnut?

Also, you have 250# of preload? Per the previous posts, I was assuming 18-27lbs (10-15% of 180lbs) of preload was good for a .621 ballscrew!

Thanks!

-Caleb105-

[SCzEngrgGroup]

Ray,

Thanks for the picture. A couple more questions:

The two ballscrews are being held together b/c the 2nd ballnut can't rotate (b/c of the pin), and therefore can't spin away from the 1st ballnut?

Also, you have 250# of preload? Per the previous posts, I was assuming 18-27lbs (10-15% of 180lbs) of preload was good for a .621 ballscrew!

Thanks!

-Caleb105-

Yes, the roll pin prevents the nuts from rotating w.r.t. each other. In commercial nuts, this is usually done with a dowel pin pressed into one nut, and a hole in the other, or something similar. These nuts were so hard, I could not make a dent in them, much less drill a hole.

Preload is 10-15 of the rated load of the screw, which is typically on the order of 1600# for a 5/8" screw. Mine are 1" screws, so the preload is higher. You ideally want the preload to be higher than the maximum machining load, as once the load on the nut exceeds the preload , you WILL get backlash due to compression of the Bellevilles. This will happen only in one direction, when the load is being carried by the "floating" nut. In the other direction, the load is carried entirely by the fixed nut, so the Bellevilles never see the load. That's why I sized the sleeve to allow 0.005" of movement, to limit the maximum deflection under this condition.

With square nuts, or nuts which are soft enough to be machined, you have many more options for how to constrain them.

Regards,
Ray L.

[cadmonkey]

Greg (CadMonkey) could you possibly be binding the second nut by tightening one of the bolts more than another thus skewing the nut so it rides the screw at an angle?

I think that is a major part of it, even though I have the bores for the countersunk SHCS a tight fit and tightened equally. I never mentioned a lot about it (I know I did mention something at some point but didn't make a big deal about it) because I left that as an "optional" item and couldn't completely blame it on the nut assembly as the stepper could be responsible too - underpowered. My thinking at this point is to go back and make a very similar floating block, but use threaded studs the block can ride on to stay oriented and then use heavy springs to push them apart or use 2 studs and 2 screws and compress (as I initially intended) instead of expand. Not sure yet. My initial desire to go with compression instead of expansion was to eliminate the possibility of the cutting forces overcoming the preload and as the screws/nuts wear, it would be possible to tighten things and continue to take up the lash as the systems wears to the point of replacement (maybe by the time I hand the archaic machine down to my grandkids at the rate I've been using it...)

I'm getting .0035' backlash from my $25 Roton ballnuts and when I saw that Nook has .01" for theirs I thought there has to be a mistake somewhere. It just doesn't sound right.

I think Nook quotes a worst case scenario so that you get what is printed or typically better - there is variation in all rolled screws just because of how they are made, they're going to elongate and contract during hardening and straightening. Also, the SBN nuts aren't intended for the XPR screws I used anyways, they are intended for the SRT screws are the general transport screws not intended for precision positioning and have an even worse lead variance. The nuts intended for the XPR screws are the same as those for a ground ballscrew - they are ground instead of cut, thus their higher cost, but preloading the cheaper nuts is less expensive and can get the same or better results than using the higher priced nuts if you do it right (which I haven't...yet...).

[caleb105]

The spring pre-load should be on the order of 10-15% of the rated load for the screw, which is typically around 180 pounds for a 5/8" screw.
Regards,
Ray L.

Preload is 10-15 of the rated load of the screw, which is typically on the order of 1600# for a 5/8" screw. Mine are 1" screws, so the preload is higher.

Regards,
Ray L.

So is it 180 lbs or 1600 lbs?

[SCzEngrgGroup]

So is it 180 lbs or 1600 lbs?

Rated load is typically around 1600 pounds. Preload is typically around 180 pounds. Most of the ballscrew manufacturers have app notes or specs defining how to determine preload, and recommended values.

Regards,
Ray L.

