Multiple Bearings in Spindle

[Iain DownsParticipant @iaindowns78295]

The spindle mainly – it's quite complicated!

Iain

[Andy_GParticipant @andy_g]

Posted by Iain Downs on 26/07/2023 16:53:13:

I'm particularly interested in how you plan to machine the spindle, Steve. It looks to me like a challenging level of precision for a mini-lathe. Certainly beyond what I think I could achieve.

I made an ER11 spindle of a very similar design on my mini lathe. I was very pleased with the way that it came out – I couldn’t detect any runout in the spindle itself with an 0.002mm indicator, and about 4µm TIR on a tool in a decent quality collet.

This was the sequence I used:

With the stock held in a 3 jaw chuck, I machined the tail end as far as I could and cut the threads for the nose bearing clamp nuts (on my design), but only roughed the bearing journals.

I then flipped the part around and held the tail end (centred) in the 4 jaw chuck. The other end was supported with a fixed steady while it was centre drilled so I could use a live centre for support while machining the outside of the collet chuck and cutting the collet nut threads.

I used the fixed steady again, without disturbing the part, to bore the collet recess and cut the internal taper for the collet. That was the bulk of the machining completed.

I then chucked a short piece of brass and cut it to match the collet taper (while the compound slide was set to the correct ER taper angle).

I could then mount the spindle between centres, using the collet taper to support the ‘nose’ end of the spindle on the brass live centre with a dead centre in the tailstock. This setup was used to machine the bearing abutments and journals.

The journals were left 0.01mm oversize and then lapped to final size. I was very pleased that I managed to get them both pretty much spot on – just a few microns above nominal.

There's a bit more description of the whole thing on the page I linked to in a previous post.

I'm sure that there are other ways of doing it, but that's what worked for me.

[Kiwi BlokeParticipant @kiwibloke62605]

Posted by Iain Downs on 26/07/2023 20:36:40:

The spindle mainly – it's quite complicated!

Iain

Yes. More simplification required!

Back in the '80s, when I was messing around with cars, I had a Ford Escort gearbox (2000) in pieces, next to an exploded diagram of a Triumph (Spitfire, IIRC) 'box. I was impressed how Ford had managed to simplify the whole thing, with far fewer components. It must have been easier, quicker and cheaper to manufacture, and far easier to strip and re-build than the Triumph, which was done the old-fashioned way. In addition, the Ford 'boxes were strong, reliable, and had a renowned, slick shift action. They were very popular (pretty well universal) among amateur motor racers, and constructors, such as the UK "Clubmans' Formula', which was getting close to Formula 3 lap times. I'm a long-time fan of simplicity!

OK, what can be done, to satisfy Ian? Well, the machining of the housing, to get the bearing housings aligned, is not a trivial job. If you're going to machine a smaller ID at the tail end, you might as well machine an ID the same as at t'other end – it isn't more difficult. Then you can simplify bearing purchasing too…

This allows the bulk of the shaft to be the same diameter, and the spacer to be plain-ended. Tail-end details will, of course, need to be adjusted also.

I'm finding this journey towards simplicity interesting. Thanks for allowing (encouraging?) me to continue disrupting…

[Kiwi BlokeParticipant @kiwibloke62605]

Addendum. Last suggestion means the tail bearing has to be fitted after the spindle has been fitted to the housing. This isn't a problem. The outer track should be a light push fit in the housing, and you have the screwed end of the shaft to take a nut and tube bearing pusher. It'll go together easily. The tail bearing bore in the housing must extend deeper along the bore, so the outer track of the bearing can float axially.

[Andy_GParticipant @andy_g]

Posted by Kiwi Bloke on 27/07/2023 02:29:27:

…the machining of the housing, to get the bearing housings aligned, is not a trivial job…

This is a powerful argument (in my mind, at least) for making the tail bearing OD smaller as it allows all of the bearing locations in the housing to be machined in the same operation, giving the best concentricity possible with given equipment. (They don't need to be *especially* accurate otherwise – the tail bearing bore is probably the most critical: to allow float without slop.)

I'm not sure what material Steve is thinking of for the housing, but in the design with a spacer between the inner races only, I would suggest that the housing be made of steel, rather than aluminium, in order to avoid variations in bearing preload as the spindle warms up – I tried something similar with an aluminium housing and the temperature effect was remarkably noticeable.

This will be the Rolls Royce of spindles when it's done!

