PDA

View Full Version : High speed spindle bearings



Tom Curlee
10-05-2010, 06:48 PM
I'm currently building a small desk top mill/engraver/(fill in the blank) machine and plan on using a 6" long ER16 chuck as the spindle. Since I plan on using this unit for (among other things) milling and drilling printed circuit boards, I want a spindle speed of at least 20,000 rpm to 25,000 rpm. ER16 chucks can be had fairly reasonably, but the bearings for that speed can be quite pricey.

So, a few questions:

What are the trade offs in shank diameter, say .5" vs .75"? Thicker would be more rigid, but there would be more mass to turn and the bearing are more expensive for the larger ID.

For this speed, shank diameter, and tool size (no greater than .375"), is the annular ball type bearing the way to go or should I also look into tapered roller bearings? Considering the load, I'm thinking about using a single bearing at top and bottom, although a dual bearing at the bottom might be better.

I guess the biggest question I have is for suggestions for bearing make and part number, with price an important part of the decision. I'm well aware that high speed bearings cost much more than the bearings for the motor on your 1500 rpm drill press. I'm just having problems paying $100 (and up!) each for bearing.

So, for those of you that have made your own high speed spindles, what did you use that worked (and didn't work)? What are the design trade offs? Suggestion and comments greatly appreciated.

Thanks,

Tom

RussZHC
10-05-2010, 07:22 PM
Don't know how much help this is but I googled "die grinder bearings" as those approximate the range you suggested and several hits got one of those exploded parts diagrams so the quickest route maybe to get a replacement bearing from some "easy" source like that and get something like a bearing series number then go for exact size you need...

TOOLZNTHINGS
10-05-2010, 08:04 PM
Hello,

Have not used their product, but saw them at the CNC Workshops in Mich. this summer and would build around one of the spindles and vfd drives they sell.
See CNC Zone for more tips.

http://www.kelinginc.net/

Brian

wierdscience
10-05-2010, 08:27 PM
+1 for the Keling spindles,why re-invent the wheel.

TGTool
10-05-2010, 09:49 PM
If you do want to build a spindle angular contact bearings would give you accuracy and should be good for the rpm. You'll need to spring load them against each other to eliminate clearances and you can use available wave or spring washers for the preload. Inner and outer races at one end (usually the load end) will be locked in place and either the inner or outer race of the bearing at the other end (usually outer) will be allowed to float to account for thermal expansion which can't be avoided. 1/2" spindle diameter should be sufficient. If you expected really heavy loads you can double up the bearings at that end but getting all the dimensions and bearings right so they actually share the load is probably more than double the trouble of a single bearing. There are existing designs for spindles like that and some can be found in the files section of the Quorn discussion group at http://groups.yahoo.com/group/quorn_owners/

JoeLee
10-05-2010, 10:14 PM
Just about every grinder spindle I took apart for service had Fafnir bearings. Like 2mm or 3mm series. High speed precision bearings. You can look them up and get tech help from them also. Hope this helps.

JL...................

claudev
10-05-2010, 11:04 PM
Routers, various Dremel(?) type hand tools, and many small high speed DC type motors run at similar speeds while some dental drills run at significantly higher speeds. You might investigate replacement bearings for these type tools. You might also investigate some of the better quality air powered die grinders since some of these run at extremely high speeds and also seem to hold up well. Of course your application may require higher precision than any of these provide.

For extremely high speeds air or fluid bearings would seem to fill your requirements.

Tom Curlee
10-06-2010, 02:08 AM
Gentlemen:

I've looked at the Keling and similar spindles and have been tempted. They solve most of the problems up front: no alignment or adjustment, collet ready to go, variable speed, and especially the ease of mounting (no machining except for a clamping block). The top end speed is fine, but what I failed to mention in my original post is the low end speed. I'd like to be able to run the spindle down to 1000 to 1500 rpm, if possible, to run bigger end mills. At 100 SFM and a .25" end mill, I need around 1500 rpm. This wide range of speed is why I've been considering a belt driven spindle (if I can find a belt that works at that speed). I may well be asking too much for the stiffness (or lack of stiffness) that the machine can provide.

As far as using die grinder bearings, what is the normal runout (and end play)of this kind of bearing? The .010" range of PCB drills need -well- under .001" runout.

I'll look carefully at the Quorn Yahoo group.

Thanks for all comments.

Tom

macona
10-06-2010, 03:09 AM
The quorn uses magneto bearings which are a cheaper angular contact bearing.

The keling spindle will go down to that low of an rpm but I dont expect to have any useful torque. It would be better if you got a true sensorless flux vector VFD to run it. But I kind of doubt you will be able to get anything useful at that RPM. You can run aluminum pretty fast.

Another thing is if you have a high speed spindle you can use small bits to take more smaller cuts faster and it will even things out a bit vs using a bigger end mill. And since you will be running carbide 100sfm is way low.

die grinder, dremel, and other bearings are all just generic bearings. Tiny PCB drills will snap if you sneeze. I have a spindle out of an excellon router and they are belt driven.

Another option over the cheap chinese spindle are the ones by Precise, now fischer-precise. They are very, very, nice. ABEC9 bearings standard. I picked up my 45000 RPM one for a couple hundred off ebay. I had tried the die-grinder/dremel/ BYO Toolpost grinder and the difference in finish was night and day. But you got to watch out with these spindles. Some of the older ones had a two phase motor and were driven by a separate motor-generator control. Also a lot of the newer ones are 3 phase but I dont know the specs. The one I have uses a brushed universal motor and has a variable reluctance shaft sensor to provide feedback for closed loop speed control.

