Ballscrew Nut Re-fit

[electrosteam]

My CNC re-furbish project on a Kondia turrett mill has languished for a few years.
But, I can now devote some time to it and the first issue has arisen.

The X-axis ball screw nut was separated from the screw after some problems.
The shaft leading up the helix is Dia 25 mm, and I 'assumed' that diameter would allow the nut to simply slide off.
Not so, the nut jammed immediately.
Tried working it back and forth and eventually removed the ball return tubes and extracted the balls.
Not easy because they are heavily greased - got 103 balls.
Quite a few gouge marks in the shaft, but of no consequence because this area is non-utilized.

Intend to clean it all up and re-fit with a central oil system.
Included is new brush wipers if they can be obtained.

Previously, the Y-axis nut but was removed by sliding onto a Dia 27.2 mm tube, retained on the tube while
machining the screw, then re-fitted without any drama.

Looking for suggestions:
- source for brush wipers,
- ideas on on how to get the nut back onto the helix.

Thanks,
John.

[rickw72]

I managed to put back together a ballscrew once by cleaning everything up and then feeding the balls in through the tube hole. Keep going until they start coming out the other end, keep letting out and feeding new ones in until you think you’re letting ones out that are properly packed. The first few in won’t be well packed because you’ll be struggling to keep the screw and nut aligned while you feed the balls in. Then fill the return tube with balls and bolt it back on.

[rickw72]

I managed to put back together a ballscrew once by cleaning everything up and then feeding the balls in through the tube hole. Keep going until they start coming out the other end, keep letting out and feeding new ones in until you think you’re letting ones out that are properly packed. The first few in won’t be well packed because you’ll be struggling to keep the screw and nut aligned while you feed the balls in. Then fill the return tube with balls and bolt it back on.

[electrosteam]

Thanks for the suggestion, I wondered whether that was a viable method.
My nut has 2 return tubes, with each end buried within the nut when the retaining clamp is fitted.
It will need quite a bit of dexterity, possibly some ingenuity, but worth a try.
One could possibly fill the nut with one tube installed, fill the the other nut at the work bench, hold the balls with a grease, then invert that tube and install.
The nut will be full when all 103 balls are inside it and it revolves smoothly around the screw !

I have started looking at Youtube clips and searching for suitable descriptions.

John

[electrosteam]

Attached is one of the ball return tubes showing the damage inflicted by a previous worker (!). 380823 This tube, and its mate, would still operate but they are clearly not ideal. The other parts required are the end wipers - some wear/distortion. Got on to the Australian representative of THK ballscrews, they have no spares and claim trying to contact the factory would be pointless. Posted a thread on LinuxCNC Forum to see if anyone there can suggest a spares supplier. This post is just to see if anyone here might have a suggestion. As an aside, placed a query through Alibaba for bellows to keep swarf out of the nut in future. John

[Michael G]

What's the bore of the tube - it looks like the sort of thing that may be able to be bent around a mandrel.

Michael

[electrosteam]

Michael,
The problem with a new fabrication is the special machining at the ends - it would take multiple attempts to get it correct.

One small step forward was a quote/order on a Chinese company through Alibaba for protective bellows.
They were not too expensive, but could easily be not suitable.
Report with photos when I get them.

I found a Korean address/site that sells redundant and used electronic and associated mechanical components, including ballscrews.
They do not have an English version of their website and it appears they do a lot of business after a walk-in inspection of the goods, with prices in Korean WON.

Found on their listings a used ballscrew of the same type as mine.
Price is not too high, but so far, they have not answered my email.

Any suggestions on how to get assistance with contacting a Korean company ?

John

[electrosteam]

From the amount of views, this subject is of some interest.
Not much progress to date, stalled out trying to contact the Korean web site.

My web searching tells me the best approach to mount the nut is to pre-fill it with greased balls, then insert a suitable holder for transfer to the screw.

For the photo enthusiasts:

381026381025

Photos shows the nut with edge of a track broken away, and one of the end brushes
This nut is hard, so it was some sort of whack, or jam-up of swarf.
Perhaps it was track debris pieces that caused the scavenging effect on the ball return tubes.
Note that the fracture is on an unused part of the ball track.

So, I have a nut that has:
- lost at least one ball (2 circuits needs an even count, not 103),
- damaged ball return tubes,
- fractured edge on one track,
- has distorted end brushes,
- operated with swarf mixed with grease.
Hence my interest in pursuing replacement parts.
A new nut is about AUD 450.

