[aarongough]

Hey all!
I've been looking for a while now for a manual lathe that would suit my requirements for CNC conversion... I haven't had much luck, so I'm going to go ahead and commit to building a CNC lathe from scratch.

Now let me just say up-front that I realize building an accurate spindle and lathe is no laughing matter... I'm treating this first and foremost as a learning exercise. If I am able to get a usable lathe at the end then great! If not, then I will investigate other options. Either way I will have learnt many things that I probably wouldn't learn during too many other projects!

So here's the plan:

• Scratch-built spindle with tapered roller bearings
• 5C spindle taper with 1-3/8" spindle bore
• 4" swing over cross-slide
• Usable bed length of about 10-12"
• 3/4HP brushed DC motor
• Encoder on spindle
• Gang tooling (I'm also sketching up concepts for a swiss-style gang lathe... We'll see about that...)
• Ballscrews on both axes with a Gecko G540 driving 380 oz/in steppers

[aarongough]

Here's a concept sketch for the headstock/spindle design:



I would like to place the roller bearings closer together to minimize the loss of preload due to thermal expansion of the spindle, however I have been unable to find a ball bearing supplier online that has standard ball bearings with a 2.125" ID and 3.5" OD to match the roller bearings.

If I can find a matching ball bearing then I will move the roller bearings closer together and then add the ball bearing toward the rear of the spindle (with a spacer between it and the roller bearings) so that the ball bearing takes most of the side-load and the roller bearings are left with the thrust loads.

If anyone has a lead on a 2.125" ID x 3.5" OD ball bearing let me know!

Ideally I would like the front roller bearing much closer to the spindle nose, however I have been unable to find roller bearings with and ID of great than 2.125" (at least affordably, the 2.125" ID roller bearings are $15 each...). That ID is only just big enough to accomodate the 5C drawtube (1-3/8") so the 5C taper would have to go ahead of the bearings.

If I can find larger bearings then the taper would be able to be moved inside the front bearing and a stiffer setup would result.

That being said I'm not hugely worried about the design as shown. It's better than a 7x12 bearing setup and roughly the same as the setup in a Sieg 10x22 lathe.

[aarongough]

Because the roller bearings are not sealed (or even shielded) I am probably going to maintain a small positive air pressure inside the headstock by way of some small pump, as well as making tight-fitting plastic shields for the nose and rear of the headstock...

Right now the design I see in my head is a pretty simple slant-bed gang tooled lathe. I will likely use dovetail ways instead of linear guides as I want to make as much of the lathe myself as practical.

I am also considering whether or not a swiss-style gang tooled lathe would be practical. In this case the headstock would move along the Z axis instead of the tool slide. Adjacent to the tool slide there would be a fixed bushing which would act effectively as a follow-rest.

This may actually be an interesting way to go because if the bushing was made removable then it could function both as a swiss-style lathe, and as a regular gang-tooled lathe. The swiss-style lathe would have the advantage of making box tools unnecessary, making tooling cheaper. It would likely be much more sensitive to variations in stock diameter though which may make it impractical.

More to come as I continue sketching and researching!

[SCzEngrgGroup]

If I can find a matching ball bearing then I will move the roller bearings closer together and then add the ball bearing toward the rear of the spindle (with a spacer between it and the roller bearings) so that the ball bearing takes most of the side-load and the roller bearings are left with the thrust loads.

No matter what you do, the bearings at the business end of the spindle are going to carry the bulk of any side-load, and the one at the other end will be lightly loaded, unless you're working pretty far away from the chuck. In fact, most of the time, the side load on the rear bearing will almost certainly be dominated by the belt tension.

You might also consider using a pair of angular contact bearings at the "front", and a radial or deep groove bearing at the "rear". This is a common configuration for spindles.

I'd suggest putting a "labyrinth seal" on the front, to keep swarf and coolant out, then use shielded bearings, to help keep the grease in place. Positive pressure really should not be required.

I would second the suggestion to find an existing spindle, it not the whole headstock. It'll be cheaper, quicker, and very likely better, than something you could build yourself. Even some of the Chinese ones are not bad.