MuellerNick
10-06-2010, 03:13 AM
Building your own good spindle is not easy!
Do you have a cylindrical grinder? No? Forget it!
You can make one on the lathe, but the expensive bearings would be a waste of money, they won't last long.

Still wanting to make your own? Can you hold a few m tolerance? A few m runout (collet cone to bearing seat)? Harden and grind the inner spindle (with the collet)?
Intend to spend money for two spindle bearings, or three (two of them paired) or even 4 (all paired)?
The Quorn or Bonelle is a starting point for a spindle, but this construction is not suited for big axial loads, so no milling.

The belt is not the problem, my ID grinder spindle on the Myford runs with 40000 RPM with a flat belt. But the big wheel on the motor has 340 mm diameter. :D Makes 180 km/h belt speed. :eek:


Buy one, you won't make it better and cheaper.



Nick

Mcgyver
10-06-2010, 06:52 AM
Building your own good spindle is not easy!
Do you have a cylindrical grinder? No? Forget it!
You can make one on the lathe, but the expensive bearings would be a waste of money, they won't last long.


Why?



.

MuellerNick
10-06-2010, 08:01 AM
Why?

Because 20kRPM are too demanding with a wrong preload, oval bore, non-coaxial bearing seats, not considered thermal expansion, ...


Nick

JoeLee
10-06-2010, 08:06 AM
If you were to make a spindle on a lathe with no furthur finish grinding to the bearing surfaces you would, or should indicate the high point of ecentricity of each bearing surface so you would know how to orient the bearing according to markings of hi or low points. Spindle bearings are a VERY light press fit, finger preassure only, too tight of a fit will result in deformation of the presicion bearing and there goes your accuracy. I would think it is best to grind those surfaces so you have a smooth finish with little resistance to mounting the bearings.

JL...................

Mcgyver
10-06-2010, 09:05 AM
Because 20kRPM are too demanding with a wrong preload, oval bore, non-coaxial bearing seats, not considered thermal expansion, ...


Nick

that doesn't explain why you say "Do you have a cylindrical grinder? No? Forget it!"

explain what there is about making a good 20k spindle that can't be done on a quality lathe between centres with a skilled operator. What tolerances do you insist are required?

MuellerNick
10-06-2010, 09:33 AM
that doesn't explain why you say "Do you have a cylindrical grinder? No? Forget it!"


It's the best answer you could get from asking a simple "Why?". Read posting #10, I explained it.


Nick

Mcgyver
10-06-2010, 10:06 AM
It's the best answer you could get from asking a simple "Why?". Read posting #10, I explained it.



post 10 quantifies that you need to hold a few um's or about a tenth of a thou in your opinion to make a quality 20k spindle. I'll accept that.



Do you have a cylindrical grinder? No? Forget it!

I want to know why you think that's not possible to do on the lathe; to hold this tolerance. Can't you hold a tenth when you have to? I think a skilled man on a quality lathe should be able to, but you claim it can't be done. I mean you didn't say its challenging or difficult, you made if fairly black and white; forget it without a cylindrical grinder. Working to a 10th is not easy, you have to hold your tongue just so, but its hardly impossible.

A cylindrical grinder lets you work hardened material. Other than that, like the lathe, the accuracy to which it will work is dependent upon its ways and infeed mechanism. With a quality lathe I can work to the same accuracy with the compound slewed over and some careful creeping up on the final dimension

I think a skilled man on a quality lathe could produce a part to a tenth same as the cylindrical grinder, but I remain curious why you don't think you could do accurate enough lathe work. Granted a hardened and ground part is nice, however not a requirement for a diy spindle that will be used both occasionally and with care.

MuellerNick
10-06-2010, 11:46 AM
Granted a hardened and ground part is nice, ...

Well, hardened. That's the point. If you think a soft spindle is worth the effort, go ahead. After having changed the collet 100 times you might think different. Add a few broken mills.

Just about what magnitudes we talk:
A standard Haimer CAT 40 holder has a guaranteed runout at the bore below 3m. A precision one below 3 m at the mill 3 * d off the bore.
Out of balance less than 1 gmm or G 2.5 at 25000 RPM.
A 2 mm mill has a feed per tooth (HSS in soft steel or aluminium; final cut) of 3 m


Nick

Mcgyver
10-06-2010, 12:52 PM
for the light, occasional and careful duty in home shop, soft will be fine....hardened would bring greater longevity but you could use prehardened or even annealed chrome moly; its still pretty tough stuff...... however your comment wasn't about longevity of the spindle nose it was to forget it without a cylindrical grinder, that the bearings wouldn't last??

My point here is that it is incorrect and dangerous to throw things out as absolutes as they are often wrong. Someone working carefully between centres can hold a tenth over the short distance required for the bearings, and its s small spindle not a 40 taper so balance is less an issue (although why a cylindrical grinder will product better balance that the lathe eludes me). I grant that a spindle with certain superior qualities can be had with a cylindrical grinder however its debatable the importance of these qualities as in all likelihood the intensity of use will not be high (home, not production). Meanwhile, when you as a knowledgeable person throw it out as an absolute "forget it" you're erroneously discouraging some budding craftsman from having a go at a challenging but doable project

A more precise statement might be "That is a challenging turning project. It's made easier if you can get on a cylindrical grinder and furthermore that would allow you to make the spindle hardened"

These spitting of tenths though just deal with bearing fits...the concentricity of taprer to axis and hence the cutter will be determined by how the spindle nose is finish machined. This is ideally done in situ in the bearings with an internal grinder, take your pick on machine...but you need a internal grinder. Soft may hold an advantage here if no internal grinder is available...without a quench movement is unlikely....first machine the spindle nose with a 60 degree reference surface to go on the tail stock centre (its a small spindle) as well as what ever is required for the collet. then you know the OD turning is concentric to the collet taper. with a quenched spindle and no internal grinding, you can't rely on the taper not to have moved

I agree with the underlying point to your statement - I'll paraphrase as ' spindle making at those rpms requires very fine work, so be up for the challenge and do it right'. As most won't have cylindrical grinders, they'd turn the spindle to a tenth and be fine.

