Need some 4 jaw opinions

[ozzie34231]

I'm working on a sharft that I turned between centers. All the critical places on the shaft dead on size, as close as I can measure, (.0001").
I now need to chuck this piece in a 4 jaw as close to perfect as I can get it.

Within a few minutes I can get to .001" TIR, and in another few I can get to about .0005" TIR, but no matter how long I fiddle, I don't seem to be able to get it better than that. I have a large collection of dial indicators so I've tried several but they all give the same result. The result is the same turning the chuck by hand or power.

My equipment is cheap, (Shoptask and stock 4jaw). At first I thought there might be that much play in the spindle bearings, but after each adjustment the results remain constant,i.e. if the high spot is at #3 jaw it stays there until I try again.

So my question is; can I get it better or is this the best I should expect? I can't imagine there are any tricks to centering, but are there?

Thanks for all ideas,
Jerry

[Geof]

.0005 TIR means the part is .00025 off-center. This is pretty close for a small inexpensive machine. I think you are as close to perfect as you can get it.

[ctate2000]

Your machine probably will not allow you get any closer than .0005. The part can be out of round which would cause the same problem. You will have trouble on any lathe in the tenth world. Most lathes are not designed to work at that level. .0005 reading on the indicator means you are .0005 TIR and .0005 off center not .00025.

[ozzie34231]

Well I don't agree with that. If you move the piece .00025 in the right direction you'll get 0 TIR.
Anyway thanks for the advise from both of you.
Yesterday I thought I'd try one more time to get closer. The high side was at Jaw #4 and I gave it one more twist. Crunch! 30 lbs of scrap chuck; it doesn't look like it's repairable.

The work piece is a spindle for a grinder, about 9" long with diameters from .625" to 1-1/16". I put the small end in a 5C collet, the other in the tailstock center and tightened the collet, and pulled the center away; .001" TIR at the collet and about .002 at the hanging end. I then put the hanging end in my steady rest and got virtually no movement on the dial Gage.

I've drilled it that way and today I'll bore and taper it. I'll be interested in seeing how it eventually runs. It will be fine for dressable wheels but I also intend to run diamond wheels and it's those I'm worried about.
If it's not right, I'll try again, it's only a hobby.

Thanks again,
Jerry

[ozzie34231]

Add:
In Chadwicks book on the Quorn grinder he says a spindle should be bored running in it's bearings. This would be a setup as I've described with one end in a 4 jaw and the other in its bearings and housing, mounted in a steady rest.
If I must remake the spindle I'll do it that way. more trouble but sounds deadly accurate.

Jerry

[handlewanker]

Hi Ozzie, I've never came across anyone who can measure to .ooo1", and that's in 40 years of hacking metal.
I once had a bet on with a group of gauge room inspectors and none of them could get the same reading on a work piece due to individual feel for mic's or tolerance on in-house calibration.
As I recall the Quorn has the wheel spindle running in angular contact bearings, and that is where the accuracy of running lies.
When you mount the wheel you dress it true each time anyway, even if they are individually mounted on their own holders.
Incidently, when you get to within .001" of concentricity working with a 4 jaw, it does not pay to crush the work to get it to move, just "tap" the opposite jaw with a copper hammer and it will move, trust me, btdt.
I currently use a 150mm diam Chinese made 3 jaw chuck, which is 10 years old, and it runs to within .02mm.
Reason is I don't force it to hold work to the "crunch stage".
I rely on it to run true and for rough work or bar stock I use an old 3 jaw or put up the 4 jaw.
Ian.

[ozzie34231]

Yes I was foolish to try to over tighten and as for the measuring, let me be happy thinking I can measure that well, at least I get the result I'm after.

Surprisingly in the 1990 reprint of the Quorn book, Prof Chaddock, doesn't talk much about the bearings; no talk of angular contact, or ABEC numbers. His quill uses only two bearings, the back bearing against a set of springs that would load them both. The same is true in an inexpensive book I picked up about building spindles.
There has been a rather thorough treatise here, in this group, on building a quill for milling purposes.

I'm using all I've read in all sources to cobble a quill for a machine much heavier than the Quorn, to be used by me for both tool sharpening and a little surface grinding.
The quill incorporates double angular bearings on the front end and a single bearing at the pulley end. I hope it works well, but as I said, it's only a hobby.

Thanks,
Jerry

P.S.
This might interest some oldtimers. My Dad taught me to use a micrometer a little over 50 years ago and the touch is very important. I spent a lot of time back then measuring standards to get the right feel, just 'cause I wanted to learn. At the time he was grinding dies to +/-0.00002" ; needless to say he wasn't measuring with a mic. He was doing the work in PA and sending the dies out of state. He was getting rejections though he thought his work was right on. His company sent him to sort out the problem. Turned out when he cleaned samples with carbon tetrachloride, they measured within the specs.; the preservative oil was the culprit.

[ctate2000]

.00025 off center is correct. Brain Fart.

[handlewanker]

Hi Ozzie, I'm currently, when I get a round tuit, remaking the quill for a Beradi vertical jig borer, which is almost the same as a mill/drill, but heavier proportions and more precision.
The design which I am using is as you suggest for your spindle, with two angular contact bearings at the business end, and one sealed radial bearing at the top end.
I'm only using one sealed bearing at the top end, to offset lube problems, and as the pulley drive has it's own double bering set-up with the splined shaft of the quill spindle going through it, there should be no side pull.
If you are going to use the spindle for a grinding operation whereby the spindle is driven from the end, then I would use two radial sealed bearings to offset the pull of the belt, which otherwise will pull the shaft and render the accuracy of the double front bearing set-up void.
On the subject of the Quorn, I assumed that angular contacts were used as they are designed for this type of requirement as opposed to radial bearings which are "pressed" into service. (didn't check in Mod Eng)
I know you can get away with this method, but who would save a few pennies this way when building a Quorn.
By the way, "just a hobby" is the most important thing a person would want to do for yourself, as opposed to "must do" for an employer, and so you tend to put your heart and a bit of your soul into it.
Ian.

