If you find this website useful, please check out my books or visit my Amazon Author page. Or even Buy Me a Coffee!


I’ve made a petition to the UK Government to restore funding for the Zoe App – will you sign it?

I just need one or two more sponsors for the petition to go public.

Thank you

Neil Wyatt


Click this link to sign the petition:

My petition:

Restore funding for the Zoe Covid Tracking App and its development.

Reverse the decision not to fund further work on the Zoe Covid Symptom Study and the widening of its work to look at a 'whole host of other health conditions', after carrying out a proper review of the risks of not funding the work and the potential benefits.

The UKHSA has put £5M into supporting the Covid Symptom Study which uses the Zoe tracking application. The study has proven one of the most reliable and unbiased sources of data on the progress of the Covid pandemic. The technology promises huge potential for looking at a wide range of other health conditions, as well as being a potential tool to help minimise the impact of any future pandemic. The cost benefit ratio of further support for the Zoe tracking app is huge, and is the sort of action that only Government can take.

Click this link to sign the petition:

I get asked this question quite often, and usually give a fairly brief summary of the process. Recently I wrote a longer response as a result of bing asked how I got from initial picture to this image of the Tadpoles Nebula IC410:

Tadpole SHO


I used a Skywatcher 130P-DS Newtonian scope which is optimised for imaging, with a matching coma corrector to give sharp round stars across the image. Attached to the scope is an electronic focuser, a filter wheel with eight filters and a cooled mono camera (to give lower noise on very long exposures). There is also a smaller guidescope and camera attached to and roughly aligned with the main scope. This is all mounted on a tracking mount, controlled by a computer that also controls the cameras, filter wheel and focuser. As well as taking photos the computer deals with pointing the mount in the right place. It uses the guidescope and camera to take short  (1s) exposures in which it tracks a star and uses this to correct the pointing of the scope in real time, typically with a precision around 1 arc-second.

The larger, cooled, camera is used to take multiple long exposures (typically several minutes, but many factors affect the optimum length, in this case it was mostly two minutes but could usefully have been two or four times as long).

Last year I collected a good set of data using a Hydrogen-alpha (Ha) filter which is a deep red light from ionised hydrogen and usually the strongest signal from nebulas like this. My filter has a 7nm bandwidth, so it suppresses pollution and reduces broadband light sources like stars.

A few nights ago I repeated this with oxygen (Oiii) (a blue-green light) and sulphur (Sii) (an even deeper red) filters.

The images from each layer (typically 20-30) are stacked automatically along with special frames to reduce 'dark noise' and distortions like vignetting (flat frames). As is typical all  three layers needed to be 'stretched' (using curves, histogram etc.) to bring out faint details, denoised and given some contrast enhancement or sharpening. Some processes remove the stars then add them back in afterwards to highlight the nebula, but I tend not to use these often.

The Sii signal was very week, and needed a particularly strong stretch, I was surprised that it eventually revealed some quite delicate filamentary structure that sadly don't show strongly in the finished image.

I combined these using the 'Hubble' palette which maps the three layers in wavelength order Sii to red, Ha to yellow and Oii to blue. This starts with Ha mapped to green, but after balancing the image selective colour is used to push the green to yellow. As 0iii is weaker than Ha it often looks very turquoise and is usually shifted to be more blue. The Sii signal gives the Ha signal a browner tinge and can be best seen as a patch to the right of the nebula where the is little Ha.

The actual colour shifts are rather arbitrary as there is no 'scientific' right or wrong for the colour balance. Unstretched images can be used for photometric measurements, but are pretty uninspiring to look at, so  just as when you process  an earthly image you might choose to alter the saturation for a more pleasing result. In this case the objective is to make the colours reflect the relative distributions of different gasses in the nebula.

I finally improved the smoothness of the image by using the Ha layer as luminance, which remove the noise from the oxygen and sulphur data, but I'm hoping to get more data soon.


The Veil Nebula is a huge supernova remnant in the constellation of Cygnus. It comprises several distinct deep sky objects such as NGC6960, NGC6992, NGC6993 with many twisting filaments glowing with hydrogen and oxygen light. Far bigger than the full Moon, this image is a big mosaic and may take some time to load. It comprises two panes taken using my home-made ED66 telescope which has a focal length of 400mm and a ZWO ASI1600MM PRO camera. I used an OVO field flattener which does not change the focal length. It is a narrowband image using ZWO 7nm filters mapped Ha - red, Sii - green, Oiii - blue.