There is an incremental benefit in hardened over soft. However as frequency of use decreases, so does its value...to the point where I hardly think its worth spending thousands just so the spindle is hardened vs soft for home shop occasional use and its certainly not true that a soft one is so bad that one should 'forget it'

TGTool
10-06-2010, 01:50 PM
I'll agree with McGyver on this. For home shop use good enough is perfect as a former manager used to say. Making a spindle accumulates experience and for home shop use the difference can hardly be worth the monetary or time cost difference. I've made an ER-32 collet adapter for my lathe and used it for five years or so. It's soft. Sure, I'd love to have at least case hardened it and ground it afterwards but I don't have the grinders to do that as it would need to be done. In my use I can't find any discernable wear in all this time and if in 20 years more it shows wear I'll take a skim cut on it to return it to original condition. Hardened and ground would be a source of pride. What I have is eminently practical and I don't have the time and money to pay for pride.

Tom Curlee
10-06-2010, 02:36 PM
Quote:

Granted a hardened and ground part is nice, ...

Well, hardened. That's the point. If you think a soft spindle is worth the effort, go ahead. After having changed the collet 100 times you might think different. Add a few broken mills.

My original idea was to use an off the shelf, hardened, ER16 collet chuck. A 6" version can be had for around $25 delivered, with a advertised runout of .0005" (I'll completely believe that number when I test it myself). Ones with runout of .0002" (5m) are not that much more. So I'm not going to worry about the actual spindle chuck wear and accuracy. I'm more concerned about the bearings. And after reading all the posts, I see that I need to be quite concerned about the bearing seat accuracy and alignment. I'm sure that will more than challenge my abilities.

Tom

MuellerNick
10-06-2010, 02:50 PM
Well then, build your own!
You need at least 3*) spindle bearings. Each one of the size we are talking about costs about 130 (- discount). Still I wouldn't make my own, except it is not obtainable what I need.

*) You can do it with 2, but not thermally stable.


Nick

small.planes
10-06-2010, 02:51 PM
Bearing seats can be easy too, especially for the outer races.
Loctite make all sorts of wonderful products ;)

Dave

Who has made 1 precision spindle, using a 50 odd year old lathe, and didnt need loctite, but did need patience.

DFMiller
10-06-2010, 03:21 PM
Have you checked out Wolfgang Engineering on Flea Bay?
It looks like he makes some nice stuff at a reasonable price.
I have not connection nor have I bought anything.
Also does Taig or Sherline have a premade part that might work?
Dave

philbur
10-06-2010, 03:21 PM
How do you know it was a precision spindle? Or do you mean it was a spindle that had precision bearings fitted.

Look at what the bearing manufacturers say about housing and shaft tolerances, concentricity, surface finish, preload requirements etc, etc. You can only claim a precision spindle if you meet these requirements, otherwise you just have a spindle that does (or doesn't do) what you need.

A commercial precision spindle has a whole set of quantifiable performance data that it must fulfil. A hobby built "precision" spindle has a much more subjective set of requirements that normally boil down to "as good as I am capable of within the limits of my budget, equipment and skill".

Just a thought
Phil:)



Bearing seats can be easy too, especially for the outer races.
Loctite make all sorts of wonderful products ;)

Dave

Who has made 1 precision spindle, using a 50 odd year old lathe, and didnt need loctite, but did need patience.

lazlo
10-06-2010, 03:31 PM
Well then, build your own!
You need at least 3*) spindle bearings.

*) You can do it with 2, but not thermally stable.

The Dumore grinding spindles, the KO Lee workhead spindles, and the spindle on my Brown & Sharpe tool & cutter grinder (and I presume many others) use a pair of angular contact bearings, one on each end. The bearing cartridge itself is the preload cylinder. They all have a Bellville washer on the drive end to accomodate thermal expansion.

willmac
10-06-2010, 03:59 PM
I think Nick is correct in his warnings about the required accuracy for making a PRECISION spindle for a quality machine tool.

I recently bought a well used copy of the SKF book 'Bearings in Machine Tools' based on a recommendation (perhaps in this group I can't remember). I need to clean and regrease the spindle bearings in my Thiel mill and this book covers the type of bearing used very thoroughly. This is an excellent book that has alerted me to some issues that I had not fully considered before in spindle design.

The requirements for cylindricity, size, form, surface finish, concentricity across housings, etc, etc are really tight if you want to get the advertised performance and life from very expensive precision bearings. SKF consider that the best or only way to get the required housing accuracy (after ID grinding!) is with a special purpose lapping tool, for which they provide a drawing. This is apart from other factors such as the housing having the correct amount of metal in the right places to ensure that it is stable and maintains an equal radial pressure on the bearing through the housing at all temperatures and for the designed speed and load range.

However, it is certainly true that you can build a reasonably good spindle in a home workshop. It won't be hardened, the bearings would probably not cost hundreds of pounds/dollars (if you are sensible) and it would not last very long if used in a production environment. It would however work reasonably well for something like a Quorn TC grinder, which would have a usage over its life of a small number of hours.

philbur
10-06-2010, 04:15 PM
I don't think you can use a spindle with spring preload for a milling spindle.

