[boyvox]

I have a worn out Thermwood that's about 12-15 years old. It needs some new linear motion components. I noticed that the ballscrew support bearings are single Timken tapered roller bearings. Most of the ballscrew support blocks that I see for sale have angular contact bearings that use round ball bearings instead of the tapered roller bearings. I'd like some opinions/information on the use of tapered roller bearings for end supports. They are in fine condition but I am wondering if the more common angular contact bearing supports I see for sale would be better? If anyone can point me to a 101 on either of these types of bearings for ballscrew supports that would be great. These seem to be loaded only by tightening the nut on the end of the ballscrew and squeezing the bearings inward. I can post some pics if you want to see. I am thinking about just replacing them all with new blocks with angular contact bearings. Any reason I shouldn't?

[Pres]

... I'd like some opinions/information on the use of tapered roller bearings for end supports. They are in fine condition but I am wondering if the more common angular contact bearing supports I see for sale would be better? .... I am thinking about just replacing them all with new blocks with angular contact bearings. Any reason I shouldn't?

Why would you want to do this?
The tapered roller bearings are bigger, bulkier and alot more $, but they will probably outlast every other moveable component on the machine -given reasonable care.

You mentioned they are "in fine condition" so it would be better you concentrate your efforts and dollars on other components, IMHO.

Pres

[CarbideBob]

Use the standard tapered bearings. Angular contact bearings for ballscrew support have a special (higher) contact angle than standard angular contact bearings, are mega expensive, and must be mounted in matched pairs. Changing styles would require a lot of machining and modifications. The tapered bearings have a much higher axial thrust load capability which you need because both bearing ends are free and you have to preload the bearings by basically stretching the screw between the bearings. Personnaly I would never design a leadscrew mount using tapered bearings but if the machine already uses them changing over is a nightmare. Don't even think about putting one standard angular bearing on each end of the shaft. The low contact angle ( typically 15 degrees ) won't carry the preload and the bearings will losen up and die very quickly.

[Zumba]

Hey Bob,

Is the stretching the screw from end to end a typical way of doing it? It'd make much more sense to have two Timken bearings face to face on one end, with the preloading confined to the machined portion of the ballscrew. The other end of the screw would be supported by a standard radial BB to allow for thermal expansion.

I believe this is how many lathe spindles are done.

[CarbideBob]

Stretching the screw from the ends is the cheap way of doing it. On most CNC machines a pair of preloaded 60 degree angular contact bearings are attached to the drive end of the screw with a floating bearing on the other end. All thrust loads are absorbed by the pair's fixed mount. This is known as fixed-simple mounting and yes it's the same as most spindle designs ( although spindle bearings have a 35 to 40 degree contact angle ) . Some machine's leadscrews don't even have the floating bearing ( the leadscrew just sticks out in the air). This is called fixed-free mounting.

(added)
I'm not familair with your machine. How long is the screw and what's the diameter? How fast does it spin? Sometimes on long small screws tensioning helps to control screw whip.

[boyvox]

The tapered roller bearings that are on my machine are arranged with just one on either end and you must tighten nuts on either end of the screw to squeeze the bearings inward on either end. If the screw expanded (got longer) the nuts would not be holding the bearings in tight and there would be miniscule slop in the bearings. It doesn't make sense to me but that is how the machine came to me (it was free). The ballscrews are just rolled warner screws with single ballnuts and the longest one is around 70 inches. They are .5 inch pitch and .5 inch diameter. I haven't set up the motors and drives yet so I dont' know what my ultimate rpms will be on these screws but probably not very fast. The shorter 39" ballscrew and ballnut is shot and I'm replacing it because it's not a very good screw. I'm working on getting a ground ballscrew with a pre-loaded ballnut. I fear that if I get the ends of my new screws ground to fit these tapered bearings I will be dissapointed because the bearings will not do a very good job holding the screw accurately. Perhaps this is an unreasonable fear....I don't know. Here are some pictures of it.

[Zumba]

I'd ditch the Timkens and go with angular contact bearings in a "fixed-simple" arrangement as CarbideBob and I discussed.

I would also recommend at the very least, a 3/4" diameter ballscrew.

[boyvox]

Yes, that is what I was initially thinking, as it just seems to be the most common method and makes more sense to me. I probably have at least 6 months before I'll start buying parts. Now....ballscrew diameter......should I post questions in another topic? I don't understand why these ones are so small (1/2
" diameter)? Is it because of sag on the long 70" one or was it just for cost effectiveness? What happens if I use too small or too big of a ballscrew diameter (is there such thing?) I was looking at some 16mm diameter ones...would that be on the small side?

-thanks for all the help,
Ben

[CarbideBob]