[Kiwi BlokeParticipant @kiwibloke62605]

Andy_G – Yes, I agree with everything you've said, except I'm less confident about the last para. It might turn out like a camel – designed by committee. Let's hope not: it should be good. it might, perhaps, become a sort-of ME standard spindle, in a similar way to Prof. Chaddock's Quorn spindle.

Regarding machining the housing in one set-up: I suppose it depends on how confident one can be, working at the bottom of a deep hole. Boring and gauging the 'far end' of the housing, and how stiff the boring bar would be, given that it would neeed to be longer than the housing, have their own problems. Boring each end, supported on a good fixed steady, is the alternative, for both of our tail-end suggestions. Then, the three same-sized bearings job is easier. Even the single set-up idea would probably need a steady at the tailstock end.

I would hope that temperature variations don't affect pre-load significantly, with an alloy housing, given the bearing pair spacing is so short, but agree steel is better. More reliable, running in a steady, too.

Keep the ideas flowing – it staves off Alzheimer!

[Steve CrowParticipant @stevecrow46066]

Here is version 8.

I've gone for a single nut at the rear. I know I could make the whole assembly in one piece but I lke the idea of a "pulley carrier" so I can swap for different sizes.

Not immediately apparent but I've put flats on the nose just behind the nut for collet tightening. This increase the front overhang but only by 2.5mm.

I want to go for different size bearing as I think I'll stand a better chance of getting the diameters optimal on two separate shorter journals rather than one long one.

When it comes to the spacer, I'm going to retain the stepped ends. Again, I stand a better chance of getting a nice snug fit if I've only got the ends to worry about. It will only take a few minutes extra work to bore out so It's hardly complicating things.

Same with the tail end of the spacer. I'l stick with the 8mm step as that is the SKF recommended abutment. Again, a few minutes extra work. I don't want to over simplify for the sake of it!

I'll describe how I plan to machine the parts later.

Steve

[Steve CrowParticipant @stevecrow46066]

How I plan to make the spindle-

Cut a piece of 10mm EN8 to slightly over total length. Face and centre drill one end.

Hold the last 10mm of the unfaced end in centred 4-jaw with live centre at the other end.

Turn the plain shank at tail end of the spindle. I will make it a bit longer than the drawing as it has to be held in a collet later.

Turn the 7mm pulley section to 6.98 then the bearing journal to 7.01. Continue along the spindle, one diameter at a time (leaving the journals 0.01 oversize), until the 9.5mm section is done.

The idea is to leave as much meat and stiffness in the spindle as possible while turning. A fixed steady might help but I think I'll get away with it.

When the turning is done, single point the thread and do any chamfering. Then lap journals to suit bearings.

Remove from chuck, reverse and hold 9.5mm diameter in a collet. Then drill, bore, taper bore, thread etc. Basically complete the ER8 chuck.

Before the next stage, you will have needed to make the spacers and rear lock nut assembly.

Assemble spindle, bearings, spacers nuts etc. and nip everything up.

Hold the tail end plain shank in a good collet or well centred 4-jaw. Support front bearings in a fixed steady and adjust so the 9.5mm diameter is as true as you can get.

Move clock to the collet taper bore and check run-out. Its only going to be as good as your 9.5mm collet during the last process.

If its good enough, your done! If not, take a light skim along the taper until it is.

This last idea was adapted from the Swindles Spindles book that started the thread.

I'll describe the housing machining later.

Steve

[Steve CrowParticipant @stevecrow46066]

I'm not sure that given the ER 8 collets and the small cutter/drill you are expecting to use whether you need two bearings at the front.

I've just upgrade a sensitive vertical drill by using a bought spindle with ER11 collets. The spindle is a straight 12mm diameter X 150 mm long and in my case runs in oilite bushes. I don't know if similar spindle are available for ER8 but it would save a lot of work.

Hi Clive, I do realise that this is probably grossly over-engineered for my purposes, but I wanted a design that could be replicated and scaled easily. Also I'm quite fond of a bit of over-engineering!

My very first thought when on embarking on this was to use a bought ER8 spindle as you said. They are available in 8mm shank at just the right length (65mm) but I wasn't sure about turning down and thread cutting as the shanks are hardened. Not to mention the satisfacion of making it myself.

My current project is upgrading a tiny sensitive drill. It runs in a 1/4" spindle which I'm remaking with an ER8 chuck. I would have used a bought spindle for this but they are not availble in imperial sizes.

This did run in steel bushes which I've replaced with Oilite.