The Veil Nebula

Roller and angular contact bearings suitable for mini lathes are available from Arc Euro Trade. These are made by Nachi, who apparently supply bearings for the legendary Japanese bullet trains. Unlike the standard ball bearings, which are shielded units, the roller bearings are open and in two parts – an outer race and the inner race with caged rollers attached. Try to overcome the temptation of taking the bearings out of their packaging before fitting them. If you do, keep them clean and do not run them ‘dry’.

See also:

On the face of it replacing the existing ball bearings with taper roller bearings is straightforward  - strip down the mandrel, remove the old bearings, lubricate and fit the new ones and re-assemble. In practice there are a few complications; the bearings are tight force fit in their housings and although the roller bearings are diametrically the same, they are longer which creates the need to make a number of adjustments.

Without tumbler

The end of the mandrel with locknuts, bull wheel and tumbler reverse plate removed.

First of all unplug and isolate the lathe, and make sure you have plenty of space in which to work. The next operation is to strip the rear of the headstock. Remove the gear cover, any change gears that are in place, the drive belt and pulley, and finally the mounting plate for the tumbler reverse. It makes good sense to have a few small boxes in which to store the various components as they are removed, especially small items such as the drive pulley key. I suggest that you fully remove the control box, after noting how the various connections are arranged. In these days of camera phones I make a photographic record whenever I dismantle a novel device, it’s an excellent way of avoiding that ‘James Bond’ moment when you have to choose which pair of wires to reconnect!


Even in the simplest of cases a simple sketch or a digital photograph is always a help with ensuring that wiring is reconnected correctly and safely.

The locknuts at the end of the mandrel can now be removed using a c-spanner. You might have a suitable one in the form of the old pressed steel multi-spanners from the days when bicycles had built up bottom brackets. Otherwise it is possible to improvise one from steel strip and a suitable bolt – it should not be subject to huge loads. To free the locknuts you will need to immobilise the mandrel, and this can be done by bolting a bar to the front flange of the mandrel, or more simply by clamping a bar in a chuck. Please don’t write in about chuck abuse – the torque I needed to free the nuts was quite modest.

Removing locknuts

Removing locknuts.jpg – This method of removing or tightening the locknuts is fine as long as only modest force is required. If the nuts are jammed solid bolt a bar to a faceplate or the mandrel flange instead.

When the locknuts came free I was dismayed to find some apparent dampness and localised surface corrosion, but only on the thread and hidden faces of the locknuts. How this came about without affecting the rest of the lathe is beyond me, perhaps it happened when the lathe was stored in a cold garage for seven months? My remedy was to clean the thread and nuts and then treat them with Jenolite.

It is now possible to remove the 45-tooth bull wheel, wangle out its key and remove the long plastic spacer, followed by the plastic bearing shield which is held in place by three cap head screws. The two ends of the mandrel are covered by these plastic bearing shields. I found that I had to recess the inside of the bore of each shield by 2mm to clear the inner races. Unfortunately this only came to light once I had swapped the front bearing. I therefore had to recess the front shield by hand using a burr in a mini drill. An alternative is to recess the rear shield before removing the mandrel, swap it for the front one at the appropriate time, and recess the other shield when the mandrel is re-assembled. Unfortunately, to do this you will have to temporarily replace much of the drive gear. Whichever way you decide to proceed, the front shield will also need to be released before the mandrel is extracted. This is easily done by poking an Allen key through a hole in the mandrel flange.

Recessing shield

The bearing shields can be recessed neatly in this way, however, a fair amount of assembly and reassembly is required if both are to be dealt with in this way.

To remove the mandrel you will need a suitable a puller. I used a length of 12mm studding, threaded through the mandrel bore, a section of angle iron and two 10mm bolts. The mandrel should come free under steady pressure, bringing the front race with it. In the unlikely event that the front race remains in the headstock, rather than staying in place on the mandrel, you will need to fabricate a longer puller as described on the Arc Euro Trade website.