Followed up with a Google Translate version of my inquiry to Korea a couple of days ago - no response.
The website is dgfa1.com: ?? ???/????? ?? ??? ?????? (http://www.dgfa1.com/)

All contributors,
Do you know someone fluent in Korean that might be interested in assisting me ?

The whole job could be done via email and telephone.
Happy to pay a reasonable fee.

Job would be some/all:
- inspect the site and translate the contents,
- determine communications channel,
- call the listed number to discuss (in Korean).
- determine the correct method for requesting a price with delivery to Sydney,
- determine payment method,
- assist in ensuring correct details submitted.

John

[electrosteam]

Never got a response from the website.

Spent a couple of weeks looking for a S. Korean speaking person to translate the site for me, and eventually found one.
Called the listed number and confirmed that they would supply quantity one, but did not deliver internationally.

My new friend suggested we could use her parents address in S. Korea for delivery, and they could forward to me.
I did not want to involve unknown people in handling a 850 mm long package weighing quite a few kg, so I did not accept the offer.

Spent some time looking for a S. Korean version of the US Shipito without success.

Then my friend discovered a S. Korean company that appears to be modelled on Shipito, but primarily delivering to S. Korean expats.
I have contacted them with a view to initiating an account, but no response at the moment.

But, when I checked the supplier's website, the desired item is no longer listed !
Too many weeks trying to get there.

So, back to Plan A - a general clean of what I have, a small amount of dressing and a re-assembly.
I hope my ball count was off and that no balls have been lost.
Cut the damaged ends of the ball return tubes back to re-establish the original tip strength.
Checked the fit in the nut and rolled balls into and out of the tube/nut interface - seems Ok.
The balls are 6.365 diameter on both my micrometers, the tubes are about 6.55 internal diameter.

I will ask my friend to call the website to confirm nil stock remaining.
And also open an account with the freight forwarder 'just in case'.

John.

[electrosteam]

Checked all the balls, 102 x 6.365 and 1 x 6.355, so at least one ball has been lost.
I will reload with the small ball in the 52 group, the 51 group being all large balls.

In addition to loading the nut onto the screw, I have the problem of positioning the ballscrew in the saddle.
The sequence is to install the Y-axis ballscrew to the saddle (simple shims), and then the X-axis ballscrew to the saddle, without the table.
Then slide the table on to engage the ballscrew bearing seat into the table endcap bearing.

Easy (relatively) to get the X-axis ballscrew parallel and level with the x-axis sideways, but difficult to ensure co-axial with the expected bearing in the endcap on the table.
The endcap has two dowels.

Presumably in the factory, with the original trapezoidal leadscrew, the endcap is offered up to engage the bearing, the endcap is bolted to the table, then the dowels added.
But I need to position the ballscrew in alignment with the endcap bearing in an endcap pre-located by dowels.

The options appear to be:
- leave the dowels off, just bolt the endcap on,
- fill the existing dowel holes and re-drill and fit dowels,
- go to extreme metrology and/or jigs to get the ballscrew in the correct location.

The perfectionist in me says I should use the jigs and metrology, the home hobbyist light duty user says 'leave out the dowels'.
The hollow ribbed design of the cast aluminium endcap prevents alternative positions of the dowels.

John

[caskwarrior]

I would leave out the dowels unless there is room to put just a small 1/8 one in just to locate it?

[electrosteam]

The table end cap has solid metal in line with the T-slots (probably to prevent damage).
I can fill the table T-slot ends and then drill for dowels after fitting the end cap and testing.

Made a transfer tube to enable desktop loading of the balls into the nut.
1214 bar125 mm long turned down to 27.1 mm swapping ends in a collett.
Held bar in the collett, equal overhang inside/outside.
Got a 3/4" boring bar at McJings and made a holder for my ancient 4-way toolpost.
Bored, carefully left to right, to 25.2 mm, so 0.95 mm wall thickness.

Made a locating bush to hold the transfer tube concentric in the nut while loading balls.
Will make a cap to cover the end of the transfer tube during ball loading.

Both ends of the screw had severe gouging on the shaft lead-ins.
These were Dia 25 mm dia when they should have been just under Dia 27.2 mm.

The screw was mounted in the lathe and lead-ins filed and emery cut.
Simply aimed for a good finish and fit to the transfer tube.