Steve

[duncan webster 1Participant @duncanwebster1]

The 9.5 diameter is very short to grip on a collet, and the collet might not be dead true either. You could copy GHT's method, fit one of the nose end bearings, support it in a fixed steady and grip the pulley end in either a collet or a 4 jaw. The collet seat must then be coaxial with the spindle.

[Clive SteerParticipant @clivesteer55943]

Hi Steve.

My "sensitive drill" was actually designed for tapping but I can find no info regarding the manufacturer. I'll post some pictures to see if anyone can identify it.

I'm currently putting together the BLDC motor system to drive it.

In my case I chose the ER11 spindle with 12mm x 150 mm shaft as it was an almost exact match for the original. I say almost because the original spindle was strangely 11.86mm diameter so hence the need to rebush. Most of the other parts on the "drill" appear to be metric so I believe it may be Swiss or German. As a tapping device it would not need to be run at high RPM so plain bushes would be fine.

CS

[Kiwi BlokeParticipant @kiwibloke62605]

I'm naturally lazy, and my scrap bin has a nasty self-filling habit. so simplicity is attractive.

The spindle and spacer tube are the easy (easier) bits, particularly if simplified further, as I've suggested. I would do as much as possible on the spindle between centres, then rough out the collet taper, and follow Duncan Webster's advice to finish bore it, supporting the shaft end either in its bearing, or directly, by a steady. Don't rely on collets (especially cheapo ones!) for concentricity, even if there's an apparently long-enough surface to grip. And remember that, whilst you can 'dial in' something in a 4-jaw chuck, to run true where the indicator's probe bears, things might be different at some distance from the chuck. You can't rely on the jaws to hold exactly parallel to the lathe's axis (they should, of course, but don't make assumptions…).

It's the housing that's to my mind the most challenging. If it's to be a quill, you presumably need the bearing seats to be concentric with its OD. You might consider making a mandrel, so that the housing OD can be finish-turned on the mandrel, between centres. I'd try to avoid doing anything tricky down at the bottom of a bore, so my previous comments stand.

[Michael GilliganParticipant @michaelgilligan61133]

Clearly impractical for Steve’s project, but I would just mention how Gepy does the concentricity thing:

**LINK** : http://anglo-swiss-tools.co.uk/gepy-centres-spindles/

MichaelG.

[Ian PParticipant @ianp]

I would say the Kiwi bloke has pretty much summed up the important bits to consider when making the spindle parts.

If I were to make one to the highest precision I could, I would machine the spindle between centres to an almost completed state except for the female cone. Leaving the ER end solid allows machining all the external surfaces, threading and bearing registers being lapped whilst the spindle can be taken can be taken on and off the lathe to check fits etc.

The conical bore can then be rough machined just using the three jaw chuck. I would finish grind the ER taper later with the spindle running in its own bearings.

I would avoid making a spacer between the (front) bearings (version 8) just because if we are talking high precision, its just another item that has to have absolutely parallel faces to avoid tilting a bearing.

The tubular spacer in version 8 has machined features at both ends which are not really needed. The outer annulus area of a flat face at the tail end will be in fresh air and shortening the shaft (front) bearing register will allow the spacer flat front face to sit against the inner race.

Maybe I have missed it earlier in this thread but I am not sure what type of ballraces are being used. As the over length of the whole assembly is quite short and as its intended use should not require it to take excessive side loads then I would just have front and back only as that would eliminate the need for the thread and but arrangement in the outer body. A thin close fitting (to the shaft) 'washer' could be pressed into the front bore to keep swarf out freeing up a bit more axial length for flats etc.

Ian P

[Michael GilliganParticipant @michaelgilligan61133]

Posted by Andy_G on 27/07/2023 10:34:49:

[…]

I'm not sure what material Steve is thinking of for the housing, but in the design with a spacer between the inner races only, I would suggest that the housing be made of steel, rather than aluminium, in order to avoid variations in bearing preload as the spindle warms up – I tried something similar with an aluminium housing and the temperature effect was remarkably noticeable.

[…]

.

I offer one of those near-random thoughts that pop into the mind unexpectedly:

If the housing was made from Machinable Technical Ceramic, the effect of warming should be to reduce the preload … making the device “fail-safe”

**LINK**

https://www.final-materials.com/gb/226-machinable-technical-ceramics

MichaelG.

[Andy_GParticipant @andy_g]

Posted by Michael Gilligan on 28/07/2023 13:42:18:.

