[ballendo]

Hu,

I notice in your drawing the bearings are still back-to-back...

While this is gonna be okay for a small mill like this thread is discussing; remember the spaced lower bearings if ya wanna "do it right" and get a MUCH stiffer spindle. I agree that labyrinth seals are the way to go.

Now, as for that pressure: It's not ridiculous, but it does take some push. There is a QC retrofit for BP mills(R8), that uses the operator as the spring "compressor". Tool change is done by pushing the quill lever in reverse(up) so that the top of the spindle meets an added fixed stop which compresses the spring, and releases the tool. My point in describing this is to show that the force required is not "tons"... Think about the multiplying effects of the Balls on the retention flare.

I was looking at a Haas VMC service manual while writing the last post, and VMC's don't have quills...

Certainly in the case of a retrofit to a quill equipped machine, you'd want to be mindful of the pressure downwards during a toolchange. possible you could have the mount for the toolchange piston ride ON the quill too; which would confine the force to the spindle bearings only. Or try to use the "upward" technique described above.

Assuming full ATC capability, you could incorporate a stop which comes in with the tool carousel shuttle. then the shuttle would need to resist the force, not the spindle/quill.

Typical cnc tapers are non-locking types. You don't want that tool stuck during an -attempted- toolchange!

Hope this helps,

Ballendo

Originally posted by HuFlungDung
Thanks for that info, Ballendo. It must take a fair bit of pressure to overcome that retention spring, would it not? So is there actually enough tension in the retention knob system that the tapers want to lock a little bit, like a regular NMTB taper with a drawbolt?

Seems like it would or could be quite a push to release, so a person would have to be mindful of creating the Z axis quillbox with adequate strength for this purpose.

[HuFlungDung]

Ballendo, what you are saying about the bearing spacer makes sense all right if there is room for it. They would also have to be perfectly flat lapped, parallel and equal, spacers when used in a precision bearing set to maintain the factory preload.

What I sketched above is similar to the arrangement in a Shizouka knee mill quill type spindle.

Does the Haas use the drive tangs for driving the tools? My retrofit doesn't use them because there is no method of finding the alignment. This means the tapers do lock together. My "Quickdraw" toolchanger used an impact wrench to suck the tool in tight with the drawbar. It "snaps" out pretty sharply when the taper releases the tool.

[ballendo]

Hu,

Yes, the Haas(and most other VMC's) use drive "tangs"... Of course the tapers "lock" rotationally, but they are not supposed to "stick together" too much axially. That's the reason for the greater angle than a locking taper like a morse... (When you pull 'em together they "connect", but you can get 'em unstuck easily for the tool change)

The Haas--and other VMC's without vector drive--typically use a shot pin to align the spindle for tool changes. It is solenoid or air cylinder controlled.

You're right about the precision needed for the spacers.

Ballendo

Originally posted by HuFlungDung
Ballendo, what you are saying about the bearing spacer makes sense all right if there is room for it. They would also have to be perfectly flat lapped, parallel and equal, spacers when used in a precision bearing set to maintain the factory preload.

What I sketched above is similar to the arrangement in a Shizouka knee mill quill type spindle.

Does the Haas use the drive tangs for driving the tools? My retrofit doesn't use them because there is no method of finding the alignment. This means the tapers do lock together. My "Quickdraw" toolchanger used an impact wrench to suck the tool in tight with the drawbar. It "snaps" out pretty sharply when the taper releases the tool.

[h3ndrix]

Hey Stevie,

I ll definately check out the Amco bronze, but I think I might go with a harder alloy steel instead of bronze, but thx

Hey John S.,

Thx for all the comments, I am always happy to get feedback (positive or negative)...thats why I am here. I am currently changing the design a little based on the feedback I got. I think I will move the bearings down closer to the nose of the spindle and use larger bearings. I am thinking about setting them up kinda the way Hu has it in the drawing, but with minor changes. The recess will definately change and get more finessed as I progress on the design. I will convert the design to belleville washers, as everyone seems to agree that this is the way to go and the picture that WMS put also has belleville washers.

Hu,

Thx again for the great replies. I think that your drawing has some important factors that I am starting to take into consideration and will implement in the next revision of the design. I will use larger bearings and move them closer to the nose, this will help me with the overhang ratio that you mentioned. I think I will implement the labyrinth for the grease, it seems like a real good idea. Because I cannot afford ABEC-7 bearings, I will have to use ABEC-1 or 3 bearings, which probably dont come with precision ground surfaces, so I dont know how good that will work out when I put the bearings back-to-back.

