[bru102]

I was thinking about the coil spring as well, and concur with Ray L. If you were right, Pat, then disc springs wouldn't work well either... their spring constant is stiff, but not stiff enough to give you anything close to zero backlash. (Rockford warns that backlash will go as high as .015" on their nuts if you exceed preload, while being zero if you stay below preload.)

Also did a little more research on disc springs and found a very thorough engineering handbook at http://schnorr.com/docs/Handbook.pdf. Maybe more than you want to know, but the best reference I've found . They show it's acceptable to push the springs flat, but it puts most spring sizes a little beyond the desired stress levels, and they are likely to have shortened lives. So I still think preloading with the washer at 75% (spec) deflection and limiting the travel between the nuts is the way to go... and the limit only matters if you are worried about backlash if and when forces exceed preload.

Ray B.

[wildwestpat]

Hi Folks

There are many types of ball-nut and the choice is down to precision - service life - operating conditions and lastly cost.

Ray B - Bellville springs have a very low total deflection for a given force and as part of their manufacturing process they are squashed flat. I think this helps ensure they have the calibrated rate specified for the less than 75% load deflection. The load versus deflection above 75% is rapidly increased and is not defined but in my experience they require a minimum of two to three times the 75% force to flatten them. As to the life of the spring there are a lot of factors and the makers run a design service but in the mechanisms I have designed life has not been an issue with springs operating at around 90%. These springs are used to ensure a minimum pressure between surfaces in an assembly which regularly rocks the washers flat. However I would expect the life to be shorter than for 75% or less variations.

Ray L. The point I was considering is that the natural backlash in a ball-nut is relatively small. The nut also has a spring deflection all of its own due to the compression of the balls into the sides of the screw wall / internal track much in the same way as a ball bearing race deflects as it is loaded radially. Since these deflections are small it is my contention that the total movement between to ball-nuts is small. Obviously the limit is brinelling of either the balls or the screw or internal trackway within the ball-nut.

The Rev Richard points out the increased torque needed on a particular ball-nut design (Rotron). This is a very valid point. Looking at the setting instructions in the Rotron information posted by Rev R. The load changes are taken by the screw locking and it is my interpretation that the spring is redundant once the setting has been locked up. This shows that Rotron have good control over the pitch of their screws as far as the length of the total nut is concerned.

Out of the low cost designs of ball-nut my preference is for the ISEL Automation ones as they have a screw with which to adjust the backlash. I find they remain set for long periods but I only have used them at low shaft rotation speeds and they are protected by gaiters as well as the supplied dust seals which act to help retain the lubricant. BUT with less cost sensitivity I would love to use ground screws and double nuts manufactured as double nuts.

Having re read this thread can I suggest that the setting should now be as follows:-

1. Bring the ball-nuts together with a disc spring between them. Use some form of clamp or screws to make this a controllable operation

2. Continue forcing the ball-nuts together whilst monitoring the torque necessary to rotate the screw. The Torque rises sharply as the slop is removed.

3. Measure the gap between the nuts and use a solid shim of that dimension to permanently space the nuts.

4. There would now be no springs to add to the chatter / unwanted movements induced by the cutting action

This would be quick to do if you can find a spare Bellville washer that is smaller in OD than the nuts but with a hole to clear the screw. Measure the gap with feeler gauges and use this to decide on the shim pack. I think the setting would remain fairly constant through out a long period and would be quick to reset.

Hope this helps. Regards - Pat

[SCzEngrgGroup]

Having re read this thread can I suggest that the setting should now be as follows:-

1. Bring the ball-nuts together with a disc spring between them. Use some form of clamp or screws to make this a controllable operation

2. Continue forcing the ball-nuts together whilst monitoring the torque necessary to rotate the screw. The Torque rises sharply as the slop is removed.

3. Measure the gap between the nuts and use a solid shim of that dimension to permanently space the nuts.

4. There would now be no springs to add to the chatter / unwanted movements induced by the cutting action

This would be quick to do if you can find a spare Bellville washer that is smaller in OD than the nuts but with a hole to clear the screw. Measure the gap with feeler gauges and use this to decide on the shim pack. I think the setting would remain fairly constant through out a long period and would be quick to reset.

Hope this helps. Regards - Pat

Sounds to me like you've never tried to actually fabricate a double-nut assembly using rolled ballscrews.... The lead error (typically +/-0.003"/foot) is such that if you attempt to rigidly mount then as you describe, you WILL end up with backlash, or binding, at certain positions, due entirely to the lead error and other tolerances on the components themselves. I built my machine using the best rolled screws out there, Nook XPRs, which have a max lead error of only +/-0.001"/foot, and saw exactly that when I tried mounting them rigidly. With rolled screws, the springs are NECESSARY. If you want zero backlash without springs, you have no option but ground screws at 3-5X the price.

