Bearing design for a spindle depends on speed and forces. In my past life I drew designs for thousands of spindles, OK many of them were straight forward for vertical drilling operations but there were many in the five to ten thousand rpm range for more precision use and that is where things get interesting.

SKF used to publish a design manual, I suspect you can still see it online, where they discussed different bearing arrangements for many uses. A lot of the older machines tools used these designs, which came first is any bodies guess. Commercially bearing manufacturers would work with designers to get the simplest bearing design, no point in sticking five bearings on a spindle is two would do.

Following a known design of a spindle which fulfils you requirements is the simplest way but be careful that you follow all the parameters of the design. Using cheap bearings in a spindle which is required to run accurately at high speed will not work, or if it does not for long. Over lubricating can cause heat within the bearing and wreck everything , as will lack of proper lubrication.

I've now got Harprit Sandhu's book in front of me. I wish to correct and add to my comments above, to avoid doing the fellow a disservice, since my comments might appear to be wrongly rubbishing everything in the book.

His 'Basic Spindle' (Ch 3) is OK, if a matched back-to-back pair of bearings is used. Alternatively, with a shim between the inner races, preload can be adjusted. So that one's potentially OK, even with standard bearings.

His smaller MT2 spindle (Ch5) is not OK, because the inherent axial play in the nose bearing cannot be adjusted out.

Micro Spindle (Ch 6), 1.000 Diameter Spindle (Ch7) similar weakness, but worse – the spindle is only prevented from pulling out by friction or adhesive holding the outer race of the nose bearing. Same for Ch 11 spindle.

Ch 9 & 10's spindles rely on accurate dimensions of inner and outer race locations, and the inherent axial play in the bearings can't be adjusted, except by trial-and-error machining.

Ch 12's spindle is potentially OK, but looks difficult to adjust – which would be required every time a cutter is changed.

Ch 13's spindle looks fine.

If you're chasing good surface finishes, you need to be able to get rid of all possible sources of unwanted movement…

Edited By Kiwi Bloke on 21/07/2023 02:53:32

Minor diameter of M16.2×0.5 is 15.7, so that is a no, would need to be at least 16.5. Spacers between both inner and outer rings at the front, ground to suit the bearings will give some preload, otherwise nut is not locked in place. Use a snap ring below the upper bearing to give a larger thrust shoulder. Use M7x0.5 for rear thread, no point in odd size. As previous bore through for simple assembly. Make front pair bearing nut M8x0.5. It is quite normal for the bearing locknut threads to be same size as bearing bore. Machine to ensure concentricity of all bores. At 15k use light air oil lubrication, will also help o keep the swarf out.

A rather complex solution that can be simplified as Kiwi bloke suggests. Also, remove the spacer between the two nose bearings, clamp them together. Maybe a circlip rather than the M16 threaded holding the outer races?

The mounting of the third bearing is as it should be.

Another thought. Why not reduce the overhang from the nose of the spindle by removing what looks like a plain diameter with spanner flats(?). Then contrive something at the other end of the spindle to hold it while tightening the collet nut.

Its good to see a design that doesn't (grossly) over constrain the bearings as many do, that are published in the world of ME.,

The tail-end bearing's inner track is sandwiched between spacer tube and nut, so that's OK. The spindle is fully axially located at the nose end. The outer track of the tail-end bearing should be allowed to 'float' (a snug, sliding fit), to take care of possible thermal expansion. Andy_G's illustration of an O ring bearing on the outside of this track is good practice – it reduces the ability of the track to slowly walk around the bore of the housing.

Agreed that you can't fully assemble the spindle before assembly into the housing, but all the important bits can be in place, except pulley and nut, of course. I'd keep the nose bearing spacer, unless you're going to buy expensive matched pair bearings. Adjustability is good…

Talk to your bearing supplier about bearing seals. It's possible to get non-contact 'seals' that don't drag on the spindle. Conventional seals can add a surprising amount of drag. And oil is probably preferable to grease. (I think someone else said that, but I'm editing this now, and can't see previous posts…)

Having now re-read the whole thread properly, I'm conscious that I've essentially re-stated Duncan Webster's ideas, although perhaps with different emphasis. It wasn't my intention to steal his thunder. Just two (great?) minds thinkng alike (or suggesting standard solutions…).

It's heading towards being a really nifty, scalable spindle. Well worth your writing it up for MEW. It would raise the tone of the mag considerably…

{Edited: it's difficult to type with fists…)

Edited By Kiwi Bloke on 23/07/2023 11:55:06

Looking good! And pretty pictures too. (Wish I could bring myself to re-learn CAD. I had an early version of TurboCAD – it drove me nuts. Ever since, it's been back to pencil & paper…).

I think the most challenging aspect of manufacture is getting the spacer-to-spindle fit right, if one is going for a snug fit in the hope that it will add rigidity. Perhaps someone could tell us how snug the fit has to be to be effective, for this geometry. I'd imagine that significant hoop stress has to be induced in the outer member, to effectively pre-load the inner, so it would need to be a press fit, which rather screws up the whole idea. The idea is more effective when the spindle is really skinny, and the spacer can increase the overall diameter a lot. It's probably an unnecessary gilding of the lilly here, because the spindle is of relatively large diameter already, and the spacer can't easily be much larger. Please can someone advise?

So, perhaps the spacer can be freed from its obligation to add stiffness, tolerances can be relaxed and simplification is achieved..

More simplification. If the spindle's 7.5 mm diameter section between the bearings is made only a little larger than the ID of the tail end bearing (say 7.1 mm), the dark green thrust washer (?) twixt spacer and inner track of tail bearing can be deleted. Also, I'd delete the groove in the shaft behind the nose bearing pack. If needed (and it looks unnecessary from your drawing, a small internal chamfer could be added to the spacer tube to clear the transition between spindle diameters here

Don't forget the pulley, the nut being outboard of it. Keep us posted!

Edited By Kiwi Bloke on 23/07/2023 23:56:38

The old skf catalogue gave lots of info about abutment dimensions. I made the mistake of binning mine on the basis it would all be on the Web. You could make the 14.5 bore 14.1 to increase the abutment. I

The spacer tube only needs to be a good fit at the ends, so if you machine the spindle so the the 8 diameter is proud of the bearing to the left by a few mm then make the dark green bit part of the long spacer it makes life easier, most of the length of the bore can be clear of the spindle, just the right hand end has to fit the 8 dia. It doesn't want to rattle about, a nice slide fit at the ends.

You need a means of locking the right hand bearing adjusting nut. Unless you actually need axial thrust both ways you could put a wavy washer in and tighten the nut fully, or take other's suggestion and get a matched set of bearings. I still think you'd be wise to draw it out to scale, I don't think it is at the moment