I will have to look into the clutch needle bearing that John S. mentioned, but I dont know how that would work out exactly. During my research, most CNC's I looked up did not have Drive dogs on the spindle, more interestinly, the collets I saw for these machines did not have spaces for drive dogs to go in anyway.

Btw, I am using Solidworks 2004 for this stuff, luckily my school has it.

Thx ppl,

H3ndriX

[HuFlungDung]

H3ndrix,

We're just giving you ideas, so take them and adapt them for your purposes.

RE: Abec 7 bearings. Yes, these can be pricey. However, I am pretty sure that if you specify when you order that you want flush ground, standard preload, that this option is available in the lower precision classes, too. FAG bearings, one of many choices, has a good selection to choose from.

The quality of the bearing surface of your spindle and housings is a very important factor in bearing temperature and life. So take your time and do a good job. A close slide fit in the housing bore is good. A light interference fit on the spindle is good, maybe .0005", which is easy to assemble by heating the bearings slightly on a hot plate and then sliding them on.

When you design the labyrinth, which I did not do a good job of in my sketch, try to remember how you will press the bearings off the spindle. You might be able to use the labyrinth plate as bait for your puller/press.

Murphy's law says you will assemble everything permanently, and after that, you will have to take them off once more

[turmite]

Hey guys is this a great place or what. I have looked for months trying to find drawings to see how an atc spindle works and now all this info.

I am not a machinist so I ask this out of pure ignorance. Does the cone on the inside of the spindle need to be ground or will cnc machining be sufficient? Since my main interest is in wood working with cnc would the labyrinth seal be prone to leaking anything down onto the table? Wood and grease don't mix.

Now for the final question. I would like to take a design like this and enlarge it for use with larger tools. Is there some set dimensions that would go with each size tool holder, i.e. BT30 as opposed to a Cat 40 etc. like the overhang Hu mention or the number of bearings, diam of the spindle itself as well as the housing? Ok so that wasn't one more!

Mike

ps a Colombo 10.5 hp 22,000 atc spinble that is made for the woodworking industry only measures 16" long and around 6"X6"
width and uh..width! Weighs 62 pounds and that includes the motor itself. It is also $7,000.00!

[ballendo]

Termite,

There are several commercial spindles on ebay right now, and pricing is looking REAL good...

Do a search on CNC router once you're at the ebay home page.

Ballendo

Originally posted by turmite
ps a Colombo 10.5 hp 22,000 atc spindle that is made for the woodworking industry only measures 16" long and around 6"X6"
width and uh..width! Weighs 62 pounds and that includes the motor itself. It is also $7,000.00!

[John S.]

Originally posted by h3ndrix
Hey Stevie,

I will have to look into the clutch needle bearing that John S. mentioned, but I dont know how that would work out exactly. During my research, most CNC's I looked up did not have Drive dogs on the spindle, more interestinly, the collets I saw for these machines did not have spaces for drive dogs to go in anyway.

Thx ppl,

H3ndriX

H3ndriX,
Virtally all metal working CNC's have drive dogs to locate and drive the holder.
The steep 30 and 40 type tapers are not driving tapers in their own right.
Some wood working machine have no drive dogs and just rely on the taper and a large spring pull to locate and hold, so it's a trade off on a small machine against drive dogs or other methods and having to apply large forces to release the holder.

Collets don't have drive dogs or any other method of driving normally There are a few exceptions be we won't go into that here.
collets like the ER series and the double angle series rely on being tightened up into the holder off the job. It's not the job of a tool changer to swap collets only holders.

The roller clutch bearing isn't a new idea and has been used for a long while.
I can't share my drawings as they are being done in partnership with someone but a quick description here.

On the holder after the taper and before the collet securing nut there is a parallel portion. When it's engaged in the spindle this is supported in a roller clutch needle bearing. This looks just like a normal needle bearing and also has the advantage of giving exta support.
If the holder attempts to rotate in the spindle thru cutting forces then the needles in the clutch bearing ride up small machined ramps in the bearing housing and grip the holder.
Some stud extractors work this way.

The advantage of this type of drivig mechanism is that the holder can be inserted into the spindle in any position making auto tool changing a lot easier.
In fact you can even have a stand on one end of the machine bed with tool in and just sent the head to each in turn and swap holders.