Hi Ben,
Aha, a picture truly is worth a thousand words. Seeing your machine I believe the tapered bearings are a must have item. I will try to explain my thinking. You have a 1/2 diameter screw 70 inches long. Think of your screw as a cable stretched between to poles. It will droop down in the center from it's own weight. Now we start to slowly rotate the cable. As it rotates it will flex and continue to hang down in the center. If we rotate it faster and faster eventually it will start to whip around like a jump rope. This (along with some other factors) determines how fast you can spin a screw. The critical speed for a 1/2 x 70 inch is 250 RPM for a simple-simple mount. Using a 20% safety factor we get a practical top speed of 200 RPM (250x0.8). With a screw pitch of .5 this gives us 100 Inch Per Minute top speed (200RPM x 0.5pitch). Looks like you have about 50 inches of travel so this means 30 seconds to get from the front to back. This length/diameter also has a real problem with column loading but it we put the screw in tension this goes away. By tensioning the screw between the two mounts it becomes like a string on a guitar. Tightening the string stiffens it and raises the frequency that it rings at. We can use this to spin the screw faster. By tensioning the screw we should get at least a 2.3 times increase in speed or 460 RPM and 230 IPM (15 seconds front to back). Higher tensions will allow higher speeds. So the only trick is to maintain a constant tension as the bearings wear and the screw grows and shrinks (and of course the machine base grows and shrinks as temperature changes.
Are you familiar with Belleville disc springs? They are cone shaped washers made of spring steel and are made in a variety of spring pressures. Put one of these spring washers on the non-driven end of the screw with the small side pushing on your bearing. Put on the nut and tighten down the nut to achieve your desired preload (you made need a flat washer if the Belleville is larger than your nut and put a drop of Locktite blue inside the nut before assembly). Now your bearing can move on the shaft as it grows and shrinks but the tension remains constant. How much tension? I'm guessing at your bearings, but from about this size in my Timken catalog they should have a thrust load capacity of about 350 to 400 lbs. I would start around 150 lbs. and see how the shaft rotates. You can stack these washers to get higher forces. You'll need at least 60 lbs. to prevent the bearings from unloading when the axis reverses and here more will be better as long as you don't exceed the bearings ratings. I buy these from McMaster-Carr for about $6.00 for a pack of 12 (pg 3573 in my catalog). Let me know the size of the bearing ID and I can get you a part number.
Why stick with the tapered bearings? Basically because you need their high axial load capabilities to absorb the tension.
CAVET EMPOR! This is a drastic oversimplification of ballscrew critical speeds and my friends in the ballscrew manufacturing business would laugh at my examples. I'm tying to come up with easy to understand analogies.
On the subject of ground vs. rolled ballscrews you should remember that a CNC needs repeatability more than accuracy. I built my first CNC grinder in 1978 and it uses a double-nut rolled thread Warner screw. It still makes parts within .0001 and has just over 3.7 million cycles on the original ballscrews. If you program it for a 6.0000 inch part the first one might be 5.9920 long but then we change the program by .0080 and were off and running. ( It's repeatable but not accurate). On my 4 and 5 axis contour grinders I have to use ultra-precision ground screws because here we are synchronize multiple axis within microns. These are absurdly expensive (about $5000.00 for a 10 inch screw). Unless your getting a heck of a deal on a ground screw consider using double-nut rolled ones.
It looks like your machine uses NEMA-34 stepper motors. If you don't have drives yet please consider using micro-stepping drives. They really are worth the extra money if your controller has a high enough step rate. They smooth out the performance of the motors eliminating resonance problems and they just plain sound better when running. All my old full and half step drives went into the dumpster years ago.
Well, enough rambling. I hope some of this makes some sense or at least gives you some something to think about. These are just some ideas off the top of my head and feel free to ignore them if you think I’m full out it. Remember there’s always more than one way to skin a cat
PS. The optimal screw dia. for a 70 inch span would be around 1.5 inches but you would need much larger motors to accel and decel that much mass.

[Zumba]

Good info, Bob. I've never heard of the Belleville washer method for stretching screws while taking care of thermal growth, but it sounds logical and pretty cool.

How does scretching the screw between two bearings affect the lead accuracy? Since you suggested 150lbs for boyvox's machine, how much would 150lbs stretch a 1/2"screw? How likely would it be that the stretch would be relatively even along the entire length of the screw (keep in mind, low cost rolled screw)?

BTW, just noticed that you joined only a few days ago. Good to have you here on the forum. You obviously have a wealth of knowledge and I hope you don't mind less experienced folks like me picking your brain. :rainfro:

[NinerSevenTango]

Be careful of spring preloading. Machine acceleration against the spring could compress it, sacrificing accuracy and unloading the other bearing in the process. Your bearing preload will be decreased by the inertia and acceleration in the direction that compresses the spring... For that reason, if it is done, the spring better be at the driven end, you don't want the unloaded bearing taking radial loads.

Because of this, you might be better off trying to control temperature in the shop and just sticking with the original scheme.

--97T--

[boyvox]

CarbideBob,

You say

The critical speed for a 1/2 x 70 inch is 250 RPM for a simple-simple mount

and also

The optimal screw dia. for a 70 inch span would be around 1.5 inches but you would need much larger motors to accel and decel that much mass

Could you give me some information on how I calculate the critical speeds for different screw dimensions, and the associated parameters such as motor size, etc. Is there a page anywhere with formulas I can learn, so I can calculate different configurations? Anyone who can provide formulas for screw performace, please post if you feel like it.

Thanks,

Ben

[Zumba]

I'll field this one.

http://www.nookindustries.com/ball/BallCharts.cfm

"INCH SRT, XPR AND SGT CRITICAL SPEED"

In your case, look at C and D (bottom of the chart). B is the fixed-simple mounting scheme that most of us (should) use... your original setup with the two Timkens and tensioned screw is actually somewhere between C and D. Not a rotating nut, but the tension gives you more stability than merely fixed ends.

The numbers on the right side of the charts should be pretty obvious... 0500-0200 means 1/2" x .200. So 70" length in the "C" row is a bit under 90IPM (note: I forgot how to read logarithmic charts). Anyway, that translates to 450RPM. CarbideBob's 250RPM rating is for a different mounting method, hence the discrepency.