Regards,
Ray L.

[Winnfield]

Wait for a Dunham 5c spindle block to pop up on eBay. I have bought several for 400-600. Building something will be that expensive anyway, and the dunhams have AC bearings, labyrinth seal, and a 4 degree Hardinge spindle taper for chucks. You won't regret it.

[aarongough]

Wait for a Dunham 5c spindle block to pop up on eBay. I have bought several for 400-600. Building something will be that expensive anyway, and the dunhams have AC bearings, labyrinth seal, and a 4 degree Hardinge spindle taper for chucks. You won't regret it.

I'm sure that buying something that's known to work would definitely make life easier... But life's a bit too easy at the moment

I would like to learn more about this kind of machining, and about machine design in general. I use a lathe almost every day so I'd like to teach myself as much as possible the important principles behind their precision.

Worst comes to worst I end up with an expensive paperweight... Hopefully that paperweight will have taught me a lot by the end.

[Winnfield]

I'm sure that buying something that's known to work would definitely make life easier... But life's a bit too easy at the moment

I would like to learn more about this kind of machining, and about machine design in general. I use a lathe almost every day so I'd like to teach myself as much as possible the important principles behind their precision.

Worst comes to worst I end up with an expensive paperweight... Hopefully that paperweight will have taught me a lot by the end.

The grinding alone will cost you more than the used spindle head. These Dunham things are over 3500.00 new. You'd certainly learn a lot by taking one of them apart. They are neat. They are almost like a baby Hardinge spindle.

[aarongough]

No matter what you do, the bearings at the business end of the spindle are going to carry the bulk of any side-load, and the one at the other end will be lightly loaded, unless you're working pretty far away from the chuck. In fact, most of the time, the side load on the rear bearing will almost certainly be dominated by the belt tension.

You might also consider using a pair of angular contact bearings at the "front", and a radial or deep groove bearing at the "rear". This is a common configuration for spindles.

I'd suggest putting a "labyrinth seal" on the front, to keep swarf and coolant out, then use shielded bearings, to help keep the grease in place. Positive pressure really should not be required.

I would second the suggestion to find an existing spindle, it not the whole headstock. It'll be cheaper, quicker, and very likely better, than something you could build yourself. Even some of the Chinese ones are not bad.

Regards,
Ray L.

Hey Ray!
Thanks for chiming in!

Looking through some bearing catalogues I've discovered that there's a lot more options available if I go with metric sized bearings... Imperial bearings seem really sparse in comparison. It seems like if I got up to 65mm bearings I have good options in deep groove, angular contact and tapered roller all for under $60 or so per bearing.

I understand that I would likely get a better/easier/quicker spindle if I was to buy one pre-made, however I am interested in learning more about precision machining and the 'origins' of machine precision, so I would like to go ahead with building a spindle...

I'll have a look at labyrinth seals!

[TiagoSantos]

For the spindle, you might spend some time on eBay looking for a larger lathe spindle and bearing set. I remember finding many Logan, Clausing, etc spindles with matched bearings for pretty cheap. Probably as cheap or not much more than buying a new pair of bearings!

[aarongough]

For the spindle, you might spend some time on eBay looking for a larger lathe spindle and bearing set. I remember finding many Logan, Clausing, etc spindles with matched bearings for pretty cheap. Probably as cheap or not much more than buying a new pair of bearings!

Thanks for the suggestion! I'll have a look on eBay and see what I can find!

[hoss2006]

Ebay is where I found my Logan 11 inch spindle with 5C collet adapter.
Make sure what you look for is capable of using 5C, the Logans smaller than 11 inch can't.
Here's some specs on Logan lathes.
Logan Lathe Model Number Table
Hoss

[wizard]

Here I'm thinking a replacement spindle from Grizzly or one of the other dealers. Of course is you want 5C support you would have to source a larger spindle. Another popular approach is a 5C indexer spindle for a very light lathe.

Of course you can buy lathe spindles as block assemblies that you can bolt to your machine. Unfortunately such spindles cost more than an imported lathe. However you can always look for a used spindle assemblies.