Phil:)


The Dumore grinding spindles, the KO Lee workhead spindles, and the spindle on my Brown & Sharpe tool & cutter grinder (and I presume many others) use a pair of angular contact bearings, one on each end. The bearing cartridge itself is the preload cylinder. They all have a Bellville washer on the drive end to accomodate thermal expansion.

.RC.
10-06-2010, 04:24 PM
The 10EE and lathes of similar quality were built for a reason...

MuellerNick
10-06-2010, 04:25 PM
the KO Lee workhead spindles, and the spindle on my Brown & Sharpe tool & cutter grinder

See posting #10.
These are grinding spindles. They can only have a "unlimited" axial load in one direction. The other direction should be less than the load from the springs.

There do exist constructions with two opposed angular contact bearings at each end, but they are very sensitive to thermal expansion. Normally, the inner part gets hotter than the outer one, and -assuming you have an X-arramgement- the play increases. Not good for high RPM.


Nick

Toolguy
10-06-2010, 04:26 PM
Maybe the easiest way for a HSM to get an accurate fit for the spindle bearings would be to bore the seats for the outer races on the lathe with 1/2 to 1 thou. slip fit, put the spindle through the inner races for alignment and pot the outer races with LocTite Retaining Compound. Then there would be 100% fit on the outer races and true alignment of the inner races. It is not the high end way to do the job, but would probably run for years for the OP.

MuellerNick
10-06-2010, 04:36 PM
Oh, and to the "it's so easy":
Why do you think a nut bought from SKF costs about 20 €? Just because it is from SKF? You could get a similar fine thread nut at the hardware store for 99 cent.
No. The thread is ground and the faces are dead perpendicular to the thread. The thread on the spindle where the nut goes on to is ground too. This ensures, that the inner bearing ring is pressed against the outer ring absolutely perpendicular. And not odd like with the cheap nut.

There are so many details ...

Nick

philbur
10-06-2010, 05:01 PM
There are of course very often low tech ways to achieve the same result. In this case possibly scraping of the nut face could achieve a similar result to precision grinding. Possibly it is not done this way by commercial manufacturers because the nut would then cost EUR 30 and not be interchangable. In fact interchangeability is one difficulty the HSMer doesn't have to take into acount.

I think the point is that you need to understand all the issues and then work out a low tech solution for each, lapping shaft and bore diameters (if you have the skill level) is also a low tech alternative to grinding.

Phil:)


Oh, and to the "it's so easy":
Why do you think a nut bought from SKF costs about 20 ? Just because it is from SKF? You could get a similar fine thread nut at the hardware store for 99 cent.
No. The thread is ground and the faces are dead perpendicular to the thread. The thread on the spindle where the nut goes on to is ground too. This ensures, that the inner bearing ring is pressed against the outer ring absolutely perpendicular. And not odd like with the cheap nut.

There are so many details ...

Nick

lazlo
10-06-2010, 05:04 PM
I don't think you can use a spindle with spring preload for a milling spindle.

It sounds like he's talking more of an engraver than a VMC:

"I plan on using this unit for (among other things) milling and drilling printed circuit boards"

MuellerNick
10-06-2010, 05:07 PM
"I plan on using this unit for (among other things) milling and drilling printed circuit boards"

Oh! Then the spring preload certainly works.


Nick

philbur
10-06-2010, 05:12 PM
True, but he did also say "among other things". I guess we need a more detailed definition of use. If his requirement only involves light axial and radial forces possibly an easily adjustable spring preload would be a useful feature.

Phil.


It sounds like he's talking more of an engraver than a VMC:

"I plan on using this unit for (among other things) milling and drilling printed circuit boards"

gzig5
10-06-2010, 05:30 PM
This talented gentleman built a scale turbojet engine from scratch. The construction of the shaft is in part 012. I would suggest that the construction techniques could be utilized to make a high speed grinding spindle. I will quote from the webpage....

"The shaft with compressor and turbine slipped on, after much sweating and angst. The critical surfaces are true to within .0001" of taper, and total runout of any surface relative to any other is maybe .00015 max. I say this not to toot my own whistle, but to show what can be done on a manual lathe with wet-dry paper and some care.

Cylindrical grinding machine? We don't need no stinkin' cylindrical grinding machine!"

http://www.5bears.com/gt.htm

The other projects are equally impressive.

lazlo
10-06-2010, 05:36 PM
"The shaft with compressor and turbine slipped on, after much sweating and angst. The critical surfaces are true to within .0001" of taper, and total runout of any surface relative to any other is maybe .00015 max. I say this not to toot my own whistle, but to show what can be done on a manual lathe with wet-dry paper and some care.

He did that with a Hardinge HLV-BK. Not your typical hobby lathe :)

Tom Curlee
10-06-2010, 06:48 PM
OK, a description of 'among other things'.

A little background first. The basic mill body is built out of .75" - 1.00" aluminum plate (my previous employer had a very generous policy toward use of recycle bin metal, as long as it was for personal use and not sold). The Y axis runs on NSK LS15 ball slides. This axis is 36" X 8", since the 36" X 8" plate already had the 35" linear bearings mounted. The X axis will likely have the same linear bearings, assuming that I can find another pair of LS15 bearing assemblies. The Z axis runs on crossed slide bearings (already mounted with a short ball screw and stepper mount). The basic mill frame can be seen here:

http://img651.imageshack.us/i/millsmall.jpg/

The X and Y axes will mount on the 1" thick vertical plate with the large hole in the center (the aluminum plate came with the hole).

Now, the 'among other things' uses. Besides the PCB milling and drilling, which is why I want to go to 25,000 RPM, I want to be able to engrave front panels of electronic equipment, mill out component holes in the panels, and do light general purpose mill work. I play around with amateur radio microwave, so I plan on using this unit to make waveguide components.