[caleb105]

Rated load is typically around 1600 pounds. Preload is typically around 180 pounds. Most of the ballscrew manufacturers have app notes or specs defining how to determine preload, and recommended values.

Regards,
Ray L.

Aaaaaaaaah. Thanks Ray!

I'm looking at the XPR line by NOOK. They don't list a "Pre-load" range:

http://www.nookindustries.com/ball/B...631%20-%200200

Do I use that static load? I'm looking at the 2nd one down on this list. For some reason that one has a much higher load (3360lb) than the rest (2110lb).

Or do I use the dynamic load? Seems like the preloaded ones on this list are limited to 450lbs of dynamic load. Does that mean that they have 450lb of Pre-load? That's about 13.4% of the 3360 static load. Thoughts?

Thanks!

-Caleb105-

[SCzEngrgGroup]

Aaaaaaaaah. Thanks Ray!

I'm looking at the XPR line by NOOK. They don't list a "Pre-load" range:

http://www.nookindustries.com/ball/B...631%20-%200200

Do I use that static load? I'm looking at the 2nd one down on this list. For some reason that one has a much higher load (3360lb) than the rest (2110lb).

Or do I use the dynamic load? Seems like the preloaded ones on this list are limited to 450lbs of dynamic load. Does that mean that they have 450lb of Pre-load? That's about 13.4% of the 3360 static load. Thoughts?

Thanks!

-Caleb105-

Use the static load. The most important thing is to make sure the preload exceeds your maximum machining load, which is not a problem on a small machine. Too loose, and you'll have backlash. Too tight, and it won't turn smoothly. On a small machine, that does not need to run 40 hours a week for 20 years, you really don't need to be too fussy about it. I just set mine to the point that I could feel going further made the motion less smooth.

Regards,
Ray L.

[wisp]

Quote from the Hiwin ballscrew catalog, page 19:

http://www.hiwin.com/pdf/bs/ballscrews.pdf

Single nut preloading
There are two ways of preloading a
single nut. One is called “the oversizedball
preloading method”. The method
is to insert balls slightly larger than the
ball groove space (oversized balls) to
allow balls to contact at four points (Fig.
4.18).
The other way is called “The offset
pitch preloading method” as shown in
Fig. 4.19. The nut is ground to have a
δ value offset on the center pitch. This
method is used to replace the traditional
double nut preloading method and has
the benefit of a compact single nut with high stiffness via small preload force. However, it should not be used in heavy
duty preloading. The best preload force is below 5% of dynamic load (C).

The CNCFusion deluxe kits use the second method to achieve preload. The ballnuts do not contain oversize balls. The balls are .125", and the nuts are preloaded by ABBA.

Hope that answers any questions you may have had on the deluxe kits.

cheers!
Michael

[scudzuki]

I have an RF45 retrofit from Chris Rosequist, who has a great deal of CNC building experience. It came with Thomson ball screws with fixed preloaded double nuts. It does not bind with the small amount of preload I have set and I have .002" backlash total in each axis (and I'm sure some of this is from the bearing). Using a floating second nut with a spring backing it up sounds great, until cutting force exceeds the spring rate, then it's anybody's guess where your table really is. I am far more confident in the machines positioning accuracy with fixed nuts.

Joe

[SCzEngrgGroup]

I have an RF45 retrofit from Chris Rosequist, who has a great deal of CNC building experience. It came with Thomson ball screws with fixed preloaded double nuts. It does not bind with the small amount of preload I have set and I have .002" backlash total in each axis (and I'm sure some of this is from the bearing). Using a floating second nut with a spring backing it up sounds great, until cutting force exceeds the spring rate, then it's anybody's guess where your table really is. I am far more confident in the machines positioning accuracy with fixed nuts.

Joe

So you'd rather have 0.002" uncertainty at ALL times, rather than just when you've exceeded the design load of the scrdouble nut assembly? Doesn't make sense. With double nuts, you can pre-load the nuts with over-size balls to achieve a very small amount of backlash, then use the spring pre-load to remove that. Only when you over-drive the machine will you see any backlash at all.

Regards,
Ray L.

