Setting Pre-load on Tapered Roller Spindle Bearings

Hummm good question. What I would do..and its worth exactly as much as you paid for it..would be to hand snug it. then run it for 30 minutes at mid speed. Then put a dial test indicator on your carraige, and indicate the nose in the Z axis (parallel with the bedway)..then get somone to really push and pull on the spindle, and tighten or loosen until you have less than .001 to .0005 backlash. This should do it as you are checking it when its warm and the bearings are expanded.
Id be curious as to what the backlash will be after the lathe has cooled to ambient.
Gunner
That rifle hanging on the wall of the working-class flat or labourer's cottage is the symbol of democracy. It is our job to see that it stays there. - George Orwell

you can get bearing standards from the bearing manufacturers
when we rebuild equipment at the shop we measure the total clearance between the roller and race at the top of the bearing and use a bearing manufacturer supplied chart to determine how much to tighten , usually a percentage of total clearance

Even .0005" sounds like quite a bit. Naturally I have no idea how much there should be but it just seems like this much movement would cause chatter and stuff... Thoughts?
Regards,
Robin

Sounds like a good plan to me, but I'd adjust for zero clearance either at temp or cold, and stick with the tighter setting. Then run for another hour keeping a close watch on temperature. Prying on the spindle with a 2x4 should work as well as applying an axial force - unlike a ball bearing, there's a direct relationship between axial and radial play in a tapered roller bearing.
The spindle will be hotter than the head casting when things warm up. Whether the preload is higher hot or cold will depend on whether the bearings are mounted cups in and cones out, or vice versa.
I just went back and reread the OP. I first assumed that the lathe has roller bearings, but that's not clear to me now. Gunner's advice is probably better if we're talking a tapered plain bearing.
Ned Simmons

The CTE difference between the spindle and head casting is very small, the distance between the bearings relatively small, and the temperature change is relatively small. Set for ZERO clearance, run it for several minutes at high speed, check for excessive heating.

Most lathes are set up so the preload DECREASES when the spindle heats up. The spindle is going to get hotter than the headstock casting, and there is often quite a length between the two bearings.
Since you mention tapered roller bearings, they sound like they are at opposite ends of the spindle, is that correct? Is this a geared-head or belt-drive lathe? If it is a geared-head or otherwise an oil-filled headstock, it won't heat much at all.
Jon

This was posted in the Yahoo group on 9 x 20 lathes in the files section: (sorta long but worth reading)
Called Emco in Austria about 3am (my time one night) and got connected with an english speaking engineer who told me how to check tension on ALL gearhead Emcos of the era. The method (he said) they use in the factory; "With the machine in neutral, disengaged from lead or hex screw (if applicable), and either the 3 or 4 jaw chuck installed, spin the chuck by gripping one of the opened jaws on top of the chuck and spin strongly. One revolution of the chuck is perfect. Less is too tight or bad bearings and more than 1.5 revolution is too loose. I've used it ever since and it's an excellent yardstick that hasn't failed me yet. My new S11CD spins 1/2"-3/4" past one revolution and that old one would only spin about 3/4 of a turn until I re-adjusted it. Apparently a "little" too tight is a LOT too tight.
Mance
Due to the number of emails I've gotten last night and this morning I thought I'd clarify this procedure.
First, it's done on a cold machine rather than one thats up to operating temp. When Emco checks theirs they are not fully assembled or at a place where power can be introduced to the machine and so they do the pre-load cold.
When spinning the chuck I locate (or make) a mark on the chuck near one of the jaws or note the number of the jaw I'm using to spin the chuck. I rotate the chuck so that the jaw I'm going to use is at the 2 o'clock position when viewed from the tailstock end of the lathe. When spinning the chuck by hand you'll notice that this releases the jaw from the spinning finger at approx the 8 o'clock position, sending the chuck on around. What you want is that jaw to end up, on it's own inertial free-spin back at the 8 o'clock position or close. A little over-rotation is preferrable to being short of the mark. If you can't get it past 1/2-3/4 turn with lots of effort it's definately too tight and if it spins more than 1.5-2 revolutions you could probably use a a little snugging up.
In my original post I said give it a "strong" spin and should've said a comfortable, firm spin. As in one that you can easily duplicate several, or even many times without causing pain or discomfort on the spinning finger. Just a comfortably firm, easily repeatable spinning action that you could repeat all day long if you had to.
And while I originally got this advise from Emco on a lathe using a 4" chuck I've used it on lathes up to and including 6" chucks. I have no idea if due to inertial loads it's applicable to 3" chucks or 8" but I would assume so tho it would take a correspondingly lighter/heavier spin to achieve the desired result.
I think the most important thing to determine a safe, proper pre-load is that the chuck/spindle assembly isn't binding or dragging and that it's not completely free-spinning either. If it's too tight it's fairly obvious very quickly. If it's too loose you can probably measure a little end play at the chuck by pulling/pushing on it. For reference my chuck indicates .0000" end play cold, when pushing/pulling by hand against a test indicator.
An additional point of reference is if I spin my S11CD chuck as hard as I can with a rag beneath my finger on the spinning jaw to cushion it my chuck will spin right at 2.25 revolutions. But in normal comfortably firm checking it only spins a hair past one revolution. Hope this helps to define the ambiguity of terms like "comfortable or firm."
Sorry to all the people I kept up last night or caused to get up early this morning to go out, start spinning their chucks. Tho admittedly I'm a little amused by the visual.
Mance
********************************************************************
Your visual may be accurate. I was spinning my chuck this morning before anyone else awoke and before the dog got fed. This maximat 7 (the yellow one in photos) spins like a perpetual motion machine. I have much chatter on heavy cuts and have tried everything (gibs, cutter overhang, locked compound, etc.). The spindle preload was going to be next and your post gave me the incentive to go and try it. I didn't realize just how bad it was. FYI, I had measured end play and it bearly moved a 0.0005" indicator. Den
Mance just gave me a new lathe with his suggestions for testing preload of the spindle bearings and acceptable limits. My maximat 7, while giving a pretty good finish, could not take deep cuts without serious chatter of the entire lathe and base no matter what else I tried.
The spindle would give many revolutions after a quick spin by hand as Mance suggested. I tightened the preload until only about one revolution was produced. I was then able to take 0.150" depth of cut on 1" CRS, and as hard as I could feed by hand (brazed carbide tool, positive rake). The chips looked like an ad for Iscar.