John S.

[ballendo]

John,

That's an interesting idea... How do you make the fit close enough that the one-way engages, without making it so that the tool holder/spindle recess "runs into" the edge/corner of the one way upon insertion?

If they make a tapered one way, this would be a GREAT place to use it<G>

Ballendo

P.S. Thinking further, if the taper upper part was a separate piece and RODE on a one-way. problem gone! (Think lathe bullnose live center, but with one way bearings)

The larger diameter part of the taper--the lower part would include the recess and nut--or whatever-- to hold the tool...

P.P.S. But there is a reason commercial CNC index to a specific
spindle angle for toolchange... Things like boring bars, when used with the G80 series of canned cycles need to know which direction "retract from surface" is. And probes may depend upon an IR link, which is directional...

[HuFlungDung]

Originally posted by turmite
Hey guys is this a great place or what. I have looked for months trying to find drawings to see how an atc spindle works and now all this info.

I am not a machinist so I ask this out of pure ignorance. Does the cone on the inside of the spindle need to be ground or will cnc machining be sufficient? Since my main interest is in wood working with cnc would the labyrinth seal be prone to leaking anything down onto the table? Wood and grease don't mix.

Now for the final question. I would like to take a design like this and enlarge it for use with larger tools. Is there some set dimensions that would go with each size tool holder, i.e. BT30 as opposed to a Cat 40 etc. like the overhang Hu mention or the number of bearings, diam of the spindle itself as well as the housing? Ok so that wasn't one more!

Mike

Mike,

The taper can be cut on a conventional lathe with a taper attachment, a standard lathe using the compound rest set at the proper angle, or a cnc lathe will also do it. You do what you gotta do, using whatever you can get your hands on.

It is possible to do fine work with a manual lathe, too. CNC is not the last word in everything. Simply use a commercially made tool shank as a gauge to get the taper angle set correctly. I check the fit on a tapered socket by using a plain felt marker, drawing a few axial lines on the tapers, then twist the shank into the socket and observing the rub off points. You should attempt to get a few points of contact all along the shank through proper setting of the tool angle. Then, you can polish it up after that, without trying to change the angle by hand.

When the taper is almost right, it gets difficult to judge which way to tweak it with polishing. Use a dial indicator (if you don't have one, get one, there are lots of cheap indicators available). By checking the tool holder for wobble near the end of the taper socket, you can detect whether the outer end of the tapered hole is doing the contacting, or the inner end. When I say check for wobble, I mean push your tool shank into the socket firmly by hand and while holding it in firmly, try to move it from side to side

RE: labyrinth and grease: yes, it will throw grease off for a time and this is annoying. However, the labyrinth does not induce grease flow or throw off, but rather slows it down. The labyrinth is a set of concentric circles, not a spiral. With some planning though, I don't see why you could not add a cup shaped cap to the spindle that would sweep around the end of the bearing housing and upwards, as a last ditch effort to catch the grease.

re: taper shank dimensions: Machinery's handbook is one easy to find source of standard taper angles and spindle noses. However, it won't tell you how to build the rest of the spindle

[HuFlungDung]

One of the "new" features that some manufacturers are using, is to create a spindle nose with such precision, (and the toolholders, too), that the flange on the toolholder actually makes contact with the end of the spindle just at the moment the tapers contact.

This takes phenomenal accuracy to achieve. I suppose on a "home-made" spindle, that a person could add some kind of a nut to the end of the spindle that you could manually jam against the back of the toolholder after it was already locked into position. This would really stiffen up the tool.

If you wanted to get really fancy, maybe you could make a very short stroke hydraulic cylinder at the end of the spindle. Then, drill some kind of oil passage through the spindle, back up to where the retention knob mechanism is working. Incorporate some kind of a small oil reservoir and piston assembly at the belleville washer seat......I dunno, I never planned it out

[John S.]

Ballendo,
The fit doesn't need to be that good. A clearance of about 5 thou is fine as the rollers can take this up easily.
The bottom of the parallel part on the holder has a short taper to allow the roller to line themselves up when entering.

John S.