But, again, as long as the spring pre-load exceeds the maximum load, the performance of a spring-loaded double-nut, and a rigidly mounted double nut or true anti-backlash nut will be IDENTICAL. And, in one direction only, you are even free to exceed the spring pre-load and will still introduce no backlash. Work out the physics of the loading on the nuts and spring, and you will see this is true.

Regards,
Ray L.

[wildwestpat]

Hi Ray L.

I have no means of checking the variations in pitch or the variation in lead in two inches increments over the entire length of the screw. The design of the backlash elimination has only to accomodate the maximum deviation in the short distance vartiation in pitch and the variation in clearances over the distance between the first and last ball typically two inches. The longest screw travel I use is 350mm which may account for my lack of appreciation of the problem with springs. With the ISEL Automation ballnut / screw combo classed as C7 the backlash is non existent with the adjustment screw correctly set. I have used double nut ball screws but these may have been precision C5 screws.

I suspect but do not know that the shape of the ground in ball roads in the nut and the number of turns used as well as the size of the balls all have an impact on the feasability of creating a solid backlash free paired nuts.

I am just glad I discovered adjustable nuts that are both low cost and maintain their settings well. I agree that the ground ball screws are a delight and have found a pair of 8mm ones on ebay by NSK. I have tried measuring the backlash but the most sensitive DTI I have only reads to 0.001mm. For me this is as good as it is going to get sub $100 including tax.

Regards - Pat

[nateman_doo]

in short, it seems to me that there is slop in ball screws just like there is acme screws, just much much less. I think as long as your spring pressure equals more then the "linear weight" of your table (the resistance of sliding the table) and more then the cutting forces, then you should be good to go. Is this correct?

[wildwestpat]

In one word 'yes'.

To keep costs and expectations in the real world determine what you need in terms of repeatability as this differs from absolute accuracy over the total size of the work piece and the production run if necessary. Determine the accuracy and repeatability of the machine under CNC control. Whilst CNC is good for both accuracy and repeatability it has one major problem and that is with co-ordinate tolerances with mating parts. (A simple example is the need to position a multi hole flange where the mating part has a known tolerance on both the pitch circle and the spacing round the pitch i.e. a pipe flange.) What I am trying to indicate is the necessity to work with realistic tolerances rather than try and eliminate all deviations from the notionally perfect.

Regards - Pat

[nateman_doo]

I will stick with the wave springs then. I found the actual belleville wave springs for ballnuts.

McMaster part # 94065K52

Ball Bearing Belleville Disc Spring Bearing #6001, 17.3mm ID, 27.7mm OD, .40mm Thk, packs of 10

Also some stainless steel washers about the same size.

[bru102]

Hi Nateman. Responding to your last two posts. The forces you need to overcome are moving the table/vise/work against friction, accelerating the same mass, and pushing the work into the cutting tool. I think you listed all the forces except acceleration... and that turns out to be important for clean cuts of curves and sharp corners.

You said earlier you are using linear bearings, so your force to keep things going is small, and acceleration and cutting are the interesting ones. How heavy do you expect the table plus work to be?? If it's 500lb, for example, you'll need 75lb to accelerate it at .15G which Bob Warfield suggested as a goal in another thread.

Your McMaster spring washers give you 18lb (80N) each. They are actually designed as "ball bearing" springs, and could be a little light depending on your desired preload. I noticed that Rockford's preloaded nut on your size screw can be adjusted up to 250lb of preload... and I think they are doing that with just 2 or 3 washers. Also, Rockford is OK with up to 30% preload, and 250lb of preload will "cost you" about 25 oz-in of torque from your drive.

Ray B.

[nateman_doo]

The top and middle plates are 79 lbs in total. Then the weight of the 2 24" rails, A 20 lbs tool plate, and thats about it. Most of my work is 1/4" thick 12x12 sheets of copper. The heaviest load my table will see is a 6" CNC vise (I forget how much they weight)

The motors being used are NEMA 23 Steppers KL23H2100-50-4B (3/8” Dual shaft with a flat) 570 oz-in

I didn't notice the 80N rating, so I found some heavier ones which are almost the exact size needed for the ballnut:

McMaster Part # 96475K273

Type
Belleville Disc Springs
Material
Stainless Steel
Stainless Steel Type
17-7 PH Stainless Steel
Minimum Inside Diameter
14.2 mm
Maximum Outside Diameter
27.9 mm
Thickness
1.0 mm
Overall Height
1.80 mm
Load
1,334 N (300 LBS)
Load Tolerance
±20%
Deflection at Load
.73 mm
Flat Load
1,431 N
Specifications Met
Deutsche Industrie Normen (DIN)
DIN Specifications (DIN)
DIN 2093

[bru102]

Sounds like an interesting application with the copper sheets. So... with an 80lb vise, you've got about 200lb to accelerate plus cutting forces. Not too bad. Thats just 30-40 lb for acceleration, plus cutting forces. That's a much lighter setup than mine, where I also have ways and gibs to contend with, and I'm designing around 500lb for tables and work.