[philbur]

Surely if you have a double nut with preload then you should not have any measurable backlash in the nut assembly (when not under load). Are you 110% sure the backlash is not somewhere else.

If it is fixed preload then I don't understand when you say "with the small amount of preload I have set".

I guess spring loaded double nuts are more forgiving that fixed preload nuts when it comes to low cost screw manufacturing tolerances.

Phil

I have an RF45 retrofit from Chris Rosequist, who has a great deal of CNC building experience. It came with Thomson ball screws with fixed preloaded double nuts. It does not bind with the small amount of preload I have set and I have .002" backlash total in each axis (and I'm sure some of this is from the bearing). Using a floating second nut with a spring backing it up sounds great, until cutting force exceeds the spring rate, then it's anybody's guess where your table really is. I am far more confident in the machines positioning accuracy with fixed nuts.

Joe

[scudzuki]

As I said, I believe that the lash I measured is introduced elsewhere. And yes, I'd rather have a guaranteed upper limit of error than uncertainty at random times. Look at the specs of a real VMC; There's no variable in there for "when forces exceed some predetermined limit".

It's fixed but adjustable preload. There's a block with a hole tapped straight through, one nut tight (and loctited in) in one end and the other nut fixed axially (that is prevented from rotating), once the desired preload is achieved, with 2 setscrews in the other side of the block. The preload still needs to be set. I tightened it up myself when I tore down the entire machine, down from .005 per axis to .002. Not sure how you measure backlash but here's what I do.

Drill a 3/4" hole in a piece of alu clamped in the vise (or clamp a piece with a hole approx that size in the vise).

Jogging only in the x+ and y+ directions, indicate the spindle in to the center of the hole, zero x and y @ the center of the hole.

z+ out of the hole, move x-1 y-1, then x0 y0 and drop into the hole again to verify/compensate. Usually takes 2 tries before I am dead on, then it will repeat within a few tenths all day.

z+ out of the hole, move several inches x+ y+, then x0 y0 (negative direction moves in both axis)

drop into the hole again. The backlash is the amount that the spindle is off center.


Joe

[SCzEngrgGroup]

As I said, I believe that the lash I measured is introduced elsewhere. And yes, I'd rather have a guaranteed upper limit of error than uncertainty at random times. Look at the specs of a real VMC; There's no variable in there for "when forces exceed some predetermined limit".

It's fixed but adjustable preload. There's a block with a hole tapped straight through, one nut tight (and loctited in) in one end and the other nut fixed axially (that is prevented from rotating), once the desired preload is achieved, with 2 setscrews in the other side of the block. The preload still needs to be set. I tightened it up myself when I tore down the entire machine, down from .005 per axis to .002. Not sure how you measure backlash but here's what I do.

Drill a 3/4" hole in a piece of alu clamped in the vise (or clamp a piece with a hole approx that size in the vise).

Jogging only in the x+ and y+ directions, indicate the spindle in to the center of the hole, zero x and y @ the center of the hole.

z+ out of the hole, move x-1 y-1, then x0 y0 and drop into the hole again to verify/compensate. Usually takes 2 tries before I am dead on, then it will repeat within a few tenths all day.

z+ out of the hole, move several inches x+ y+, then x0 y0 (negative direction moves in both axis)

drop into the hole again. The backlash is the amount that the spindle is off center.


Joe

"I'd rather have a guaranteed upper limit of error than uncertainty at random times." - Even though that "uncertainty at random times" is of *exactly* the same magnitude as the uncertainty you have ALL the time.... VMCs use ground screws which don't require spring pre-load. But, then those screws cost 5-10X what rolled screws do....

[scudzuki]

You're saying that the maximum distance the "floating" nut backup up by disc springs can deflect is .002"? You can't be serious. What is the total accumulated backlash per axis with your setup? I bet we're talking about a little x3 table, right?

And what does the quality of the screw have to do with backlash-induced positioning error?