Pulling mandrel

This fairly robust arrangement was needed to pull the mandrel. There is a nut and thick washer at the far end of the mandrel bore.

Once the mandrel comes free it will leave the gears and a spacer inside the headstock. Another spacer will probably come out with the mandrel, remove this and the long key (that engages the internal gears) and put them to one side.

ARC remove bearing in press

The ideal way to remove the front bearing is with a proper bearing splitter and an hydraulic press. (Photo courtesy Arc Euro Trade)

Lacking a press I had to resort to driving to remove the old bearing from the mandrel and replace it with the new inner race. I used two pieces of 3/8” square bar to make a support the old bearing as close to the mandrel as possible. After applying some light oil I drove the mandrel most of the way out, using an aluminium alloy drift to protect its end. Once the bearing was far enough up, I was able to use a proper Picador puller to finish the job. If I was to do this again I would probably either make extensions for the proper puller, or make a custom puller from scratch. With the bearing removed it is now possible to swap over the previously recessed bearing shield.

Pulling front bearing

Unfortunately, the mandrel was too long for this Picador puller to be used the whole way.

I followed the advice on the Arc website and drove the new inner race into place using a tubular drift – a section of 30mm light alloy scaffold pole, probably the ideal material for the job. Whatever you use, make sure that it fits the inner race without overlapping the caged rollers. It is also important to protect the face of the mandrel by resting it on a block of softwood.

With the mandrel removed, I decided to remove the headstock from the lathe, partly out of curiosity, partly out of convenience. To do this requires the removal of the leadscrew, the motor and their respective guards. The headstock itself is retained by three 8mm cap screws. Once the headstock casting was removed the lathe bed beneath was covered in a layer of ‘plastic fluff’, the result of several year’s wear on the internal gears. Fortunately the gears are of generous dimensions and they did not appear badly worn. Even so it was obvious that it would be worth cleaning out the inside of the headstock and applying a generous amount of grease to all the moving parts before replacing it.

Inside Headstock

The inside of the headstock was devoid of any lubricant, the ‘dust’ on the lathe bed is the result of wear of the gears.

With the headstock removed it was a straightforward job to pull the rear bearing, using the home made puller and a short piece of 3/16” bar across the inside of the bearing. To fit the outer races to the headstock another suitable length of angle iron was required.

Pulling rear bearing

The rear bearing can be pulled in situ, but it is easier to carry out with the headstock removed from the lathe.

The front race went in with no trouble, but the rear one started to go in skewed an I had to adjust the puller. When I came to test the lathe it was apparent that one or both of these races was not completely pulled home, so watch for the puller contacting the headstock rather than the race.

ARC blind puller

A blind puller with a sliding hammer can be used to remove the rear bearing. (Photo courtesy Arc Euro Trade)

Refitting the mandrel to the headstock is best done after refitting it to the bed of the lathe. Grease the internal gears and clean the bed before doing so. Fit the long key and one of the short spacers, then load the front bearing with suitable grease. I used a heavy duty moly-grease. Ideally you shoudl fill no more than 2/3 of the space in the bearing to avoid it slipping rather than rolling. Fortunately, there is plenty of space for excess grease to escape into the void of the headstock, so don’t worry to much about slightly overpacking the bearing.

Moly grease

The front bearing and its race liberally dosed with molybdenum sulphide grease.

The mandrel should slip easily into the headstock, threading into the internal gears and the second small spacer. Now load the rear bearing with grease and slip it onto the lightly oiled mandrel. You can now use the long spacer and one of the locknuts to draw it into place until there is no play, locking the mandrel with a bar in a chuck as before. Check the headstock gears engage properly and refit the tumbler reverse plate, motor, drive belt, leadscrew, control box and any other miscellaneous parts removed earlier. Check there are no obstructions and everything turns freely, then start up the lathe in low gear, gradually accelerating to full speed over a few minutes, then stop and check the spindle has not come loose. Repeat this procedure in high gear.

Spacer too long

The bull wheel spacer before being shortened.