If you wish to DIY it isn't totally impossible you can get good results buy machining any tapers after final assembly. As for your design; for a 5C spindle you probably can do just as well with Angular contact bearings up front and a standard ball bearing at the rear. Ball bearings would allow for higher speeds which if you expect to do a lot of collet work could be advantageous. As for bearings don't get extremely uptight about having the same size at each end. In fact it would be better to retain each bearing separately. Also the front bearings need to be close to the spindle nose as reasonable. The front bearing group would be retained in place in the spindle housing while the rear ball bearing is free to move.

For the DIY approach there is lots of info on the net. Larger spindles and housings are harder to do well. Of course this depends upon how well your shop is equipped. Further the housing is just as import as the spindle itself. Review as many designs as possible until you reach the point of enlightenment.

[aarongough]

Here I'm thinking a replacement spindle from Grizzly or one of the other dealers. Of course is you want 5C support you would have to source a larger spindle. Another popular approach is a 5C indexer spindle for a very light lathe.

Of course you can buy lathe spindles as block assemblies that you can bolt to your machine. Unfortunately such spindles cost more than an imported lathe. However you can always look for a used spindle assemblies.

If you wish to DIY it isn't totally impossible you can get good results buy machining any tapers after final assembly. As for your design; for a 5C spindle you probably can do just as well with Angular contact bearings up front and a standard ball bearing at the rear. Ball bearings would allow for higher speeds which if you expect to do a lot of collet work could be advantageous. As for bearings don't get extremely uptight about having the same size at each end. In fact it would be better to retain each bearing separately. Also the front bearings need to be close to the spindle nose as reasonable. The front bearing group would be retained in place in the spindle housing while the rear ball bearing is free to move.

For the DIY approach there is lots of info on the net. Larger spindles and housings are harder to do well. Of course this depends upon how well your shop is equipped. Further the housing is just as import as the spindle itself. Review as many designs as possible until you reach the point of enlightenment.

I've managed to find some inexpensive larger bearings, which means I can move the bearing mounts closer to the spindle nose.

I'm essentially thinking that there will be a 'stack' of parts and bearings that go like this (from back to front):

lock nuts - drive belt - spacer - deep groove ball bearing - spacer - roller bearing - roller bearing

The entire stack would be pre-loaded by the lock nuts, but there would only be shoulders for the roller bearings... The deep groove bearing would be a tight slip fit and would be free to move when the spindle expands. Does that make sense? Does it sound reasonable?

I'm definitely going to be doing more research on spindle design over the next few weeks...

My shop is not equipped very well right now... But I expect to be taking delivery of a new lathe and mill within a month or so. I've put in my time making/selling pens, now I get to reap the benefits of all the crazy hours

[SCzEngrgGroup]

You might also want to go to the ArtSoft forum, and search for a thread from a couple of years back by user Simpson36, who has built several very high performance 4th axis units from scratch. He now sells them. He learned a LOT about building high quality spindles, and his adventures could save you a LOT of time, money and effort. It's a LOT harder than it appears. I'm sure he has a thread on CNCZone as well, but I don't know where it is, or what his username is here. Steve's a VERY sharp guy.

Regards,
Ray L.

[elfrench]

If I were building a lathe, I'd do it like this: Stepperhead

[aarongough]

If I were building a lathe, I'd do it like this: Stepperhead

Looks interesting! I like the idea of having a dedicated MPG for each axis...

[aarongough]

You might also want to go to the ArtSoft forum, and search for a thread from a couple of years back by user Simpson36, who has built several very high performance 4th axis units from scratch. He now sells them. He learned a LOT about building high quality spindles, and his adventures could save you a LOT of time, money and effort. It's a LOT harder than it appears. I'm sure he has a thread on CNCZone as well, but I don't know where it is, or what his username is here. Steve's a VERY sharp guy.

Regards,
Ray L.

Sounds awesome! I will definitely have a look tomorrow!

It's worth noting that I don't think this will be easy... I expect it to be a hard slog. But honestly that's actually part of the appeal. I'm ok at running a lathe now, but I expect I'll be much better when I've managed to successfully build a spindle!