[ozzie34231]

Hmmm. I hadn't thought about the side pull on the pulley end being enough to warp the spindle during operation. The pulley will be mounted about an inch away from the bearing on a 5/8" shaft, and since I've already bored the housing, changing the plan now would mean a new housing. I'll try this way, bearing in mind what you've suggested if there is a problem when it's done.

Thanks for the help,
Jerry

[NC Cams]

A new term to learn: repetitive runout which is not to be confused with non-repetitive runout.

You need to consider that there are tolerances in the bearing: inner ring raceway runout, outer ring raceway runout, inner ring roundness, outer ring roundness plus comparable roundness tolerance in each and every ball.

Add to this any mounting deficienecies such as housing or shaft runouts or out of round and you can really end up with some bizarre "off" conditions that will affect how the shaft shaft rotates or "orbit" when it turns

Due to the interaction of ALL these normal and customary bearing tolerances, although the shafts "rotate" just fine, the axis actually moves around in an "orbit" that is extablished by the ever changing "instantaneous axis" that is established by the raceway and balls and the tolerances contained thereing.

Depending on the ball count, the ball pitch diameter, whether the balls are constantly loaded or they unload and can skid, the orbit pattern will inevitably and eventually "repeat". The problem is the the axis is orbiting all over the place until the stuff realigns at the start position and the path can be traced over. Depending on the ball pitch diameter, ball count, tolearnces, and raceway diameters, this "orbit" will essentially repeat in 15 to 22 revolutions of the shaft.

THis is often referred to as non-repetitive runout because you don't know nor can you predict the orbit pattern of the axis.

Thus, the "runout" which is created by the normal and customary orbiting produced by the tolearances noted above is essentially unavoidable and unsolveable unless very special care is taken in the construction of the spindle and, more importantly the component selection thereof. THis is why spindles are ID ground in a fully assembled condition - to reduce/minimize the effect of cumulitive runout at the tool holding interface.

Now, if you add a belt load to the mix and any runout that a belt load variation may induce, your situation gets even MORE tenuous.

Now, if you make what is essentially a "perfect bearing" (ABEC 9 runouts, grade 3 balls, precision preload, hand selected and matched in all ways), you can create a bearing that will repetitively runout. Each and every rev, the axis will orbit in the same, relatively predictable orbit path. Yes, there still will/may be an orbit BUT it will tend to be much less or at least manageable.

Such a bearing would be very helpful in a precision, tool room lathe and/or a spindle for a HDD (Think about it: every time you put a piece of data on a HDD, you want to go back to that point and find the data bit - non-repetitive runout would kill you, REPETITIVE RUNOUT can be lived with/tolerated - think about it for a while, it makes sense).

When you work in tenths to try to hold runout, anything less than an ABEC 9 bearing with Class 3 balls in a preloaded bearing will result in an inability to "re-find" the prior axis of rotation if the part is taken out of and reinstalled into the spindle.

Your chances of finding the EXACT same axis of rotation that the part was initially cut on will usually be somewheres between slim and none (reinsert belt load comment here).

BTW: Handlewanker's comments about repetitively measuring to a tenth is true. We learned the same lesson sorting piston pins. NO two people could repeated remeasure the same pin to the same diametral reading using standardized tool room gages in an environmentally controlled room.

The chances of even an expert machinist with a god knows what grade spindle measuring runout to a tenth with general shop micrometers/dial indicators in a regular shop environment would involve more luck than anything else....

[handlewanker]

Hi NC, I take my hat off to you man, you must have really been round the traps to accumulate such a vast wealth of technical knowledge.
It's good to be able to pose a problem and get a lot of viable answers from people who've been there and actually done that.
As much as I dislike bronze bearings and all the problems with adjustments etc, yeah I know there are tons of highly precision machines that rely on the "steady effect" that a bronze bearing gives as opposed to the rattling good fit of the average ball and roller bearing, I still love a good ball or roller bearing set up.
When I was apprenticed in the late 50's we had a Churchill crankshaft grinder which did the crankshafts for the D8 Caterpiller engines, and to supplement it the company bought another German grinder in.
The spindle bearing ran in a hydraulically adjusted bronze bearing, and after a couple of seize ups due to the bearing overheating, it was found out (by reading the instruction manual in German) that the lube mix was 50:50 sae 30 engine oil and diesel fuel mix. After that, no problems.
As far as Ozzie's problem goes I don't think he would have to worry too much, at first, it is only when the radial type bearing slackens up a bit from the belt pull that the shaft could get a miniscule bit of a wobble, which would be better carried on two bearings.
On one high speed milling spindle design, in the FAG catalogue from some years back, the upper bearing is an angular contact with spring loading to keep it tight, so I suppose a similar design could be applied to Ozzie's single radial bearing to offset the belt pull.
I don't like the idea of putting any end loading on radial bearings as it just accelarates the wear, unless the loading is fairly light and under a constant spring preload.
Some of the Taiwanese drills that came out in the 80's had plain radial bearings in the quills and they soon gave trouble under load.
Ian.