Rotating Ballnut - design ideas

[Andrew Wilding]

Hi Jonathan,
If you would like some ideas or a rotating nut you can look at my mill build log. I have just uploaded some pictures showing the assembly of the rotating nut. I used 2 AC bearings and preloaded with a nut. I bought a double AC bearing from EBAY but was not happy with the amount of clearance between the balls and the race so I resorted to 2 seperate AC bearings. The nut is not a lock nut but the pulley below has a grub screw over the key. The nut stack is preloaded then clamped with the grub screw. If this causes problems the nut is thick enough to slit radially and add an M3 clamp screw. Hope this gives food for thought

[Jonathan]

Hi Andrew

If you would like some ideas or a rotating nut you can look at my mill build log.

That did get my attention - I was going to post but hadn't finished thinking!

I have just uploaded some pictures showing the assembly of the rotating nut. I used 2 AC bearings and preloaded with a nut. I bought a double AC bearing from EBAY but was not happy with the amount of clearance between the balls and the race so I resorted to 2 seperate AC bearings.

It is interesting what you say about the bearings. I was indeed intending to get them from eBay, two of these to be precise:

http://cgi.ebay.co.uk/5208-ZZ-DOUBLE...item53e20e2fe6

I'm hoping if I press fit the bearing onto the shaft that should expand the inner ring, and tighten the bearing up nicely. Similarly with the outer ring, except if I follow m_c's excellent suggestion of boring it slightly oversize and clamping it then that's not going to happen. Did you get the single row angular contact bearings on eBay, and what size are they?

The problem I've got is that I don't have the constant force applied to my bearings due to gravity, unlike on the various milling machines that have a rotating nut.

I'm really not at all sure what to do now. I've said already two single row angular contact bearings is a lot of money...and yet it looks like that is certainly the best way to do it. Having said that there's someone on CNC zone who has done it with two standard deep groove bearings.

If I could get away with a 16mm. 10mm pitch, ballscrew then I would save so much on the screw, that I could easily afford the cheaper smaller bearings. However I would prefer to have the 25mm screw just in case it fails completely and I have to spin the screw.

[Andrew Wilding]

Here is a rambling collection of my thoughts some maybe relevant most probably not!

The bearings I purchased were from Bolton Bearings and had a 30mm bore (too small for you?) and were about 8 quid each. As I am sure you are aware that bearing is not an ac bearing and probably not a deep groove bearing. Deep groove can take some axial load but a normal bearing is quite limited in this respect. The bearing shown will not have any preload but as you say you could acheive this by pressing the bearing onto the shaft (nut). You will need a fairly close tolerence shaft to control the fit and you will need to haver a fair idea of the clearance in the bearing to work out how much of an interference fit you will need. If you are pressing up to a shoulder than you will not be able to remove the bearing. Mine are a push fit (couple of tenths interference) which can be removed without wrecking the bearing. Compressing externally would give you more control and would allow you dismantle the assy.

The real solution is to try to avoid using a rotating nut, particulary when you are talking about 40mm bore bearings and all the associated pulleys etc that go with it. I expect this is a last resort as you have already looked at other design solutions.

I would be surprised if you are getting anwhere near the design axial loads of a suitable AC or deep groove 40mm bore bearing in your arrangement. The load figures may be based on L10 life rather than ultimate limit, or the bearing you were looking at is not designed for axial loads.

The worst thing for bearing life is not having enough load and the balls sliding rather than rolling and so preload is sometimes used for this reason as much as combatting lash.

[Jonathan]

I've machined plenty of parts with bearings before, so I'm not worried about getting the required accuracy. The lathe I've just bought should help there too :)

30mm is too small I think, on the drawing currently I've bored the shaft 27mm which does not leave much for the bearing.

How do you tell if the bearing is classified as 'deep groove', or is it somewhat arbitrary? The standard bearings I had selected were the ones with a bigger outer diameter (72mm), though not the biggest. I've got 2 good quality, FAG and SKF if I recall correctly, 45x85mm bearings.

I'm not entertaining a belt or rack and pinion drive so the only other option as far as I'm aware is to tension the ballscrew sufficiently to stop whipping. That does not help with the inertia of the screw though, which for 2000mm is similar in magnitude to the other forces.

[Andrew Wilding]

deep groove is usually in the description of the bearing. On closer inspection of the diagram that you linked the bearing you have shown probably is a deep groove. These are the norm but still not ideal for use in an application where relative to radial load the axial load is significant.

Some more thought (sorry if all this has been covered in other threads)

Moment of inertia is proportional to diameter squared so I am surprised that there is much difference between the shaft and nut? I am sure you have done the sums.

Is the moment of inertia significant when compared to the mass of the 'carriage' (when calculating acceleration)?

If you have to tension the shaft to raise its critical frequency then will have to start considering beefy shaft support bearings, I can now see why the rotating nut idea is appealing!

a 2000mm leadscrew poses a lot of design challenges even when cost is not an issue (which I assume is, if you were as poor as I was when I was a student!) have you considered other methods with feedback or your real accuracy requirement? One solution maybe to use a rack and pinion/belt and calibrate the travel using a dial indicator and blocks of a known dimension. The smaller the increments the better idea you would have of variation along its length. Even a crude survey (100mm steps?) would give you significant improvement in accuracy. Obviously this would require a decent home switch.

I look forward to seeing your solutions. This is not an easy project too keep in budget but will be satisfying when cracked.

[Jonathan]

Moment of inertia is proportional to diameter squared so I am surprised that there is much difference between the shaft and nut? I am sure you have done the sums

The problem is mass is proportional to the radius squared, so you end up with 4th power. Hollow ballscrew would be nice!

Some more thought (sorry if all this has been covered in other threads)

Keep thinking! It's good to have it all in one place.

Is the moment of inertia significant when compared to the mass of the 'carriage' (when calculating acceleration)?

Yes it is, I'll post the numbers when I've verified them.

If you have to tension the shaft to raise its critical frequency then will have to start considering beefy shaft support bearings, I can now see why the rotating nut idea is appealing!

Just what I was thinking. I suppose tapered roller bearings would do the trick, but then you're adding friction.

For a rotating nut have you considered how much the timing belt tension would deflect the nut on a 2000mm length <25mm shaft and the effect of out of balance forces of the nut on the shaft?

Surely the bearings are going to take the radial load and stop the screw deflecting, or am I misunderstanding you? So as long as the ballnut is held on centre I should be ok. That could be an issue if the flange on the ballnut isn't concentric. Another issue I've found is that at sufficiently high rpm with a standard ballnut centripetal force will stop the balls rolling properly.

a 2000mm leadscrew poses a lot of design challenges even when cost is not an issue (which I assume is,...calibrate the travel using a dial indicator and blocks of a known dimension....survey (100mm steps?) would give you significant improvement in accuracy.

I think that method of measurement would have a cumulative error, which is especially significantly over this distance. It could maybe be done with a digital calliper and carefully fixing to consecutive points along the bed. Theoretically the process could be automated.

Obviously this would require a decent home switch.

Easily done with a cheap laser pointer I reckon.

[Andrew Wilding]

Jonathan,
You can ignore the stuff about critical frequecy/deflection of the screw with the spinning nut. I had a dim moment and forgot you were supporting the nut with bearings. Its late or at least that is my excuse!:cry:

Agreed with the 4th power bit but that would make the nut inertia even larger than the screw. How does moment of inerta or rotating bits compare with mass of linear bits when you look at a good old newton second law equation?

going tp bed now.