So the use will be everything from very fine drilling, through engraving, to light mill work.

Tom

S_J_H
10-06-2010, 10:44 PM
I think Mcgyver's post #18 was very well said and I agree fully.

I do understand where Nick is coming from too, It's the "if you have to ask?" scenario.

I try and not under estimate others abilities. We have all seen some of the incredible work done by others in this hobby/profession. Mindblowing work at times. Those guys don't know the words " it can't be done".

I built a high speed spindle a while ago. I used a er11 chuck and cheap ball bearings such as found in a diegrinder. I used 3 bearings,2 at the tool end with a very thin brass shim between the outer races, a floating 3rd bearing was located at the belt driven end. A take-up nut and a soft nylon spacer to help allow some growth from heat.
It was driven with a 3mm Gates polyflex belt which is designed for that sort of speed, and a soft start, 6amp variable speed laminate trimmer as the drive motor.
It was just air cooled with heat sink fins. At slower speeds of around 20,000 rpm, they would run for quite a while without overheating. At full speed of around 43,000 rpm, I would keep the operating time under 30 minutes. I made 2 of these little cartridge spindles and they both worked very good for the intended purpose's of light milling and engraving using small cutting tools. For drilling they worked fine if the bit was kept under 1/8".

Setting up the bearing preload will be the biggest hurdle. To loose and it'll be sloppy, chatter and break tiny cutters. To tight and it will overheat or even seize up. Play with it until you get it right. It is a very do-able project for use in a home shop.

http://i109.photobucket.com/albums/n48/S_J_H/x3%20mill/SPINDLEHIGHSPEED001.jpg

http://i109.photobucket.com/albums/n48/S_J_H/x3%20mill/th_cartridgespeedtest43500rpm.jpg (http://s109.photobucket.com/albums/n48/S_J_H/x3%20mill/?action=view&current=cartridgespeedtest43500rpm.mp4)
-
http://i109.photobucket.com/albums/n48/S_J_H/x3%20mill/th_highspeedcircles.jpg (http://s109.photobucket.com/albums/n48/S_J_H/x3%20mill/?action=view&current=highspeedcircles.flv)

This pic is the result of the above early cutting tests. I was pretty happy with it's operation at this point.
http://i109.photobucket.com/albums/n48/S_J_H/x3%20mill/circles004.jpg

Steve

S_J_H
10-06-2010, 11:09 PM
I should probably add the ER chuck's I used for the spindles were USA made Craftsman.
http://www.craftsmanindustries.com/

I used ETM collets. Run out was ~.0002" with this combo.

Steve

wierdscience
10-06-2010, 11:16 PM
Like Macona said,magneto bearings can be had for $35 for a 19mm bore bearing,they work fine for light milling and engraving.

Starting with a off the shelf ER collet shank is a good idea,it works and I have done it.

The housing,start with a dead material(de-stressed) Durabar,Brass or stress proof steel would be my choices,you can add aluminum cooling fins to the outside if needed.

Like said already,bore the housing fit either .000-.000 or +.0005" and use Loctite to retain the races.Forget a press or shrink fit on a precision bearing set.

I would use two angular contact bearings back to back on the business end mounted rigid.On the other end one single angular contact bearing with a wave washer to provide preload to the outer race.

MuellerNick
10-07-2010, 03:34 AM
I would use two angular contact bearings back to back on the business end mounted rigid

You can call me names, but this is prone to fail.
You need paired bearings (they are not sold as pairs, but do have very tight tolerances = double price of the same non-paired bearing). You can select what preload you want when ordering. Low, normal, high. The inner ring sticks out a bit (we are talking about 2 m). When the two bearings are mounted back to back and compressed by a nut, the bearing gets preloaded.
Standard bearings have no guarantee whatsoever in regard to the preload. You can't mount them in a row.

Trying to get preload with shims is also a sure way to fail. Except you do have the metrology (indicator with a resolution well below 1 m), a surface grinder and the skill to hold half a m.


This pic is the result of the above early cutting tests. I was pretty happy with it's operation at this point.


While I'm at being the a$$hole, the surface wouldn't make me happy. There is a lot of chatter. Don't know how you milled the circles and what was the cause for the result. Looks like a lot of play involved.


I'm nitpicking, I know. But you have to be aware of the facts.

Nick

S_J_H
10-07-2010, 07:36 AM
The pic is greatly magnified Nick, to the naked eye it looked pretty good.. :rolleyes: That was cut with a cheap 2 flute ball endmill.

Here is similar magnification of two 3/8" endmills. Note how you can easily see the poor s flute surface on the Putnam vs Metal Removal endmill. With the naked eye you really can't even see it-
http://i109.photobucket.com/albums/n48/S_J_H/misc/IMG_1857.jpg
Earlier version-
http://i109.photobucket.com/albums/n48/S_J_H/x3%20mill/cartridgespindle008.jpg



Steve

MuellerNick
10-07-2010, 08:14 AM
That was cut with a cheap 2 flute ball endmill.

Ball endmill, that explains it to some extend.


Nick

wierdscience
10-07-2010, 02:07 PM
You can call me names, but this is prone to fail.
You need paired bearings (they are not sold as pairs, but do have very tight tolerances = double price of the same non-paired bearing). You can select what preload you want when ordering. Low, normal, high. The inner ring sticks out a bit (we are talking about 2 m). When the two bearings are mounted back to back and compressed by a nut, the bearing gets preloaded.
Standard bearings have no guarantee whatsoever in regard to the preload. You can't mount them in a row.