Now remove the locknut and thread on the bull wheel, it will be apparent that the bull wheel no longer lines up with the tumbler reverse. I had to shorten the spacer by 4mm, which I did in the lathe, relying on friction to hold the rear race in place against the light load. Once shortened I had to use a rat-tailed file to enlarge the notch in the spacer for the end of the bull-wheel key, before replacing the rear bearing shield, fitting the key and bull wheel. I then refitted both lock nuts and re-tightened the mandrel. Appropriate preload for a roller bearing this size and precision application is very small, of the order of zero to four thous. The pitch of the mandrel thread is 1.5mm or 60 thou so, once all shake has been removed with the first locknut, tighten it by no more than a further 24 degrees and lock it securely with the second nut.

Shortening spacer

 The long bull wheel spacer was the only one I found it necessary to modify, shortening it by about 4mm.


The first task I undertook was deliberately a light one, turning two 1 1/2” light alloy rings for a tailstock micrometer index. For the record the pitch of the tailstock leadscrew is 1.5mm, so 60 divisions for the index is 0.025mm, a close approximation to 1 thou per division. This did not go without incident, as one of the inner races shifted a little and I had to readjust the bearings. If the outer races snug in a bit further when you first use the lathe, it will become apparent as noise when first putting on a cut. Should this happen stop the lathe and retighten the bearing and no harm will be done.

Could I replicate George Thomas’ achivement of parting off 1 1/8” FCMS at 600rpm?

I have successfully parted off 1 1/2” mild steel to make the parts in photograph X and 1 1/8” alloy steel, to make the 0.025” thick disk. It took a little practice to develop the right technique. The tool was 3/32” wide with the cutting section 1/2” long, set dead on centre height. I got the best results starting at 220-250 rpm and then speeding up as the work progressed. It was essential to force the tool in positively at the start of the cut, then to maintain a steady feed rate, but being careful to watch for the work slowing down too much. Lubrication, using neat cutting oil on a small paintbrush helped keep the speed of the lathe up. Chatter was only evident if the cut was not aggressive enough. Feeding the tool in too fast did not cause dig-in, instead the ‘motor slowed and the fused blowed’. This suggests that the limiting factor is now the motor power, not the rigidity of the lathe.

Parting examples

Examples of items parted off with the modified lathe. The disc is 0.025” thick, and could easily have been made thinner. The swarf is 3/32” wide by about 0.004” thick, a flatter topped tool would have made less tightly wound swarf.

A final and more extreme test was to cut a 3/32” groove in 2” mild steel. Running the lathe at about 310 rpm I set up my digital camera to video the event. Remarkably the tool went in like a knife into butter – no chatter and short coils of swarf flying out like chips from brass!

Parting off

Parting off.jpg – After cutting a 3/32” wide groove in 2” diameter mild steel. The lathe was run at over 300 rpm, and no chatter was experienced.

I had not expected such a dramatic improvement in parting off, particularly from the point of view of being able to speed up and increase the depth of cut so markedly. I certainly did not expect any other changes, but there were two. The first came to light when running in the new bearings – the reduced friction meant the lathe ran far more smoothly at low speed – it would now turn at just 4 rpm without stutter. This could be very useful if I decide to mill or grind helical threads in the future.

The second and welcome change was the improvement in surface finish on tougher materials. I was not expecting this, yet there was a noticeable improvement in the surface finish when turning both mild and medium carbon steels. So far I have seen none of the ripples or waviness that troubled LBSC over eighty years ago. I wonder if his problem was poor bearing quality or just too much preload?

I did have two reservations about the change. One is the apparent 4mm of increased mandrel overhang, however because the new bearings have a linear contact patch instead of a point, the new bearings actually support the mandrel with /less/ overhang. The second is that the new bearings are unshielded. Provided the plastic shield are undamaged and the bearings well packed with grease this ought not to be an issue. Assuming normal levels of use it would be a good idea to draw the mandrel and inspect and repack the bearings every couple of years to ensure long life.

I had considered changing the bearings a last resort to rescue a worn or otherwise unsatisfactory machine, would I be able to detect any difference? In practice, for a day’s effort and at very modest cost, the results really surprised me. Even with a lathe that appears to be performing well with the standard ball bearings, consider changing to roller bearings and taking full advantage of the basic precision and rigidity of the mini lathe design.