Mounted the screw to the fixed steady and used 5 mm sash cord and oil to clean/burnish the ball tracks.
Also wool thread for the groove in the helix peak and emery tape over the top.
The screw came out looking good.

New thought on ball loading - use a suitable hand-held tube and a vacuum pump to hold the ball while loading.

Checked all the balls under a 200x microscope.
Half the balls are worn, but probably re-usable.
The others either show scuff marks or pitting.
The nut has two separate circuits, so one half may have been the problem
I need new balls.

I measure 102 x 6.365 mm and 1 x 6.355 mm (last digit is midway between marks on the micrometer scale).
The micrometer measures spot-on at 5.1 mm and 7.7 mm on slip gauges.
My inspection grade DTI station puts the diameter at closer to 6.37 mm.
Probably 1/4" nominal oversized to fit.

Can anyone comment on how to assess the required ball diameter ?
And a source for 110 balls at not too high a price ?

John

[electrosteam]

Made the transfer tube end plug.

Went down to CBC in Penrith for a chat, left with 110 x 1/4" balls for under $14.
Apparently special sizes are ordered through their system and I could do better myself directly overseas.

Knocked up a simple nut holder and found a cheap (acquarium ?) vacuum pump with matching rubber tubing.

Tried one of the new 1/4" balls and did a trial (nil lubrication).
Positioned the transfer tube in the nut with its end just below the ball return tube capture zone.
Simple to pick up the ball and release it opposite the capture zone.

Worked a treat !!!!!!
Fascinating watching the ball disappear down the helical tunnel presented to it.

Will provide further update and photos tomorrow.

Plan now is to load the nut with the 1/4" balls, move the nut to the screw, and try the 'feel' of the nut on the screw.
Then re-load the nut with the old balls (+ at least one 1/4" to make up the numbers), move to the screw and re-try the 'feel'.
Depending on the results of these 'feels', a decision will be made as to acquisition of oversized balls.

John

[caskwarrior]

This thread is great thanks for all the detail.

[electrosteam]

Continued the trial fit, no need for the vacuum, just drop the ball into the vertical jigged assembly and let gravity do its job.
Can even drop multiple balls and stir with a small screwdriver !
Filled up one circuit with 54 x 1/4" balls, so the old nut may have lost 5 balls.

With the present jig, cannot get an accurate view of the fit of the last ball.
My assessment is that the gap left between the first and last ball is about 1/10 of a ball, say 0.6 mm.
With 54 x 1/4" balls, each ball could be 0.01 mm larger to fill the helical path length.
But the transfer tube is slightly smaller than the target screw minor diameter, so no point assessing fit/size at this point.

Plan to fit the nut to the screw without end brushes and circlips so that a view of the balls on the screw might be available.

Busy at the moment, but will try and get a photo of the process and result later today.

John

[electrosteam]

Loaded the nut and mounted on the screw, twice
The complete process is simple and trouble free.
The nut even balances on the transfer tube and spacer without need of a supporting stand.
Easy to finger drop about 10 balls, then fish them around with the blunt end of tweezers until they either enter the return tube or go down the helix.

First run was 108 x 1/4" balls.
Used thread wires to estimate gap between balls at 2 mm.
Feels 'loose' and the nut drops under gravity over full length of the screw.

Second run was 54 x 3.65 mm on one circuit, and 47 x 3.65 mm plus 7 x 1/4" on the other circuit.
The tread wires were marginally tighter to enter the ball gap, but still about 2 mm.
Can definitely 'feel' that the nut is firmer, but still loose in my mind.
The nut tries to drop down the screw, but is reluctant, especially near the ends, but it does rotate several turns before stopping.

Note total ball counts:
- I reported 103 originally, now 54+47+1(not used small one) = 102 (damnit),
- originally 103 (or 102) as found, now 108 loaded by me.

My judgement is that the 1/4" balls are too loose, but what size to purchase ?
US price for a 6.37 mm ball is about USD 0.72, so 110 balls is a significant purchase, have to get it correct.
Have no idea about the effect of a lubricant film on the fit, or the allowance for heat expansion.

John

[pippin88]

Do the balls need to fill the ball path length? I don't think so.

[electrosteam]

I think the ball size is determined by the fit to the screw groove, it affects the pre-load, backlash and friction factors.
The balls in the path length simply need 'working clearance', currently about 2 mm.

20 microns for lubrication plus 20 microns for heat over 54 balls is 2 mm, the whole current working clearance.
If I increase ball size by 5 microns I would probably have to drop the ball count to 53.