Trying to get preload with shims is also a sure way to fail. Except you do have the metrology (indicator with a resolution well below 1 m), a surface grinder and the skill to hold half a m.
Nick

Sorry Nick I know full well about preload selection,there are several million vertical mills in the world that disagree.Nearly all Bridgeports and they're clones use this arrangement and quite a few feature spacers to set the preload.

Since a spacer is nothing but a thick shim we are back to simply buying precison shims and inserting them between races.While we are at it we can also buy a prescision spindle nut in the sizes we are talking for about $6.There is no reason in the speeds and sizes we are talking that the bearings,shims and nut should cost more than $125.

RussZHC
10-09-2010, 03:21 AM
Close to it? In kit form

http://www.hemingwaykits.com/acatalog/Grinding_Spindle.html

MuellerNick
10-09-2010, 03:58 AM
Since a spacer is nothing but a thick shim we are back to simply buying precison shims and inserting them between races.

And you adjust the preload with the spacer? No, with a nut.

If you do have two spacers, an inner one and an outer one, they do have to be of exactly the same length. The length is ground as a set. There is no other way! If you want to change preload, one has to be shorter. Again, we are talking about m.


There is no reason in the speeds and sizes we are talking that the bearings,shims and nut should cost more than $125.

The subject is 20 kRPM. If you want to go cheap, go cheap. But then, start with cheap bearings.


Nick

macona
10-09-2010, 04:14 AM
Sorry Nick I know full well about preload selection,there are several million vertical mills in the world that disagree.Nearly all Bridgeports and they're clones use this arrangement and quite a few feature spacers to set the preload.

Since a spacer is nothing but a thick shim we are back to simply buying precison shims and inserting them between races.While we are at it we can also buy a prescision spindle nut in the sizes we are talking for about $6.There is no reason in the speeds and sizes we are talking that the bearings,shims and nut should cost more than $125.

Bridgeport and the like dont preload with the spacers, its done with preloaded bearing sets. There are a few out there that do but they are rare. You need a surface grinder to rebuild one of those. With a bridgeport you just get a set of -DUM preload bearings and match the witness marks on the races. You really cant get shim material thin enough to set a proper preload. You just cant slap a pair of generic unmatched bearings in a mill and expect to get good results.

You can build a spindle for $125, for $50 more you can get one with a motor:

http://cgi.ebay.com/0-8-KW-CNC-router-Water-cooling-spindle-motor-/180570854353?pt=LH_DefaultDomain_0&hash=item2a0adc93d1#ht_2219wt_907

And the home made one wont even be balanced.

Near the end of this PDF from Fafnir is a nice section on spindle design:

http://yabe.algebra.com/~ichudov/misc/ebay/Fafnir-3MM209WI-DUL-Abec-7-Bearing/Fafnir.pdf

wierdscience
10-09-2010, 11:41 AM
Bridgeport and the like dont preload with the spacers, its done with preloaded bearing sets. There are a few out there that do but they are rare. You need a surface grinder to rebuild one of those. With a bridgeport you just get a set of -DUM preload bearings and match the witness marks on the races. You really cant get shim material thin enough to set a proper preload. You just cant slap a pair of generic unmatched bearings in a mill and expect to get good results.


I have calculated the preload dims and ground the spacers accordingly.

Yes,like I said,been there done that.The idea that it's impossible to build a working spindle at home is all I am saying BS to.

I also wasn't talking about using shims on a B-port quill,just a home brew spindle.Since they can be had down to .003-.001 in .0005" increments.Paint also makes a good shim when .00015 increments are needed.

BobWarfield
10-11-2010, 12:27 AM
FYI, SKF provides data on spacer requirements for preloading AC bearings here:

http://skf.us/portal/skf/home/products?maincatalogue=1&lang=en&newlink=9_1_30

If you run through the math, you will need to do spacers that result in a differential spacing of 3 to 8 ten thousandths of an inch. I say differential, because if you read through the material, that doesn't mean the spacer itself must be that thickness, just that the mounting results in that difference for the bearings.

Cheers,

BW

Tom Curlee
10-11-2010, 03:13 PM
My, ask for a simple bearing recommendation and get 50 replies.

I want to thank all of the people who replied: Nick, 'Wierd science', Steve, 'Macona', Bob, 'mcgyver', and all the others for your comments, suggestions, and links. I've certainly learned a lot about precision bearings and high speed spindles the last few days.

I'll probably still try to make a high speed spindle (Fools tread where ---, etc, etc.), but if melts down at 25,000 rpm, I'll just run it slower and find another solution if I just have to have the higher speed.

I was digging through a friends warehouse Saturday (he owns an electronics retail/surplus store) and ran across a number of bearings of the right bore size that look interesting. These are Fafnir AS5KDD units that I think are annular type bearing, but I'm having problems locating full data on them. Some internet listings refer to them as annular, while others deep groove. I've seen them described as allowing both radial and axial loads. Any data sheets out there? I'm interested in maximum speed and proper perload

Another question is that these bearings have the inner and outer races marked with a small 'X' (I wouldn't have thought to look without comments previously made in this thread - thanks). Pictures can be seen here:

http://img524.imageshack.us/gal.php?g=as5kddbearingouterracee.jpg

The small Xs on the inner and outer races are on opposite sides of the bearing, one on each side. Is the X side the direction from which the external load is applied or what?

I've noticed that most spindles, both commercial and home made, have a round housing that is usually mounted in a split mounting block. Is this due to wanting even clamping pressure, just easier to make a round housing, or what? Is there anything inherently wrong with making the spindle out of a rectangular block of aluminum (discounting thermal expansion and stresses), with the bearing mounted in the ends?

Thanks to all for all the great info.