See also:

Over recent years various 3 1/2” centre height “mini-lathes” have become increasingly popular as an entry-point into model engineering. Personally, I don’t like the term ‘mini lathe’ – it suggests that they are toy-like or unsuited to ‘real model engineering’. Fifty or sixty years ago, if you could not afford a large, full-featured lathe one choice you had was a true lathe such as the Adept, Flexspeed or Centrix Micro – each of which could truly be described as ‘mini’ and lacked any screwcutting capability. Even so these lathes were capable of producing superb results in the right hands, and in the hands of some owners were modified to become full featured and true precision instruments.

CL300M lathe

My mini lathe, a Clarke CL300M, taken before the bearing change. At least eight other modifications can be seen in this view!

The modern ‘mini-lathe’ is technically far advanced from these historic examples, though it fills the same niche. It has much greater capacity, a built in variable-speed motor, screwcutting and fine feed, is far more rigid and, with the benefit of modern mass production, is probably more accurate. The one issue that does get raised by users of these lathes, however, is difficulty in parting off. Parting is one of the most demanding tasks for any lathe, as it involves taking a broad cut with an overhung tool, often at the bottom of a deep groove.

You may ask what all this has to do with roller bearings and angular contact? Well there are several requirements for successful parting, and the quality and adjustment of the lathe mandrel bearings are one. Mini lathes are supplied with the mandrel fitted with ball bearings as standard, and for some users these have proven to be a source of trouble when parting off or even, in extreme cases, when trying to achieve good surface finishes. In lathes where these bearings are not up to scratch, changing to roller bearings is an economic and realistic way to address the problem.

See also:

But there are many other pitfalls for the careless parter-offer, and unless these are addressed changing your mandrel bearings is pointless, and if they are, you may not find the need to change them anyway. So in this article (sorry!) I will explore the mysteries of parting off and explore the various solutions to allowing our mandrels to rotate.

Perhaps every model engineer wishes to attain LBSC’s nirvana of ‘parting steel with a sound like frying bacon’, but it was the greatly respected George Thomas who made parting off a fine art. He combined a well-adjusted lathe, a rear toolpost, the right tool and the right speed. With a special tool shape he reached a point where he could part off 1 1/2” mild steel at 615 rpm! By his own admission this was a piece of showmanship reserved for exhibitions, and more usually he found a suitable speed (for mild steel) of 200 rpm per inch of diameter.

A well adjusted lathe means ensuring the saddle, cross-slide and top slide all move smoothly, but without any trace of shake or movement. Rock the toolpost and watch anywhere two slides rub together – if you see a bead of oil moving in and out, then they are probably too loose. Make sure there is no play in the mandrel in its bearings. If there is, cautiously tighten the locknuts at the rear of the mandrel.

A rear toolpost, with upside down tool, is widely recognised as improving parting off, by reducing the chance of ‘digging in’. Apparently the rear-mounted tool will tend to spring out of, not into, the work and for materials which produce short chips, rather than curls of swarf, these will tend to fall clear of the cut. I have never seen a completely convincing explanation of the geometry involved, but several respected model engineers have testified to the effectiveness of this approach, and it has been a popular technique since the 1940’s. Mini-lathes do not mount chucks on a screwed mandrel nose and have reversing at the flick of a switch. This means to explore the benefits of this geometry one simply has to invert and pack up a parting tool in the normal toolpost, and reverse the lathe rotation.

George Thomas’ ideal tool had a v-groove on top and the end ground to a v-point, to encourage the formation of curled, narrow swarf (Duplex advised a rounded convex-ended tool for the same reason). I’m sure that both would have admitted that this is a counsel of perfection and that what really matters is a sharp tool with appropriate top rake and adequate front and side clearance. Two essential pieces of George Thomas’ advice, however, are to match the size of the tool to the job and to keep the end of the tool square to the job. Small tools are not rigid enough to tackle large diameters and the angled tool end that avoids making an end pip produces sideways forces that can bend the tool off line and jam the job.

Parting Tool


This drawing shows ideal angles for a normal parting tool as recommended by Duplex.