Note that I have to get new balls, the old balls are faulty and the 1/4" too small.

I currently think 6.37 mm balls are worth trying.
These may be too tight, then I can try alternating 6.37 mm with 1/4".
But, the risk is that the screw root diameter plus ball diamater is too great, then I would have to re-order balls at 6.365 mm.
Count will be such as to leave at least 2 mm in the path length.

John.

[Michael G]

Can you use a gauge pin to accurately size the groove? That would at least give you a maxiumum ball diameter. Not sure how loose is too loose but will some of that play go if the ends of the screw and the nut are constrained so that they can only move in one direction (that is mounted as they would be on a machine)?

Michael

[electrosteam]

Ball size measurement was discussed in a parallel thread:
https://metalworkforums.com/f65/t203957-chrome-steel-ball-measure-digital-micrometer

The size determined was a distribution:
- 1% bigger.
- 12% 6.356 mm,
- 84% 6.355 mm,
- 5% smaller.

Some sample balls from the largest group were measured independently at 6.357 mm.
I think I should take 6.356 mm as the best estimate of the diameter.

Loaded 54 balls from the 84% group into half the nut, a very neat fit.
Installed the nut on the screw and clamped the screw to my small mill table with the table clamped.
DTI on the nut showed total backlash of 25 micron when the nut pushed/pulled by hand.
Tested rocking of the nut and table movement, both negligible effects.
Moved the balls to the other half of the nut, tested as above, and got the same result.

Loaded all 102 0riginal balls plus 6 new 1/4" balls in the nut and installed on the screw.
Clamped the nut in the vice and measured the vertical looseness at both ends of the screw.
Some simple maths reveals a vertical slack in the nut of 5 micron.

As the nut has demonstrated excessive looseness, the new balls need to be increased in size.
Vertical results suggest 5 micron, the backlash suggests 25 micron.
Searched the web for hints on the effects of increasing ball size, without any definitive result.
My personal guess is that 6 microns should be Ok, and doesn't seem excessive.

If final backlash on the mill is about 20 micron, I would be happy, compensation should handle that.

The new size is estimated to be 6.362 mm, with actual size determined by the stock of balls available.
Balls are graded for dimension and range and are available from Grade 10 to Grade 100 (in micro inches).
Some Grade 25, about 0.0007 mm, are listed with the lowest price offered (USD 0.79).
Some batches have very high prices, so it is a matter of being selective on size and Grade.

Spoke to Universal Bearings in Sydney as they list balls on their website.
They only stock standard imperial and metric sizes from China.
They were vague when I mentioned certification, batch number and grade.

There is potential for this to come out badly.
I am sure about my measurements, but not sure about the additional 6 microns.
The firmness of the ball screw nut on the screw could still be loose, or too tight.

Checked price for 110 balls with Bal-tec in the US.
The price is horrendous with postage exceeding the price of the balls, total of almost AUD3 delivered per ball.
You can find Bal-tec at website "precisionballs.com".
I have contacted Bal-tec by email enquiring as to availability of cheaper postage options.

John

[caskwarrior]

I would try an SKF distributor maybe, they have a good presence here in Aus and I have bought some strange stuff through them no problem.

[electrosteam]

Michael,
My understanding is that the groove on the nut and screw is two separate curves, and each may not be circular.
Depending on the direction, the nut has single-point contact to a ball on one curve, and the screw single-point contact on the ball on the diagonal opposite curve.
Then the curves in contact swap around for the opposite direction.
It all sounds too difficult to measure or assess, I will have to go on the nut/screw assembled test results.
John

[electrosteam]

Why is it so hard !

Was always troubled by the discrepancy between the DTI and micrometers for ball diameter.
So I re-did all the measurements with two inspection grade B&S and Mitutoyo bench-mounted micrometers and adjusted my estimate of current size upwards a few microns, to 6.365 mm.

Placed an order on Bal-tec in California at 6.368+ mm for 110 balls at USD 0.79 each.

Order seemed to go through, then link across to the credit card payment facility, to have my credit card declined.

Tried two cards from one of the Big-4, and a card from a multi-national bank, all declined.
Checked in my local branch, no problems found.
The cards have 3, 5 and 7 months expiry.

I will email Bal-tec today (Monday), then call them early in the morning (currently 18 hrs).