Tom

MuellerNick
10-11-2010, 03:29 PM
The small Xs on the inner and outer races are on opposite sides of the bearing, one on each side. Is the X side the direction from which the external load is applied or what?


The marking should be the high point (point with most runout).

Let me tell you, that a 2RS at 25000 RPM is a nice way to generate heat. A lot of heat! Data is hard to get, but it is something like 50 ... 100 W.
If you intend to make a good spindle, use non-contact seals like the ones from GMN (http://www.gmn.de/en/seals/products/labyrinth-seals-made-of-metal/examples.html). Or make your own.
Also very common is "lock air" Don't know how you call that. You blow a certain amount of air (together with a bit of oil = 1 drop per minute) through the bearings. Thus cool them and avoid dirt coming in from outside.
That air has to be filtered, not right out of the compressor!


As I said: So many details ...
Nick

lazlo
10-11-2010, 03:34 PM
Also very common is "lock air" Don't know how you call that. You blow a certain amount of air (together with a bit of oil = 1 drop per minute) through the bearings. Thus cool them and avoid dirt coming in from outside.
That air has to be filtered, not right out of the compressor!

"Oil mister"

It can be done a lot simpler than it sounds -- the Dumore spindles have a piece of felt on the inside of the bearing, and there's an oil reservoir in the center of the spindle sleeve. When the spindle spins, oil is driven out, it saturates the felt, and the spinning felt creates a mist of oil...

lazlo
10-11-2010, 03:39 PM
I was digging through a friends warehouse Saturday (he owns an electronics retail/surplus store) and ran across a number of bearings of the right bore size that look interesting. These are Fafnir AS5KDD units that I think are annular type bearing, but I'm having problems locating full data on them.

'K' means deep-groove (non angular contact) bearing. 'DD' means double shield.

They'd be OK, but they're not precision bearings, and they're not meant for preloading.

MuellerNick
10-11-2010, 03:48 PM
"Oil mister"

That device is called oil mister, right.
But in fact, it doesn't make an oil mist. It generates a thin film of oil in the tube that is slowly driven into the bearings. So you should use a transparent tube to see wether it works. You also have to consider how that oil film flows within the spindle. It is really bad if the oil collects at the bottom and gets spilled in relatively huge amounts into the bearing from time to time.

What I don't know is how you call the air that prevents dirt coming into the bearings.


Nick

Tom Curlee
10-11-2010, 04:00 PM
Nick;

You said that :

The marking should be the high point (point with most runout).


Is this the high spot relative to the mating surface of the bearing (the outer surface relative to the inner)? What do I do with the markings when assembling the bearings into the spindle? Match them up relative to the mark on another bearing, set them opposite, or what?


Let me tell you, that a 2RS at 25000 RPM is a nice way to generate heat.

What is a 2RS?

Thanks,

Tom

p.s. I spent some time in the lat 1980's and early 1990's in your area I was in Freising, Bad Mergentheim, and Trier on business. Munich was a great city to visit (the Deutsches Museum has a great collection).

lazlo
10-11-2010, 04:05 PM
What is a 2RS?

2 Rubber Seals. Bearings come open, shielded, and sealed.

Shields are thin metal foils that don't touch the inner ring, and therefore aren't tightly sealed, but have virtually no drag. Shields will have a "2Z" or "ZZ" suffix.

Seals are rubber seals that a pressed into the inner ring. They're completely sealed, but have a lot of drag.

For a high-speed spindle, I don't think seals would be a good idea. A compromise solution is a "non-contact" seal, which uses a labyrinth-type arrangement to be mostly sealed, with much less drag.

MuellerNick
10-11-2010, 04:15 PM
Is this the high spot relative to the mating surface of the bearing (the outer surface relative to the inner)? What do I do with the markings when assembling the bearings into the spindle? Match them up relative to the mark on another bearing, set them opposite, or what?


It is the high spot relative to the mating/mounting surface. For the inner ring it is: bore::race for balls; for the outer:circumfence::race

If you mount two (or more) of those bearings, the X for the two inner rings should be aligned, same for the outer ring. You should also place the high spot of the bearing at the low spot of your spindle.

So many details ... (I'm just waiting for Mister knowsitallanditwillbreak to womit his "wisdom" once more)

Nick

Machtool
10-11-2010, 07:22 PM
What I don't know is how you call the air that prevents dirt coming into the bearings.
Nick.

Ive always known that as Air Purge

Regards Phil

lazlo
10-11-2010, 08:18 PM
Ive always known that as Air Purge

Thanks Phil -- I've seen oil misters, but I'd never seen a spindle "Air Purge" before. Can you buy the Setco air seals off the shelf?

http://www.moldmakingtechnology.com/mag_images/010703a.jpg

http://www.youtube.com/v/khN6A4d2kCk

By the way, want your grease back? :)

Machtool
10-11-2010, 09:57 PM
Can you buy the Setco air seals off the shelf?
G’day Robert.

Interesting that you linked Setco, I was going to do that, they give the best example of it.

I don’t think they will sell that seal cartridge, I tried years ago and they wouldn't sell to me.

If you read these links, it claims there only available with Setco Spindle Rebuild services.

http://www.setcousa.com/files/pdf/S-0008-1a.pdf
http://www.setcousa.com/files/pdf/S-0006-1a.pdf

I always thought they looked like something Chicago Rawhide may have made. But I can’t find them direct. The active part of the seal if just a Vee-Ring.

http://www.moore-international.com/v_ring_s.html

We do a modification on a lot of spindles to include a ring like that running on a ceramic metal sprayed face that is ground and polished. Then you port air so that it unloads the lip. Setco claims to have invented that in 1999. The company I did my apprenticeship in were doing that back in the early 80’s.

