My “Elliot mini mill” did not produce a smooth finish when taking light facing cuts with an endmill. The machine’s spindle had 0.02mm runout measured inside the spindle taper so I took the bold step to take it apart. I did not measure axial play which I should have done in hindsight. The bearings were in a very poor condition. The upper was grinding and the lower pair full of grainy gunk that must have built up over the years.
It has an R&M LJ25WRR 25x52x20.6 deep groove sealed ball bearing at the top of the quill and a pair of open R&M angular contact bearings mounted back to back on the nose marked LP25 S7P 25x52x15. According to lathes.co.uk the machine was fitted with a matched pair of AC bearings with a preload of 50lb. I could not identify any “matching” grinding on the races but the inner races measure 0.03mm narrower than the outer races.
I managed to identify a modern replacement pair of universal bearings with the excellent help of Gareth at Orinoco bearings in Leeds. They do however cost more than I paid for the machine itself. After intensive internet research on the subject of matched AC bearings and preload I decided to order an unmatched pair of old stock Hoffman bearings just to try to get it going again.
I will firstly mount them on a slightly undersized mandrel and compress the inner races together to assess if there is any preload / deflection present on the outers races. If there is play I intend to shim the outer races to provide some preload before mounting them on the spindle.
Have any members replaced bearings on this type of machine? Which bearings did you use? I would also appreciate any comments on the subject of preloading unmatched AC bearings.
I’ve no idea how much bearing preload your Elliot / Downham jig borer might have used. So for whatever this is worth? Because of there much better accuracy to meet that jig borer designation, some jig borer manufacturers and Moore Tools in the U.S. in particular used some higher than usual bearing preloads. Very high on those Moore machines in fact, and with a known shorter bearing life span because of it. Moore obviously thought the trade off was worth it. Only my best guessing, but you might have to try multiple times before you find the correct setting. Checking the heat produced at the higher rpms by hand over an hr or so and touching the outside of the head casting might be your best indicator for when that preload is too high or low if there’s almost no heat being produced at all.
I’d also guess that 50 ft. lb number should be at least a decent starting point. And even non matched replacement bearings would obviously make a large improvement over what you pulled out of it. One thought and it’s hard to say if they would be cooperative or not, but a few phone calls to some proper spindle repair or rebuild shops might get you some better information than any of my guessing does. I do know enough to say for sure that no impacts at all are ever allowed when replacing spindle bearings. Gentle use of a press only in case you don’t already know it. I’ve read that bearing brinelling can start on very high precision bearings with as little as 4-6 ft. lbs of impact force. Probably more than that with your bearings, but I don’t know that for sure. And almost surgical cleanliness for everything touching or anywhere near those new bearings. Sorry I can’t be of more help.
Ivan, if the pair of AC bearings in your assembly are actually preloaded by spring force, you don’t need bearings with any kind of face adjustment. (Face adjustment is only required when bearings are mounted in direct contact with each other OR separated by equal length spacers.)
If you have stripped the spindle, the presence of spring-preloading will be obvious as you will have found either wavy washers, or in more sophisticated spindles, a sliding abutment with a series of coil springs built into it. If this is the case, then the ‘special’ features of your bearings may just be limited to indications of where in the tolerance bands (i.e. upper, middle or lower) these particular rings are, and the positions of max runout. Neither of these things are particularly important in my mind, in view of the vast improvements in manufacturing consistency of todays bearings compared with those made up to the early 1970s when the R&M name would have been replaced.
The ‘matching’ of bearings was probably not done by machining them. Rather it would be the bearing manufacturer making millions of them and after testing selecting ones that matched.
There is a film around I think of the way Colchester lathe bearings were made to such high quality – each ball measured and sets made up of matching ones……
With close coupled bearings like that the really important thing was concentricity. With a decent spacing half a thou error between the pair wouldn’t matter that much. Not that you are likely to find anything that bad from a decent brand. Clamped up as a close coupled pair any error will put nasty loads on the bearing.
Modern production processes are much better and more reliable than in days of yore so shouldn’t be any issues given a pair of premium quality bearings off the same production line. It’s only very high end applications where the top precision classes are really needed. Even the at the modest speeds our machines generally run at really precise as against good premium quality shows up more in machine life than performance. We will never get the hours in to notice.