Any suggestions ?
John

[caskwarrior]

American banks are absolutely crazy, especially when it comes to debit cards that draw from an account rather than line of credit. Just call and see if they can do it over the phone, otherwise you might have to mail a money order or use some ancient thing like western union.

[BaronJ]

Hi Guys,

You might find that this is a "prevention of money laundering" issue ! Its starting to happen here with certain cards declined. I tried to buy fuel, in France, and had my UK card declined. No explanation. Yet a few miles further on, I had no such problem. It seemed to be one of the super store chains that did this, because it only got refused at their fuel stations.

[RustyArc]

Maybe see if they take PayPal?

[electrosteam]

Some success.

Sent email to Bal-tec and one of the staff replied with suggestion I call them and ask for her.
I called and she took down my credit card details, no problem at all.

Couple of days later the appropriate charge appeared on my credit card.
No delivery documents yet.

John

[electrosteam]

Success, re-balled.

Balls arrived today in a nice parcel.
They glisten in oil in a vacuum sealed plastic bag.

They switched batches on me, but the size is the same, 6.368542 mm.
Their inventory listed a number of batches of identical size, my selection was arbitrary.
Size was invoiced at 0.25073 ", so better in future to deal with them in imperial.

Apparently for very critical applications batch tracking is important.

Delivery documents state 'certified' to meet the relevant standards.

Never got advice of tracking details.

My cheap Chinese micrometer confirms 6.365 (or so).
I will get the precision micrometers back from their owners to see what they say.
But looking good.

Difficult to get time to load the nut and trial the screw for a day or so.
(Very busy running trains for a charity event)

John

[electrosteam]

Loaded the nut with 2x54 new balls, clearance seems adequate, but way less than another ball.

Installed on screw and tested the vertical friction with the shipment oil (very light).
Still too loose for a new build in my opinion.
Not inclined to self-start from stationary, but will continue the full length if started.
My +3 microns size increase was a bit too conservative.

Mounted ballscrew on the mill table and tested the backlash.
Tested as 18 microns.
Still plenty good enough to install on the mill.

The backlash improvement follows the ball size increase.
I wonder where I would be if I had risked all and gone for a 20 micron or so increase ?

I guess that completes this thread.
Thanks to all for the suggestions and comments.
John

[rcaffin]

In my experience of servicing ball nuts, you cannot get a zero backlash result from a single ball screw nut. It isn't possible, and you should not even try. A slight temperature change and you will have lock-up. If you want zero backlash you must used a double ball nut with adjustable spacing, usually with some sort of 'spring' loading. Pretty stiff spring. Double ball nuts are stock items and are almost always what is used on a CNC. I am taking about 1-2 microns backlash here.

Yes, you reload the balls through the outside tubes. A bit of grease and a bit of a wiggle till they are all in place. You should not try to pack the grooves full of balls: leave room for a bit of play and some thermal expansion.

Incidentally, unloading the balls from a nut should ALWAYS be done over a bucket of some sort!
382999

Cheers
Roger

[electrosteam]

Roger,
Thanks for the comments.
Loved the bucket photo, I use a re-cycled food tray on the bench.
The reason I chickened out on the size increase was just the risks you mentioned.
I was chasing the "stickyness" of the nut on the screw, a few rotations before coasting to a stop after starting by hand.

It seems to me, about 18 microns backlash in the ballscrew is OK for a turret mill X-axis.
I am sure there will be more in the toothed rubber belts.

Can you provide any references for the effect of ball size increase or acceptable backlash in the nut ?
I have found it very hard to get this sort of information.

John

[rcaffin]

It seems to me, about 18 microns backlash in the ballscrew is OK for a turret mill X-axis.
Depends on what you want from the machine. My system resolution is 0.8 um, and I routinely zero it to 10 um. I can go tighter if warranted (5 um is easy enough), but that is only for some jobs.

I am sure there will be more in the toothed rubber belts.
With the older timing belts such are used on cars, yes, but if you change to the GT2 or GT3 profile there will be essentially zero backlash. And those GT2 belts are VERY strong. You just need to have the right tension on them to seat the tooth profile. That is, the European ones, not the Chinese ones. The latter may still be fairly good.

The whole subject of 5 um size changes in ball nuts is a bit obsolete today. You can try to get minimal backlash with a single nut, but it is not worth while. For CNC work we use double ball nuts spring-loaded apart, to reduce the backlash to very close to <1 um. That changes everything.

Cheers
Roger