Joe

[SCzEngrgGroup]

You're saying that the maximum distance the "floating" nut backup up by disc springs can deflect is .002"? You can't be serious. What is the total accumulated backlash per axis with your setup? I bet we're talking about a little x3 table, right?

And what does the quality of the screw have to do with backlash-induced positioning error?

Joe

My machine is a very fast, very strong, very accurate servo-driven 9x49 Bridgeport Clone with a 3HP spindle, not a "little x3". I can do things on this machine no X3 could ever even dream of doing. I typically run a 4 cubic inch per minute removal rate in aluminum, so I'm applying FAR greater machining forces through my screws than any benchtop machine could possibly apply.

Your statement indicates you understand how double-nuts actaully work, and the forces acting on them. At, and above, the load where you overcome the pre-load, the maximum backlash you will ever see is the same backlash you would see if the springs were not there. If you load over-sized balls, this can easily be no more than 0.001" if you're using good quality screws and nuts. In any case, it is *EXACTLY*, no better, no worse, than the backlash you can achieve by rigidly mounting both nuts and pre-loading balls alone using whatever screws and nuts you're using.

The quality of the screw has everything to do with how much compliance you must provide to accomodate lead error in the screw. If you rigidly mount the nuts, you will have backlash in some screw locations, and binding in others, as the pitch of rolled screws is *not* perfect - hence the "lead error" spec on the screws - typically 0.004"/foot for rolled screws (other than Nook XPRs). This is precisely why spring-loaded double nuts are necessary with rolled screws. Ground screws are FAR more precise, typically having a maximum lead error spec of less than 0.0005"/foot. This requires less compliance in the nuts.

Regards,
Ray L.

[scudzuki]

What is the spring force exerted by the preload springs on the floating nut? If the tool pressure exceeds that spring pressure, the floating nut moves relative to the table (introducing positioning error) until coil bind occurs. I have an exact understanding of how preloaded double nuts work, thanks. If you can't understand that, then believe what you want.

If my Thomson rolled screws have an accuracy of .003" per foot, that's less than .001" of deviation in the less than 4" that my 2 ballnuts occupy.

Joe

[SCzEngrgGroup]

What is the spring force exerted by the preload springs on the floating nut? If the tool pressure exceeds that spring pressure, the floating nut moves relative to the table (introducing positioning error) until coil bind occurs. I have an exact understanding of how preloaded double nuts work, thanks. If you can't understand that, then believe what you want.

If my Thomson rolled screws have an accuracy of .003" per foot, that's less than .001" of deviation in the less than 4" that my 2 ballnuts occupy.

Joe

"If the tool pressure exceeds that spring pressure, the floating nut moves relative to the table (introducing positioning error) until coil bind occurs." - Nope, not true. The screw can move off-position *only* as far as it takes for the load to be picked up by the balls on the fixed nut, and that distance will be *precisely* equal to the amount of backlash in the fixed nut. It is 100% impossible to achieve "coil bind" in a double-nut assembly, unless you have a HUGE amount of backlash in the fixed nut. Or perhaps you can explain how further motion would be possible?

"If my Thomson rolled screws have an accuracy of .003" per foot, that's less than .001" of deviation in the less than 4" that my 2 ballnuts occupy." - Also not true. You are assuming the spec implies a constant, monotonic lead error, which it absolutely does not. In fact, quite the contrary. The spec simply defines the maximum error you will ever see between any two points on a one foot length of screw. It says absolutely nothing about how the error varies over that one foot length. If they reliably could produce a screw with a constant, monotonic error, they would have the ability to fix the error and sell perfect screws. But manufacturing processes simply don't work that way, hence the need for that spec. And hence the reason even high-end VMC use large screw mapping tables to correct the lead errors in their screws. A screw which had an error that increased to +0.0015" over nominal during the first two inches, then decreased to -0.0015" under nominal during the second two inches, and so on, would MEET the spec to the letter, but your 4" long double ballnut would see a worst case error of +0.003" in some positions, and -0.003" in other positions. This extreme error profile would be unusual, but the spec does NOT preclude this possibility, so designers must design around it.

Regards,
Ray L.