If the housing was made from Machinable Technical Ceramic, the effect of warming should be to reduce the preload … making the device “fail-safe”

The problem I had with the aluminium housing was that the preload reduced / disappeared when it got warm. If the preload was adjusted to 'take the slack out' in this condition, then the spindle almost seized when it cooled down.

This was with a solid spacer between the inner races (a circlip for various reasons) and an adjustable nut to preload the outer races against each other:

Unless I'm missing something, the same arrangement with a [low expansion] ceramic housing would do the opposite: i.e. preload would tighten up as the temperature increases.

Incidentally, something to consider is that matched pairs of angular contact bearings suitable for face to face mounting are available, fairly cheaply, as spare parts for Chinese made CNC router spindles. Face to face mounting takes the expansion of the housing out of the equation. The quality of the ones that I have bought has been pretty good.

Edited By Andy_G on 28/07/2023 15:00:04

[Michael GilliganParticipant @michaelgilligan61133]

You are probably right, Andy

As I said it was just an unexpected thought

… I should have ‘thunk it through’ before posting.

MichaelG.

[Iain DownsParticipant @iaindowns78295]

Just wanted to say thankyou for the useful comments.

My own use case is to put the spindle cartridge in the spindle of my mill and drive the (high speed) spindle from a motor hanging off the cartridge.

The drawing below my give an idea.

Left is the spindle (cheap Chinese ER11 'extension rod&#39 which runs in a 'cartridge' (some imagination is required to see the bottom part as an ER11 collet and above that a timing pulley). Attached to the bottom of the cartridge is a bracket that holds the motor (in my case a 200W Chinese thing). Again imagine the timing belt between the two,

The spindle is supported on bearings at the top and bottom of the cartridge, loctited in. I've also made a version with preload applied by nut with a thread on the top of the HS spindle (cuts fine with single point threading) though that performed somewhat worse than the one which was just glued.

The specific thing I wanted to achieve is to be able to mount the high speed spindle in the mill and have it be correctly registered. I could, for example rough something out with a 10mm end mill and tidy up with a 1mm mill.

This kind of works ( drilling with sub 1mm drills is entirely possible now), but the run out is bigger than I'd like. I'm not sure if all the complications of Steve's spindle is really required for what I want. In particular, going mad for rigidity may make little sense if I have a bit sticking out with a drive on it!

I'd be interested in any suggestions or comments before I build my next prototype.

Iain

[Andy_GParticipant @andy_g]

Posted by Iain Downs on 28/07/2023 16:46:38:

The specific thing I wanted to achieve is to be able to mount the high speed spindle in the mill and have it be correctly registered. I could, for example rough something out with a 10mm end mill and tidy up with a 1mm mill.

This kind of works ( drilling with sub 1mm drills is entirely possible now), but the run out is bigger than I'd like. I'm not sure if all the complications of Steve's spindle is really required for what I want. In particular, going mad for rigidity may make little sense if I have a bit sticking out with a drive on it

Have you checked the runout of the taper on the ER11 collet extension with respect to its own shaft? I have had a couple of shockingly bad ones and one very good one (roughly similar prices). Similarly, 'cheap' ER11 collets vary widely in their ability to run true.

Could you put the drive pulley on the rear of the bearing arrangement so that there was less overhang at the front bearing? IMHO, it's well worth trying to minimise tool overhang, to minimise the effect of spindle runout / lack of straightness. The smaller the tool diameter, the better the runout needs to be.

This is the arrangement for a toolpost spindle I made using the good ER11 collet extension. It is driven from the rear by a timing belt pulley:

The motor is mounted piggy-back:

I wouldn't use the double row A/C bearing again (it still needs to be preloaded to remove slop – I didn't realise until after I'd made it). If I were to make one using an ER collet extension holder again, I'd probably use a similar configuration, but with a single bearing at each end (maybe A/C or maybe plain ball bearing, depending on likely loads & budget at the time). I think it would be a compromise compared to machining a spindle to one of the designs above, though.

(Sorry, I didn't intend to monopolise this thread – it's just a topic that I have given quite a bit of thought to over the last few years!)

Edited By Andy_G on 28/07/2023 18:36:00

[Steve CrowParticipant @stevecrow46066]

I've only just read the latest postings as I've been in the workshop all day. Those boring bar holders don't make themselves! And I've got a feeling I might need them.

A word about collets and workholding regarding the spindle machining.

