Press Fit

[slow-poke]

Hi,

First post...

I'm an electrical engineer that tinkers with mechanics as a hobby. I'm adding CNC capability to my 10x22 lathe.

I have a ballscrew for the Z axis. It's long enough to allow full carriage travel, however if I could extend the not threaded floating end by 5" it would make mounting the floating end much easier. The captive end is mounted in a bearing block with a lock-nut that prevents movement along the length of the ballscrew. So the floating end just floats.

The shaft size at the floating end (reduced diameter stub) is 15mm in diameter to match the bearing in the floating end bearing block.

Plan is to make a 5" stub starting with a 20mm steel round I happen to have on hand and boring a 15mm hole (interference fit) to mate with the 15mm stub on the end of the ballscrew, and then machine the other end of the round to 15mm to match the bearing in the bearing block.

Plan is to put the ballscrew in the freezer, and the new 5" stub in the oven and then quickly tap the two together to hopefully lock the new stub to the end of the ballscrew.

Question #1 how much under size should that 15mm hole be to; be tight enough to lock the new stub to the ballscrew but not be so tight that the frozen ballscrew and heated stub won't go together with a few taps of a brass hammer?

Question # 2 in a perfect world the new stub will be perfectly aligned with the ballscrew, but this is not a perfect world so I'm thinking I should probably machine the 15mm end of the new stub after it's mated to the ballscrew even though it would be easier to do it before hand. Thoughts?

 

[Stress_Eng]

The amount of interference needed I think will be defined by what type of loading it's going to see. What are the operational loads? For this type of joint, you can calculate the radial interference pressure seen at the contacting surfaces. This and friction are the only things that will be holding the joint together. There's two types of loading you'll have to consider, your operational axial loads (tension / compression) and the loading seen during machining (torque due to cutter contact). This type of interference fit joint will result in the extension stub cylinder region seeing a constant build stress. The cylinder region can be assumed fixed at the base (bottom of hole) and will be forced open radially by the interference fit, thereby putting the inner surface of the cylinder at the base of the hole into tension.

 

[slow-poke]

I thing there will be very little load at this joint because the other end of the ballscrew is the driven end and that end is captured in a bearing with thrust surfaces on both sides of the bearing. The stub end is just along for the ride and sits in the floating side bearing (no thrust surfaces on the stub end, at the far end of the lathe.

 

[dvd]

I think your 20mm bar does not have enough wall thickness to get a good interference fit, but the worst that would happen is that you will create a new problem to solve and learn from. Have a look at this ISO 286 fit info summarized in a Flender catalog to get an shrink fit with no key - as indicated you would need to calculate the hub stress.

Edit - that sucks, the shrink fit classes H6 and u6, v6, x6 are not listed in that literature. It woud require finding that data.

 

[TugboatEng]

I second using ISO 286 as a reference for fits. It's freely available on most bearing manufacturers' websites.

For assembling interference fit items, I'll set a bore gauge to 0.005" over the shaft diameter. Heat until the bore gauge passes freely and then assemble. A temperature differential of around 180 degrees F is also sufficient for assembly of press fits. The gearbox techs all spit on their couplings. If the drop bounces around on the surface you're ready for assembly.

Also consider a clearance fit and retaining compound if your wall thickness doesn't support an interference fit.

 

[3DDave]

Far easier to find a longer screw and reduce the length and make the end feature.

Other than that, tough to guess when nothing else about the installation is known.

 

[rb1957]

yeah, a 20mm dia shaft with a 15mm dia hole is only 2.5mm wall ...
if you go this way, be vewy, vewy careful ... I'd machine the shaft, and measure carefully, and then cut the hole to suit. a 15.00mm shaft may want a 13.00mm hole. IDK the ISO spec, maybe it tells you this ?

but what about a cross-bolt (instead of thermal interference) ?

 

[mfgenggear]

Not feasible, and possibly dangerous.
Buy a cnc lathe. Prices are resonable.

 

[slow-poke]

Not feasible, and possibly dangerous.
Buy a cnc lathe. Prices are resonable.

Dangerous, please explain?

That's reasonable not "resonable"

Original post states tinker and hobby, I don't find your suggestion practical or helpful. The cost of a CNC lathe is not particularly reasonable for most hobby type use.

I converted my knee mil to CNC in 2024 for peanuts and it's accurate to 0.0004"

 

[TugboatEng]

Maybe the danger is that if the joint separates during an operation you're going to lose the stock.

As for some off your questions, even if it isn't perfectly true, a little radial run out isn't going to have much effect on a flexible shaft that is primarily positioning axially.

Most lathes have some type of weak link in the feed screw. It's not mean to generate tremendous force.

 

[mfgenggear]

Dangerous, please explain?

That's reasonable not "resonable"

Original post states tinker and hobby, I don't find your suggestion practical or helpful. The cost of a CNC lathe is not particularly reasonable for most hobby type use.

I converted my knee mil to CNC in 2024 for peanuts and it's accurate to 0.0004"

From your post it is 10 inch swing x 22 inch length , probably 10 horse. Cutting in the z axis.
What is the produce torque required.
Very small diameter lead screw.
I am a little rusty on lead screws.
So I will get AI help.

Maybe I over reacted but I am an OG who had been in sheet metal and machine shops .
I use to turn a engine and turret lathes.
With power feeds. The torque and hp required
To feed , and take depth of cut is impressive.

I seen many accidents from industrial accidents.
And I cringe when someone alters a machine.
The because of the size of the lead screw it maybe at the limit axial and bending forces.
Maybe cause even at slow feed , for it to break and whip around. Can you take pictures.

The best bet would be to have a shop reverse engineer the lead screw. Take hardness test.
It's good possibility it is a case harden for wear.
Or premium steel. Surface wear qualities.
A lead screw can be ground from solid like a thread.

 

[mfgenggear]

Great info

Lead screw sizing in 4 steps

Lead screw sizing is a bit of a hybrid exercise, with similarities to the process of ball screw sizing, but incorporating some of the considerations that come with sizing a plain bearing.

www.linearmotiontips.com

 

[slow-poke]

From your post it is 10 inch swing x 22 inch length , probably 10 horse. Cutting in the z axis.

It's a hobby lathe 5" swing, 1HP

Thanks to everyone for their help.

So I made the stub, bored it to ~0.0005" under, I think it was probably closer to 0.0007" hard to tell because I was using telescoping gauge to measure bore and even after heating the stub and freezing the ballscrew it took a few raps with a brass hammer to get it all the way in. I put a few drops of Loctite 609 on the joint, so between the Loctite and the interference fit it think it's pretty solid. There is no power or force displaced across this joint, this end is simply rotating in a "floating bearing" at the far end of the lathe where the tailstock is.

Final step is to trim the stub length and turn the end to match the bearing, but that's easy.

 

[IRstuff]

Note also, changes to the lead screw can alter its resonance frequency and its vibration.

 

[mfgenggear]

OP
What is the pitch run out near the assembled joint. While it is in the bearing.
It should .0015 or less.

 

[slow-poke]

OP
What is the pitch run out near the assembled joint. While it is in the bearing.
It should .0015 or less.

<0.001"

 

[slow-poke]

Note also, changes to the lead screw can alter its resonance frequency and its vibration.

Good point, I'm using a (free) donated Yaskawa servo, that will be mated directly to the ballscrew. The Yaskawa tuning software has some anti-resonance capabilities, so if there is an issue hopefully it will help. I have had zero issues with resonance on my CNC converted mill that also uses servos. My now sold first mill that I also converted used steppers now those things sang like a drunken choir of miss-fits. No more steppers for me, I'm 100% ac servo. these things are next to silent.