You’d best find a use for that grease, remember it has a 2 year shelf life, the oil separates out of the base over time.

Regards Phil.

wierdscience
10-11-2010, 10:50 PM
FYI, SKF provides data on spacer requirements for preloading AC bearings here:

http://skf.us/portal/skf/home/products?maincatalogue=1&lang=en&newlink=9_1_30

If you run through the math, you will need to do spacers that result in a differential spacing of 3 to 8 ten thousandths of an inch. I say differential, because if you read through the material, that doesn't mean the spacer itself must be that thickness, just that the mounting results in that difference for the bearings.

Cheers,

BW

Yes,and infact using spacers mean the bearings can run at higher speeds-

http://www.skf.com/portal/skf/home/products?maincatalogue=1&lang=en&newlink=9_1_21

wierdscience
10-11-2010, 11:15 PM
And you adjust the preload with the spacer? No, with a nut.

If you do have two spacers, an inner one and an outer one, they do have to be of exactly the same length. The length is ground as a set. There is no other way! If you want to change preload, one has to be shorter. Again, we are talking about m.



The subject is 20 kRPM. If you want to go cheap, go cheap. But then, start with cheap bearings.


Nick

Yes you do preload with the spacer,as all milling spindles do.There are two spacers involved in this,one for the inner race and another for the outer race.As Bob pointed out it's the differential measurement between the two that determines the amount of preload.That differential is between .0003 and .0008" this is in the realm of a tenths reading micrometer and a lapping plate.It is not strictly the domain of Nasa and NIST.

So,you are violently agreeing with me and don't even know it:)

25,000k is not rocket science.Skf shows they're angular contact bearings running 60,000 on grease.

If you are worried about heat,water cool the housing which is.......just exactly what the commercial router spindles do.

I for one would build the spindle,fit it with angular contact ball,preload it with spacers and grease the bearings with SKF's uber good LGLT-2 grease.Add on a water cooling jacket and run it.

.RC.
10-12-2010, 01:26 AM
I preloaded the el-cheapo angular contact bearings in my bridgeport clone simply by lapping one of the spaces a tiny bit...


I am not making parts for NASA so who really gives a **** if I took off 2.6543 microns too much.... It turned the mill from a crap performer to one that performs good enough for my needs....

It amuses me when HSM people go on about the extreme accuracy they think they need... I have repaired a not small amount of general stuff over the years and have not yet needed a cylindrical grinder for some extreme accuracy that I can't measure anyway....

Richard-TX
10-12-2010, 02:00 AM
2 Rubber Seals. Bearings come open, shielded, and sealed.

Shields are thin metal foils that don't touch the inner ring, and therefore aren't tightly sealed, but have virtually no drag. Shields will have a "2Z" or "ZZ" suffix.

Seals are rubber seals that a pressed into the inner ring. They're completely sealed, but have a lot of drag.

For a high-speed spindle, I don't think seals would be a good idea. A compromise solution is a "non-contact" seal, which uses a labyrinth-type arrangement to be mostly sealed, with much less drag.

NTN uses a no contact seal om some of their bearings and that style of bearing is all that my supplier stocks. I have tried other manufacturers bearings that use a contact seal and to date have not been able to measure any difference whether the contact seals are in place or not. Bearing size was 6204 spinning at 14,000 RPM, common EM class grease lubricated.

If you want a low friction external seal, I would look at the Nilos ring seals. Very low friction and they seal well. They also have a labyrinth seal that is excellent. Nilos rings are available from your SKF dealer.

Nilos rings makes sealing 7200 series (angular contact) bearings a snap.

Black_Moons
10-12-2010, 03:28 AM
I preloaded the el-cheapo angular contact bearings in my bridgeport clone simply by lapping one of the spaces a tiny bit...


I am not making parts for NASA so who really gives a **** if I took off 2.6543 microns too much.... It turned the mill from a crap performer to one that performs good enough for my needs....

It amuses me when HSM people go on about the extreme accuracy they think they need... I have repaired a not small amount of general stuff over the years and have not yet needed a cylindrical grinder for some extreme accuracy that I can't measure anyway....

Heh, I saw a guy mentioning about CNC being great for one offs of a shifting rod cam unit ( http://homepage.ntlworld.com/stevenson.engineers/lsteve/files/4th%20axis%20cam.jpg ), Saying 'In the past you would have needed probably 3 master cams, produced at great cost plus a complicated follower drive system, totally out of context for a one off.' But all I could think was: 'That cam moves dogs in and out that have a tollerance of at least 1/8" I bet.. ie, you could make that thing by drawing lines with a pen after measurements taken with a ruler, and just following the lines by eye while milling with a rotary table and X axis.

Excessive accuracy can make a problem much harder and longer then it needs to be. It might be good to 'learn' to be accurate, but its not good to force yourself to be excessivly more accurate then you need to be for the job at hand.

MuellerNick
10-12-2010, 06:10 AM
25,000k is not rocket science.Skf shows they're angular contact bearings running 60,000 on grease.

Well, I somehow overstressed it regarding cooling air etc. I was getting more general.
But a B7003C (ID 17 mm, OD 35) makes 38 kRPM in grease and 60 kRPM with oilmist.

Regarding my comments, as some start to tell that it's nonsense and a skater bearing did work for them etc.:
It might have worked for you. Others who read this and intend to build a internal grinding spindle that not just doesn't remove material by making sparks but want a good surface, will fail. First you have to understand how things work and how the ends connect, then you can reduce to your needs. If the result is a failure, at least you know why.

If you want a high speed milling spindle, the needs are not much less. Runout and axial play plays a huge role in surface finish and tool life.


Nick