A couple of comments mentioned concentricity of collets. I've recently made an ER16 chuck for my headstock. Machined in place, it has no measurable run out. I bought a set of ultra-high precision collets (5 micron) and they are excellent, on an other level to any I've had before. Checking them with a test bar of nominal diameter (so there is hardly any compression) I get a run out of 2 to 3 um and I can half that by careful tightening (final 2 nips at 180 to each other) so I have confidence in my collets.

Saying that, Duncan Webster was right, that 9.5mm diameter is far too short (4mm) to be gripped accurately. ER collets should grip on their full length for best accuracy. That's 27mm for a ER16.

I could hold on the 8mm journal length instead which is 16mm long. As its being held in a 8mm collet, all should be good. After all, I'll be checking and skimming the taper in the final stage, when held true by the front bearings.

When I do the final stage, I'll hold the tail end in a watchmakers collet which grips on a shorter length.

The only time I'll be using a 4-jaw is the first stage, supported by a centre. As all diameters are turned in the same set up, everything should be concentric. I could even do it in the 3-jaw with oversize stock and turn down to 10mm for the collet when I reverse it.

I will get round to describing how I plan the make the housing eventually but it's beer o'clock.

Steve

[Iain DownsParticipant @iaindowns78295]

Andy – the runouts do seem to vary quite a bit, though I found the worst part is generally the collet nut (some of which are unusable. Some are quite good and I have collets which claim to be decent.

the point of my design is to put it in a collet in a (locked) mill spindle -not much chance of driving that from behind!

Iain

[Andy_GParticipant @andy_g]

Posted by Iain Downs on 28/07/2023 19:39:59:

the point of my design is to put it in a collet in a (locked) mill spindle -not much chance of driving that from behind!

Sorry, I misunderstood – I read it that you had a separate high speed spindle, offset from the mill spindle, with some sort of mounting bracket held in the mill collet – my apologies.

[Steve CrowParticipant @stevecrow46066]

I do want to make a longer quill version of this. That will have to have the housing trued on a mandrel as mentioned by Kiwi Bloke. However, I'm going to concentrate on the short version first to prove the concept, so to speak.

The housing will be steel, probably EN1A. Much as I would like to bore everything from the front, I have zero confidence in getting the rear bearing diameter right. And as for cutting the O-ring groove? Forget about it.

I'm going to cut and face to size then centre in a 4-jaw. A few minutes with a little nylon hammer on the tail end and careful tightening and in my experience you can get both ends near true. The whole housing is less than 48mm long and a third of that is in the chuck jaws so I'm not sure a steady is needed.

Drill and bore to suit rear bearing all the way through. Cut O-ring groove.

Reverse in chuck and clock both ends but this time on the chuck end and the bore. Bore for clearance, nose bearings and nose inner nut. Have said nose inner nut ready (that you made earlier) for testing the fit and cut internal threads.

I don't know the first thing about O-rings by the way. If anyone can enlighten me to what I need for this application?

Steve

[Andy_GParticipant @andy_g]

Posted by Steve Crow on 29/07/2023 18:22:07:

I don't know the first thing about O-rings by the way. If anyone can enlighten me to what I need for this application?

The O-ring isn't particularly critical – find one with an ID that matches the bearing OD (within a few %) and has a section diameter that is less than ~25% of the bearing width.

Cut the groove in the housing to ~90% of the O-ring section deep by somewhere between ~125% and ~150% of the O-ring section wide (not critical). The groove is roughly centered on the bearing location.

(The groove could be machined in a different setup to machining the bore because any sight lack of concentricity doesn't matter.)

[Steve CrowParticipant @stevecrow46066]

Posted by Andy_G on 29/07/2023 21:14:41:

Posted by Steve Crow on 29/07/2023 18:22:07:

I don't know the first thing about O-rings by the way. If anyone can enlighten me to what I need for this application?

The O-ring isn't particularly critical – find one with an ID that matches the bearing OD (within a few %) and has a section diameter that is less than ~25% of the bearing width.

Cut the groove in the housing to ~90% of the O-ring section deep by somewhere between ~125% and ~150% of the O-ring section wide (not critical). The groove is roughly centered on the bearing location.

(The groove could be machined in a different setup to machining the bore because any sight lack of concentricity doesn't matter.)

 

Thanks for that, In that case, I've seen O-rings that will fit.

I'd rather cut the groove when boring rather than go through the whole faff of setting up again though.

Edited By Steve Crow on 29/07/2023 23:29:37

[Author]

Posts