I found this informative video on YouTube that provides tips on rebuilding a spindle and how the matched pairs are pre loaded. It’s quite long but I’ve watched it twice skipping the parts about grease massaging as the Eliot uses oil. No impact and cleanliness is highlighted as being very important and I will try and follow it. https://www.youtube.com/watch?v=grUdsTTRGl4
I agree that fitting the unmatched pair as is will improve the condition compared to the worn bearings and in my use case of running the machine around 2 hours per week will suffice.
The old bearings did not have concentricity markings and neither do the new ones. I don’t have a way to check it either so concentricity has to be ignored.
Regarding preload:
– The preload is applied by the axial clearance between the bearing pair inner and outer races. pushing the inner races together axially clamps the outers together. There are no spring washers in the system.
– To assess this I made a mandrel with a clearance to the inner races. Then compressed the inner races of old and new bearings (unmatched 7205). There is no discernable difference in resistance when rotating the outer races. I was also able to rotate the outer races independently with a strong hand grip.
– The lower bearing pair is pushed up into the quill with the lower spindle bearing nut. This is how the spindle is located in Z in the quill. Here is a very basic sketch of how it looks and how I think the force path runs through the system. The top bearing inner race is pulled down by the upper nut. This compresses all inner races. I suspect the top bearing outer race is floating in Z and not pushed fully into the quill rebate as my sketch suggests.
– All the info I found on forums suggest that some preload is required and that unmatched pairs have to be shimmed if used in a back to back arrangement. I will follow this suggestion. I have found some 0.01mm 304 shim stock and will try to fabricate a shim for the outer races and try this on the mandrel first before I re assemble.
This is taking a bit longer than expected and I will update my findings.
Ivan, thanks for making the sketch which helps a lot.
Regarding the pair on the left, if you can rotate the outers relative to each other after clamping the inners tightly, then there is no preload in the unit. As you suggest shims between the outers will create this. You should aim to get the unit about 10microns tight, so depending on how much clearance you actually have, you may need need one or several shims. The normal way to decide the correct shim thickness would be to measure the end-float and add shims to that value, PLUS 10 microns.
The problem with putting shims between the outer is that they are not very flat and this will make the process of determining/adjusting the preload a little hit and miss. Another and hopefully more accurate way to adjust the unit would be to disassemble one or both ACs and rub the appropriate inner-ring faces on a surface plate with fine carbide paper on it. A micrometer will tell you how much material you have removed.
Regarding the single bearing at the other end, this looks a little odd, at least as its drawn. IDEALLY there should be a light wavy washer between the outer-ring face and the housing abutment to ensure that the radial play in this bearing is removed. This would improve the rigidity of the spindle a little, but you could get away without it. Without a wavy washer though, the abutment face should be clear of the outer ring face. If its not, the uncontrolled face-protrusions on this bearing could, if you are unlucky, interfere with the preloading of the pair at the other end.
Preload AC bearings in the spindle nose. (LH side of sketch)
– There is some preload when mounting the bearings on the test mandrel.
– You have to grip the outers very firmly to be able to rotate them independently. The amount of force required felt similar between the old and new pair. I was surprised by this as the width of the inner and outer race of one of the new bearings measured the same.
– Before attempting the shim between the outer races today I measured both new bearings and found that the inner race is 0.06mm narrower than the inner on the other bearing. The bearings doesn’t look like they are from the same supplier.
– I plan to mount them, without a shim, despite them being from different suppliers.
Preload on deep groove bearing at the top of the spindle (RH side of sketch)
– I did not consider that this bearing should also have some axial pre load that could be provided with a wavy washer, thanks for explaining that. There was none present when stripping the spindle. I assumed the bearing was floating in Z and not making contact to the abutment in the quill.
– I measured the quill and the spacer that is fitted between the upper and lower bearings. The distance between the abutments in the quill is 0.2mm larger than the length of the spacer. here is a new sketch. The top bearing is on the LH side in this case. I should have drawn it in the same orientation as sketch 1. Hindsight is 20/20.
– This means the both the outer races on both the bottom and top bearings are making contact with the abutments. This will place significant axial preload on the top bearing. Could it be that this is designed to stretch the spindle?
I could fit it back as is or introduce another spacer to enable a wavy washer to be fitted under the top bearing.
I should have mentioned this earlier but the widths of bearing rings is cannot be used as an indication of face adjustment, preload or clearance. There is a raceway to face tolerance in there as well so a bearing with identical ring widths could still have significant offset (ie a ‘step’) over the faces of the inner and outer rings. The only way to measure this ‘step’ is to put a very light axial load on the bearing (in the appropriate direction) and measure it directly.
