[musicmkr]

Anyone who has considered using pairs of angular contact ball bearings for their lead screws was undoubtedly shocked by the price of these little gems. Unless you are a ebay-aholic then you will have to find another solution, or shell out a couple hundred dollars.

Unless you are building a very large mill, don't waste your money.

Deep Groove Ball Bearings are designed for "Combined Loading" meaning they will take both radial and axial loads at the same time. Skate bearings for Rollerblades, and skateboards are 8 mm "Deep Groove" bearings, and they are the least expensive bearings on the market.

According to SKF, 8mm Deep Groove ball bearings will support an axial load that is 25% of their rated static radial load. So for 8x22x7 deep groove bearings with a rated static radial load of 1.37 kN (308 lbs), their suggested maximum axial load is 343 N (77 lbs). So you can just use two skate bearings with a small pre-load on each lead screw, and you've just saved yourself a couple hundred bucks.

They may wear out before angular contact ball bearings would, but at $.50 each who cares?

[ger21]

That's what I did. You'll find that as you preload them you'll reach a point where they don't roll smoothly. I just back it off a little and they're good, with no play.

[NC Cams]

Having formerly done a lot of machine tool bearing enginering work, I will concur that deep groove balls can absorb an axial thrust. BUT that is not how they are designed to absorb load AND STILL MEET RATED FATIGUE LIFE.

IF you axially preload a ball to 1% to 2% of the rated load, you will have nice smooth motion and little if any axial clearance. HOWEVER, go above that and things get different FAST. Simply put, the sides of the raceways are NOT finished in a deep groove ball as they are in an a/c bearing. Thus, the sides aren't optimized for continuous service. Besides, ball and raceway geometry are not matched well for long term and/or high force axial loads.

Example: The ball screw bearings in a typical Bridgeport mill are basically preloaded 6204's (thus turning them into a psuedo a/c bearing). Our measurements show that they have about 100lb or so preload.

More Rigid?? yes
Rigid enough??? no.
As rigid as an A/C??? not even close.

If you lean on the table with about 150-200 lb force with the oem bearing, you can see table deflection approacing 0.001". This compliance really becomes apparent when you do a direction change when milling around a profile.

We experimented and shimmed them to get rid of the clearance/slop. When we got rid of the clearance, you could hardly turn the handle and when it did trun, it felt like turning a ground pepper mill. Deflection was stlll about 0.0003".

We replaced them with true 60 deg, ball screw bearing (same dimension) meeting ABEC 7 having a preload of 480 lbs.

Result: no measureable deflection when you lean on the table. Less than 0.0001" flat at direction change. And, more importantly, we found that the torque to turn the handle was LOWER than with the OEM "speudo a/c's" in either stock or shimmed form.

Yes, deep groove ball bearings may statically support up to 24% of their rated load BUT that does NOT mean that you can/will see 100% of the rated fatigue life if you do so. This 24% factor accounts for occasional thrust load that MAY occur. It does not mean that you can/should run them there at that level as a matter of course. One really has to be careful when reading handbooks. There are usually well placed caveats contained therein to keep people from misapplying the technology.

In that regard, I'd be inclined to bet that the SKF catalog also advises to contact the engineering dept when the intent is to load a bearing in a fashion other than its prime intent - IE deep groove balls with high axial thrust.

Reason: Axial thrust in a ball bearing is something that has to be carefully evalutated to assure proper performance and fatigue life. Yes, you can preload them axially and expect to see more accurate, less sloppy operation.

You can NOT expect a deep groove ball bearing to provide the same level of performance and accuracy and fatigue life as that of a true A/C bearing when operated under high levels of axial thrust/preload.

Yes, skate bearings are cheap and absorb radial and axial thrust. They are, in effect "throw aways". Did you ever consider that tather than throw the bearings out, where might a bearing supplier find a market to unload perhaps non conforming bearings, overstocked bearings and still recover some of their production costs???

[ger21]

I think the original post was directed more to the DIY router crowd, with smaller machines built on very small budgets. As I mentioned, it doesn't take a whole lot of preload before they no longer work smoothly. But, for light loads, they work great.

[Mcgyver]

musicmkr, I agree for a lite duty diy router type project the cheapies may work, but the angular contact bearings don't have to cost that much. I suspect you are looking at matched pairs, ie ABEC7's or such. you can buy regular (not matched, sorry don't know the class) angular contacts for about $15 each. Its a mystery to me why some machines use the abec7's for leadscrews (for a spindle yes, i understand), ie the regular angular contact bearing's accuracy exceeds that of even a ground screw - there is likely a good reason, it just has escaped me so far.

[NC Cams]

Here is the runout specs for a 608 bearing being discussed

ABEC 1 (ISO P0) = 10 microns.
ABEC 3 (ISO P6) = 5 microns
ABEC 5 (ISO P5) = 4 microns
ABEC 7 (ISO P4) = 2.5 microns
ABEC 9 (ISO P2) = 1.5 microns

When you want to run this bearing at high speeds (as I'm sure routers run) and obtain low tool runout/chatter potential, it is easy to see how better quality bearings will help you do just that. It should be noted that other accuracies also improve such as side face runout with respect to raceway and so on and and so forth.

You can literally FEEL and HEAR the difference between a 6206P4 and 6207P4 bearing as compared to and a regular 6206/207 combo of electric motor quality bearings in the same application on an A-B comparison. It is amazing even at speeds as low as 1700 rpm or so.

Example: I was lucky enough to obtain some surplus 207P4's from an old, old hard disk drive and used them in a grinding spindle and some electric motors and saw/heard first hand the difference. The difference in grinding finish on the ABEC7 equipped grinding spindle was HUGE, especially since the spindle was running at about 4000 rpm.

Yes, a slight (1%-2% as stated in my earlier post) axial preload will greatly improve running accuracy of a deep groove ball bearing. The wave washers used in many electric motors do just this by keeping the balls lightly loaded so they don't roll into and out of a load zone - the balls tend to chuckle/skid when they do this. My point of exception was the blanket implication that a 24% load factor was OK of any 608 bearing (7x22x7).

This general axial load factor as quoted from an SKF catalog is used for shock or occasional thrust loading consideration - not for a 100% duty cycle. Bearings bearings will TOLERATE this type of load but can't/won't live as long under such duress on a continuing basis.

Whenever, when you start to combine axial and radial loads, you need to compute the resultant load vector from combined Fr and Fa and figure life on that combined factor. Thus, if you take the 77 Fa and combine it with the 308 Fr, you end up with a resultant 317.5 or instant overload (take square root of (77*77+308*308).

Point being made is this: one really needs to take ALL the load factors (direction and magnitude) into consideration when doing bearing application work. Once you know the loads and directions of same, run the numbers and see if you'r original assumptions are still correct. IF in doubt, seek the help of a bearing applications engineer. NSK, NTN, SKF, et al have them and they are there to help.....

I've seen many an engineer use axial load factors impropely and get burned with poor life and/or poor performance. We learned this lesson when someone used some generic bearings in a Dumore high speed electric die grinder - they failed to consider the speeds and loads properly.

I'd be inclined to think a high speed router would react in a similar fashion which is why I expressed that caveat as I did in my earlier post - sadly without going into excruciating detail.

[Mcgyver]

NC, my comments on using a less-than-abec7 were directed towards using the bearings on a feedscrew - completely agree with you re a spindle, router, grinder, mill or otherwise. You obviously know a lot about bearings - thanks for the info.