Tapered roller bearings in general are more tolerant of zero clearance, or even true preload, conditions. I think the OP's approach of measuring spindle drag is pretty good, that's how the manufacturer does it.
Another way would be to do that same thing a bit more scientifically by measureing breakway torque on the spindle. Wrap a string around the chuck, and connect the end to a spring scale.
Then measure how much tangential force is required to force the chuck to spin. The more the preload, the larger the force. This will be a bit more repeatable than the "one turn spin" method.
Jim
================================================== please reply to: JRR(zero) at yktvmv (dot) vnet (dot) ibm (dot) com ==================================================

It has been years since I read about bearings, but as I remember preload means the amount of additional tightening of the bearing AFTER ALL SLOP HAS BEEN REMOVED. Without checking a book on bearings, I would remove all clearance, run it to get it warm, and then adjust to remove all clearance and tighten it a 1/16 th to 1/8 th of a turn more. Then run it some more and see if it gets hot.
Dan
I have always used when setting preload on tapered spindle Is this a true assumptuion? Thanks, Steve

The procedure from the factory for replacing the tapered bearings in one of my CNC lathe headstocks involves a preliminary tightening of the adjustment nuts until you get a certain amount of Z axis endplay with a certain amount of end force on the spindle.
Then run the spindle at a low rpm for an hour or so. Check temperature.
Adjust nut for less endplay. Run again at higher rpm while watching temp.
Repeat the above steps until you get a situation where there's only .0002" endplay with a certain end force on the spindle and the machine will run continuously at 4000rpm without overheating the spindle.
To follow their exact procedure will take at least a full day or more.

Technically this is correct. In common use however you will see folks talking about "preload" where they really mean 'get the clearance down below some certain level' even though there is still some positive number that represents the clearance. It's not zero, and it's not negative. But they still say 'preload' where that term really should be used to describe the case where the clearance number has gone below zero.
Jim
================================================== please reply to: JRR(zero) at yktvmv (dot) vnet (dot) ibm (dot) com ==================================================

As others have replied, at least in a rigid setup - massive cast headstock - you need to deal with the temperature expansion coefficient and make sure the preload isn't too high when cold or hot. Presumably cold for a typical lathe.
That said, I experimented with the spindle bearing preload on a Grizzly 12" gear head lathe I used to own. While the headstock was cast iron, it was just a lightweight square box, with fairly thin end walls that supported the taper roller bearings at each end of the spindle.
What I learned is that the headstock casting was so flexible, you could tighten the bearing far beyond where it took up the free play, and it just pulled in the outboard wall of the headstock, without increasing the drag on the spindle by much at all.
So I figure for this class of machinery, maybe there is a different approach - take up the preload entirely and let the casting function as a large spring, that maintains fairly contstant tension on the bearing with no free play over the full temperature range. If you Sheldon is like this, I have no idea. Perhaps not.
I'd like to know if this is a deliberate aspect of the design of a machine with a lightweight casting like this one.
Bob

As far as I can tell, the spindle on my 13" Sheldon has never been messed with. It has a surprisingly high breakaway torque. In fact, it's a PITA when dialing in something in the 4-jaw.
When I installed new bearings in my 11" Sheldon I stopped short of preloading them that much. For all practical purposes I have zero runout on the spindle no matter whether the machine is idle and cold or in use and warmed up.
IIRC, I tightened the 11" bearings until I could feel a drag on the spindle, then backed off just a smidgen. I don't know if it's the correct way to do it, or not, but it seems to be working well.
BTW, I've gotten into the habit of *always* checking the level in my Sheldons' oil cups before starting the machine each day. The castings are porous enough to allow the oil to slowly seep out.
Orrin
Orrin

I had noticed that one of the engineers at work has a book on bearings, so I borrowed it and read up on preload. I also googled a bit and found some stuff by Timken and another bearing manufacturer ( I think it was SKF ). Anyway all the data I have says that a slight preload ( tightening after all the clearance is removed ) will give the longest bearing life more than with zero clearance or some slop. It will also greatly increase the rigidity of the spindle.
Preload does two things. One on roller bearings is change the contact from a point to a line contact. Second it causes all the rollers to contact the cups, not just the ones on one side. Now when you put a radial load on the bearing, the load on the other side decreases while the load on the other side increases. So you have at least twice the spring constant of the case where you have just zero clearance.
But do not over do the preload. A little more preload causes the tempertures to rise and the life to drop rapidly.
Dan