[aarongough]

You might also want to go to the ArtSoft forum, and search for a thread from a couple of years back by user Simpson36, who has built several very high performance 4th axis units from scratch. He now sells them. He learned a LOT about building high quality spindles, and his adventures could save you a LOT of time, money and effort. It's a LOT harder than it appears. I'm sure he has a thread on CNCZone as well, but I don't know where it is, or what his username is here. Steve's a VERY sharp guy.

Regards,
Ray L.

Thanks for the great link there! As soon as I saw the thread I realized I had seen his work before... I've been watching Steve's YouTube channel for a while!

[aarongough]

I was thinking today and I have decided that if I'm going to do a small spindle I may as well make it a 3C spindle rather than an ER... For me a 3C spindle would actually have some utility, so on the off chance I actually pull it off then I will be able to use it.

I spent some time this afternoon drawing up a concept:



(excuse the ****ty ball bearing drawings, apparently I don't have a compass at home and had to rough it in by hand)

The drawing is pretty much 1:1 scale, but I have avoided dimensioning it as I have drawn it without having the roller bearings in front of me for proper measurement.

The bearings in the spindle nose are both Nachi 32007 tapered roller bearings ($21 each), the bearing at the rear of the spindle is a 6207-RS sealed deep groove ball bearing (~$8). The lower bearing costs and decreased material cost of the smaller spindle means it's not the end of the world if I screw things up.

I also chose the rear bearing with a larger OD than the nose bearings so that the parts of the headstock bore that need to be close tolerance can be made as short as possible... If all the bearings were the same OD then the entire rearward bore of the headstock would have to be machined to the same close tolerance.

From looking at the bearing specs the spindle should be fine up to about 8000 RPM... Unlikely that I would run it that high though, more like 2500 RPM, we'll see.

Any thoughts on the design as it stands?

I'm not going to be starting on this for at least another few weeks as I have to wait to take delivery of the new lathe.

[wizard]

I was thinking today and I have decided that if I'm going to do a small spindle I may as well make it a 3C spindle rather than an ER... For me a 3C spindle would actually have some utility, so on the off chance I actually pull it off then I will be able to use it.

3C is very small and of course ER could imply a number of sizes. So I'd have to say which is better here is highly debatable.

I spent some time this afternoon drawing up a concept:

If you where doing angular contact bearings you could do them back to back and have the bearings seat against on shoulder and be clamped at the other end.

(excuse the ****ty ball bearing drawings, apparently I don't have a compass at home and had to rough it in by hand)

A pocketful of change has solved that problem.

The drawing is pretty much 1:1 scale, but I have avoided dimensioning it as I have drawn it without having the roller bearings in front of me for proper measurement.

The bearings in the spindle nose are both Nachi 32007 tapered roller bearings ($21 each), the bearing at the rear of the spindle is a 6207-RS sealed deep groove ball bearing (~$8). The lower bearing costs and decreased material cost of the smaller spindle means it's not the end of the world if I screw things up.

Always an important point in the learning process.

I also chose the rear bearing with a larger OD than the nose bearings so that the parts of the headstock bore that need to be close tolerance can be made as short as possible... If all the bearings were the same OD then the entire rearward bore of the headstock would have to be machined to the same close tolerance.

There are a number of ways to deal with that issue. If the nose bearing where seated in a pocket what happens at the other end of the spindle doesn't matter.

From looking at the bearing specs the spindle should be fine up to about 8000 RPM... Unlikely that I would run it that high though, more like 2500 RPM, we'll see.

That is assuming you follow lubrication recommendations, achieve the same level of precision in the machine as they will have and that you assemble everything like a surgeon.

Any thoughts on the design as it stands?

The big issue as I see it is getting the shoulders that the outer races butt up against parallel. It would be a critical feature for your design. You might want to reconsider that arrangement depending upon how well equipped your machine shop is.

I'm not going to be starting on this for at least another few weeks as I have to wait to take delivery of the new lathe.

This should give you plenty of time to research ideas. Spend some time on the various bearing manufactures sits to down load app notes and design materials. There is a wealth of stuff out there to hone your design with.