Regarding being able to turn the outer-rings of the ‘preloaded pair’, if the pair was properly preloaded I don’t think you could do this as there would be too much friction. So I’m wondering if in fact the inner-ring faces are not properly clamped when you did this test ? This may explain why it happened with two different set pairs of non-face adjusted bearings Were the inner rings clamped against the shaft abutment by a nut? …. just looking at your earlier picture again it looks like you have clamped them with a torx screw…Was it very tight ?
Is it possible to make a partial build with just the pair clamped on the spindle and in the housing ? If so you could check for axial play in the spindle with a DI or, alternatively, in the bearings are preloaded, check that the torque in the spindle is not of a ‘stick-slip’ variety that would indicate excessive preload. I think its important to try to check this. Excess preload will make things hot, clearance will give you a poor finish.
SINGLE BEARING
I’m not clear on your measurements here. You give 114.4 and 144.27 but say the difference is 0.2 Am I misunderstanding something? Should 114.4 be 144.4 ? If so this would be a difference of 0.13 rather than 0.2 If the diff. is 0.13, this is still very high and with the long spacer being shorter than the outer abutment span this would be bad. Can you clarify A-B and C-D.
– I used a m8 cap screw and tightened it with a 150mm hex key as hard as I could. It was very tight. I used a 2mm mild steel washer to compress the inners and it dished. I will turn a thicker washer, get a high tensile bolt and test again on the old pair using a torque wrench to see at which torque the outers stop rotating.
– When I extracted the original bearings the lower outer race disengaged from the inner and the balls spilled out. I have read that disengaging is not good for them so I will experiment on the original pair before doing a partial build with the new set.
– I’m waiting for the oil seal to be delivered before it try the build up. The final decision I have to make is whether to shim the outer races by 0.01 or to fit them as is.
SINGLE BEARING
I made a mistake on the sketch and measured again. Measurements were taken with a digital calliper.
– Abutment faces AB 114.42mm
– Spacer length CD 114.27mm
The spacer is 0.15mm shorter than the distance between the abutment faces. Thanks for your advice that this in not good. I will address this by adding an additional spacer washer and then pre loading the bearing with a W61580 wave spring washer.
The spec for the wave spring washer is 0.56mm thick and loaded height 1.75mm. The additional spacer should therefore be at lest 1.6mm thick to prevent load onto the lower pair. Is this correct?
Ok I think you have convinced me that the inners are positively clamped :). Therefore if you are able to still just counter-rotate the outers by hand, your two bearings are, by hook or by crook, effectively flushed-faced on the inside faces (ie for back to back configuration). As you say, you just need to decide if you want to now preload them with a shim between the outers or leave as be.
If your cutting forces could be high, I would be inclined to head towards the 0.010 preload direction. I noticed you mentioned earlier that the test mandrel was undersize. When the inners are mounted on a shaft with some interference, that will induce some preload on top of what you have. (The same would be true with the outers but there should be very little interference here.) So unless you can measure bores and ODs to microns, it will have to be a little bit ‘suck it and see’.
If you do a trial on the spindle, is there a means of pulling the bearings off the shaft via the inner rings?
Single bearing.
This bearing should ideally be given an axial load from the wavey washer of around 5~10% of its Cor catalogue rating. (This is not super critical though if you are a little higher or lower.) So the dimensioning of the abutments will depend on the stiffness characteristics of your selected wavey washer. If this guideline axial load is achieved when the wavey spring is compressed to 1.75mm, then your shaft-spacer washer should be: 1.5mm + 1.75mm=3.25mm thick. (The 1.5 brings the inner and outer abutments into line. The 1.75 creates the gap between the outer ring face of the bearing and its adjacent abutment, to achieve the correct squash of the spring. Does this make sense ?).
Regarding this shaft spacing washer, it must be flat and even. Give it a good rub on carbide paper on a surface plate to get it flat and check the thickness variation with your micrometer. Aim for about 5 microns or less variation, if you can.
Ivan, quick error correction. I must have been having a funny five minutes when I said that the spring load on the single bearing should be around 5~10% of the catalogue static rating. This is too high. 0.5% to 1% should be adequate for the application, achievable with a wavey washer and wont interfere significantly with the preloaded pair at the important end.
Gerry
Author
Posts
Viewing 12 posts - 1 through 12 (of 12 total)
Please log in to reply to this topic. Registering is free and easy using the links on the menu at the top of this page.