[h3ndrix]

John S.,

I have looked at the roller needle clutch bearing and wanted to ask you if this is the one you are talking about. If this is the one, then to lock the spindle in a certain position, the spindle direction has to be reversed first and then the spindle will stop at the designated location, at least thats what I got from McMaster's site, please check;

Clutch bearings roll in one direction, then lock to transmit torque load when the rotation of the shaft is reversed (an arrow on the drawn-steel outer shell indicates load direction). Use for indexing, backstopping, and overrunning operations on hardened and ground shafts. Low profile is great for limited spaces. They may be used in conjunction with needle-roller bearings to provide a load-carrying capacity in addition to the clutch action.

Is this the one?

Thank you,

H3ndriX

[h3ndrix]

So, I did some work on the sindle, incoprated some of the changes;

- put the drives on the spindle nose
- doubled up the bearings, front 40mm ID; 80mm OD, back 30mm ID; 62 mm OD
- put front and end caps
- put a sleeve running in the middle

- havent put the clutch bearing yet, as I am mistified on how it would work in my design
- havent put the grease labyrinth in yet
- havent put the drive pulley on the end of the spindle



thx

H3ndriX

[John S.]

H3ndriX
That's a lot better.
Couple of points though. Space the two lower bearings our a bit to get preload on the bearings. At the moment you are relying on the factory settings for preload by just putting them back to back.
You have no setting options.

Loose the double bearing at the top, it's too complex and you will find one set of preloads will be fighting the other.
All the support and loading will be on the lower two beraings so there is no need for a bouble bearing at the top. A single radial ball race will be satisfactory.

John S.

[HuFlungDung]

Looks good, Hendrix.

John brings up a good point about preloading both sets of bearings against one another.

If you purchase bearings in pairs, specify flush ground, standard preload and you will be okay. I don't think that amateurs have the wherewithall to know when a bearing as been properly preloaded. You are more likely to screw up the bearing alignment with homemade spacers than you are to gain much benefit.

If you rely on preload with some kind of a locknut to hold the preload, you will be disappointed. Using the "preload grind " bearings allows you to snug the retaining nuts up tight, without affecting the preload whatsoever. I'm not implying that this is what John was implying , but just for the record, its been said.

I disagree that standard radial bearings are good enough for spindles. They have too much internal clearance and there is nothing you can do about it. But, the upper set of bearings needs to float in the housing. So what you do is remove the outer retaining nut from the design. Keep the inner race nut and this will lock the pair of flush ground bearings together. The outer races will be trapped, provided you assemble them correctly, just as you have them drawn there.

Make sure there is a little bit of space below and above the upper bearing pair, so it can expand or contract a wee bit without fouling on the shoulders.

Can you also design some kind of retaining locks in the extended bearing housings? I don't know if you can reach from the inside to insert any bolts, etc. You need to ensure that it cannot drift out of the main housing bore.

Edit: I take it that there is a little bit of clearance between the spacer sleeve and the lower bearing housing. It just looks tight

[John S.]

Originally posted by HuFlungDung


I disagree that standard radial bearings are good enough for spindles. They have too much internal clearance and there is nothing you can do about it. But, the upper set of bearings needs to float in the housing. So what you do is remove the outer retaining nut from the design. Keep the inner race nut and this will lock the pair of flush ground bearings together. The outer races will be trapped, provided you assemble them correctly, just as you have them drawn there.

Make sure there is a little bit of space below and above the upper bearing pair, so it can expand or contract a wee bit without fouling on the shoulders.

Hu,
That yardstick of milling machine design, the Bridgeport uses two angular contact bearings with spacers at the bottom of the spindle for thrust and preload and a standard off the shelf radial bearing at the top that's allowed to float as you say.
All the top of the spindle is doing is supporting itself against thrust loads from the pullies.

If we dare to mention Mill /Drills in the same post as Bridgeports they use two standard radial bearings top and bottom on the spindle and I have seen some nice accurate work performed on these.
In fact the mill that Hendrix is converting and the Taig that I am doing work exactly the same way.

John S.

[HuFlungDung]

Yes, John, I've seen the construction just as you described it. There's lots of things you can get away with.

But my way is better, if you want better.

In a spindle that I rebuilt, I found that it ran a lot cooler with the bearing setup as I described. I guess the radial ball bearing is just not in a precision class that can stand 5000 rpm for 10 hours Less precision = more heat.

The top bearing does provide additional stiffness to the spindle, if it is actually running preloaded. Whether this machine warrants it is questionable. But the heat issue is a real one to think about.

[ballendo]

Hello,

Yes, it's looking MUCH better!