So a spring that max's out at 300lb (spec, not flat) is probably about perfect...and interestingly very much like Rockfords for a 5/8" screw. You could start with a 100lb preload, and tighten things up after some testing if necessary, all the way up to 300lb. (The force you get is pretty linear with compression.)

But is 14.2mm minimum ID big enough?? Thought you had 16mm screws.

Also, I know you wanted a longer spring... did you know if you put the springs in series like ()()()( you can put as many as you want... the deflection for 7 springs, for example, is 7 times the amount for one, but the force is unchanged (still 300lb for your spring).

Ray B.

[nateman_doo]

CRRRAAAAAP! good catch.

McMaster Part: 96445K284

Type
Belleville Disc Springs
Material
Steel
Steel Type
Grade 6150 Chrome-Vanadium Steel
Minimum Inside Diameter
16.3 mm :wave:
Maximum Outside Diameter
31.5 mm
Thickness
1.25 mm
Overall Height
2.15 mm
Load
1,913 N
Load Tolerance
+15%, -7.5%
Deflection at Load
.675 mm
Flat Load
2,359 N
Specifications Met
Deutsche Industrie Normen (DIN)
DIN Specifications (DIN)
DIN 2093
Notes
Springs are cold formed and fine blanked. IDs and ODs are machined.

Its a hair larger then I want, but I guess reading that the OD is machined, I can turn it to trim the OD to 1.1" (shave off 3 mm)

[bru102]

I wonder if 16.3mm is enough clearance, and how you will get and keep things aligned so nothing rubs on the ball screw...

BTW...MSC direct, Reid Supply, and others also carry the washers, and in some different sizes. It can be a little tough to get an ID big enough with an OD small enough. One possibility would be a different mount where the OD was less restricted. (Or you can still go to the coil spring since length seems to not be an issue, and from Ray L's input, I think the "soft spring" became a non issue with enough preload.)

Ray B.

[nateman_doo]

I'ts pretty close. if I can't get those to work, I still have the existing wave springs that I do have. Plus they are machined. I think I can adapt them to fit my needs. May be extra work, but I got the time.

[bru102]

Sounds like you've got a plan. Thanks for starting the thread, I think several of us learned a bit through the discussion. Best of luck with your machine.

Ray B.

[nateman_doo]

I will keep the progress posted. Very slow, but it will be made regardless.

[nateman_doo]

Sidenote on oversized ball bearings:

I couldn't get the next size bearing to fit in the 1605 ballnuts. The next size up from .125, was 3.5mm (.1378).

Maybe it works for larger screws, but unless there is a size larger then 1/8, and smaller then 3.5mm I will just leave well enough alone.

[rowbare]

Sidenote on oversized ball bearings:

I couldn't get the next size bearing to fit in the 1605 ballnuts. The next size up from .125, was 3.5mm (.1378).

Maybe it works for larger screws, but unless there is a size larger then 1/8, and smaller then 3.5mm I will just leave well enough alone.

You should be able to find balls in about .0001 increments. One seller on eBay is ToolSupply, go to his store and select the ballscrew balls section. I think Hoss mentions another source in one of his threads or on his site.

bob

[nateman_doo]

So if the standard ones are 0.125, what size should I get? 0.126? I feel like the limiting factor is that plastic return. Thats where the 3.5mm ones got stuck in. I was only limiting myself to what I could find on McMasters page.

[bru102]

Is your 1605 an "anti-backlash" nut? I have a 2005 that is, and it feels smooth and tight, using offset tracks to limit "clearance". I can't measure backlash yet, but I found a post claiming a measured .002" backlash for the 1605 anti-backlash.

I believe his number since the screw isn't precision ground, and they need to allow for some lead variation in the screw. So the only way to get zero backlash (vs "anti-backlash!") is by adding a spring preload.

If your goal is zero...I'd worry that balls big enough to get that in one section of the screw are going to cause damage in a tighter section. (Or just the opposite, with a perfect fit in one section, but significant lash in another).

Ray B.

[wildwestpat]

Fitting over size balls means means that the ball is being ever so slightly distorted as is the screw track and the internal race way. This acts as a spring. Balls sticking in the return path / tube / pipe are often more a sign of either too many balls in circuit - or - dirt - or - wrong lubricant.

Some nut deigns respond better to over sive balls than others. Manufacturers are remarkable reticent about this as it possibly reduces sales of their precision grades.

Regards - Pat