I happen to agree with both Hu AND john about the upper bearing. As Hu sez, it IS better. As John sez, it's probably overkill for this level of machine. In THIS case, I'd tend to favor John. But HU IS right TOO, IMO.

I see a few problems still. First, how did you put that upper bearing lock on? Because the the threads don't start until you get TO the retainer (relating to this, you usually have an increase--or decrease, depends on your point of view-- in spindle diameter whenever you encounter a set of threads. Your spindle doesn't show this...

Next, get rid of the sleeve. It is doing nothing for you except making the thing more complex. Just increase the spindle diameter to replace it. Then implement Hu's suggestion to lose the upper OUTER race retainer. Which appears likely as there's nothing now holding it on! (I'm not sure why Hu is worried about the extensions. I'd use red loctite and forget about it (IN THIS CASE!) as the upper inner retainer will keep any "play" to the level of the shaft thermal expansion. And you should be using a press or shrink fit for at least the lower extension at assembly anyway...

Finally, get rid of the "neck" in the spindle interior. It's not needed. The larger upper bore diameter will provide a stop for the spacer at the bottom of the belleville washer stack; so it's just extra difficulty in the making that gains you nothing...

And while you're at it, be sure to draw the intended way to hold that lower outer race retainer. That's a KEY part of this whole thing!

Hope this helps,

Ballendo

P.S. One factor that John mentioned with the BP example is that those bearings ARE spaced. This IMO is why the upper radial is ok. Wothout the spacer--for the reasons Hu mentions-- having the upper preloaded set is a good thing. But on the size and overall quality of the machine THIS is intended for, I'm still with John...

P.P.S. You'll want to draw the threads on the locking ring (light blue) for the lower inner races.

And just for clarity's sake: The "space" Hu mentions for expansion will be in the upper bearing housing extension. Assuming yo follow the changes I've outlined here. In other words, the bore for the upper bearing(s) will be a bit deeper than you have it drawn currently. This allows the upper bearing(s) to float with thermal changes. (They'll float in the housing extension, since they're constrained on the shaft.)

One more: There's no reason for the "necking down inside the spindle BELOW the sliding sleeve (which has the toolholder locking bearings) Again, this just makes the machining harder, and promotes the idea that designers have never been in a shop and had to make REAL parts

Originally posted by h3ndrix
So, I did some work on the spindle, incoprated some of the changes;

- put the drives on the spindle nose
- doubled up the bearings, front 40mm ID; 80mm OD, back 30mm ID; 62 mm OD
- put front and end caps
- put a sleeve running in the middle

- havent put the clutch bearing yet, as I am mistified on how it would work in my design
- havent put the grease labyrinth in yet
- havent put the drive pulley on the end of the spindle



thx

H3ndriX

[h3ndrix]

Thx guyz, great feedback as always

About the bearings on the bottom, I have to go with Hu on this one because of space limitations. The front end of the spindle has already reached 3" without any tool holder or bit. The more I add to this dimension, the more I have to raise the gantry to regain this space and the less stable the CNC gets.

The upper bearings I want to agree with John S., this time because of mass and weight. As the weight increases on the head, the holding force of the head to the Z travelling axis has to be increased for a stable hold. I will show my plans for the whole CNC real soon as I am finalizing my decisions on it for revision 1.0.
Although if it is really curicial that I use 2 Back to back bearings, I will, but If its not, I will move to a single Angular contact bearing facing down, to help pick up the forces that will push up the spindle or maybe go with a deep groove bearing (similar to Bridgeport spindle) so it will pick up radial forces.

Ballendo, I did not really understand which neck you are talking about in the sentence below. If it is the one that the draw bar used to pass thru, then I have already removed that one in my recent revision (I will post soon). Like you said, the belleville springs only sit on the very edge, so I have enough of an edge at that point. But If you are talking about the groove for the balls to fall into when the collet is being released, I dont see how that could be removed without sacrificing from the taper that holds the collet. Although that neck doesnt seem like much, when a line parallel to the spindle is drawn from the top of the groove to the taper, 1/3 of the taper has to be sacrificied for that groove to be removed. Unless I understood that wrong and I am blabbing for no reason. BTW, designers make damn good machinist (well...some)

One more: There's no reason for the "necking down inside the spindle BELOW the sliding sleeve (which has the toolholder locking bearings) Again, this just makes the machining harder, and promotes the idea that designers have never been in a shop and had to make REAL parts

thx all

H3ndriX