Getting the right speed for any parting job depends on so many variables (the material, its diameter, the tool and the state of the lathe, and some would say the phase of the moon) that there has to be an element of trial and error. The same applies to getting the correct rate to feed in the tool. Fortunately the variable speed control of mini lathes comes into its own for parting – you can gently vary the speed to get the best results, and also speed up as the diameter drops.

Finally, don’t be afraid of the job. A jam up and a broken tool-tip can make you understandably cautious. George Thomas was adamant that a confident and positive approach was essential, and that over-cautious pecking at the work will never bring good results. Don’t just watch either – let your ears and your fingertips tell you what is happening.

A 1940’s article by LBSC advocated a useful tool, not for parting off, but for the similar task of making deep, wide grooves in large diameters. The tool has the shape of a fish’s tail in plan view, the opposite of George Thomas’ tool. I ground a shallow vee in the end of a normal parting tool using a mini drill mounted stone, in order to turn a wide groove in 3” diameter cast iron. I found this shape of cutter was less liable to chatter than a normal parting tool. In use it is fed in five to ten thou at one side of the groove, moved across to the other side of the groove and fed in again before the return journey. The ‘double pronged’ design cuts the full width of the slot to the full depth with ease.

Turning multiple blanks

Producing a series of gear blanks in a medium-tensile steel was essentially a multiple parting off operation. Note chatter marks at the bottom of some diameters.

At first I had all the parting off problems recounted by many mini-lathe owners, and had my fair share of ‘dig ins’ and chatter. In time I was able to overcome these by following the above advice. I also found the huge difference made by a brush full of neat cutting oil in the groove. Most of all, I found the benefit of having confidence to feed in and get a positive cut. After a fair share of jams and scrappers, I eventually reached the point where I could reliably part off 1 1/8” alloy steel. With standard bearings I even used a parting tool to cut a deep groove in some (freely machining) 2 1/4” diameter cast iron.

Making a gear cutter

Using a broad form tool, such as when making a gear cutter in silver steel, is possibly one of the few tasks even more demanding than parting off.

Let us assume that you are in the position of some mini-lathe owners, having tried every tip and wrinkle and still not getting successful parting off. The final option is to replace the mandrel bearings in the headstock. There are a number of requirements for lathe bearings. They need concentric accuracy, they must be able to take both high axial loads and thrust loads, they must have low friction and ideally they must have a long life.

Roller bearings are the ideal choice when turning the barrels of a battleship’s 16” guns, simply because of the loads involved, but for small lathes the choice is less clear as almost any type of bearing could be used. Once, many had mandrels running in a plain cast-iron headstock, which is a perfectly good solution with adequate lubrication. Before the Second World War plain and taper bushes and sometimes ball bearings were the rule. Roller bearings were scarce during the first half of the twentieth century, but their use took off during the Second World War. Even so many top quality post-war lathes, notably Myford’s ML7, used plain bushed bearings and a ball thrust bearing, whilst the Super 7 used a hand scraped tapered bronze bush at the head of the mandrel. There is no doubt that these bearings could handle any task within their lathe’s capacity. On the other hand slightly larger lathes, such as those by Boxford and Colchester, typically had roller bearings.

Milnes mandrel

The mandrel of a Milnes Type R lathe, fitted with two taper roller bearings at the front of the mandrel and two further ordinary roller bearings.

At a slightly earlier time LBSC was in no doubt – anything but roller bearings! In one of his “Lobby Chats” he told the tale of new “Type R” lathe he had received in 1923 from Henry Milnes, of Bradford, to replace his 3 1/2” Drummond. If I dare paraphrase a tale he told in his inimitable fashion, it went something like this: He didn’t care for the specified roller bearings, but obtained a written guarantee that, if unsatisfactory, they would be replaced with plain ones. Within three weeks he had detected ‘fine lines” and got onto the maker, who replaced the front bearing with a bronze cone. Another three weeks and the problem arose again, and this time it was agreed to be the fault of the tail-end roller bearing. The patient Mr Milne agreed to supply a specially made roller bearing, but there would be a few weeks delay. Now LBSC depended on his lathe for his living, so the manufacturer sent him a temporary bronze bush, which he fitted, immediately curing the trouble. In due course the new ball bearing arrived, but as the bush was doing fine, he thought he’d keep it for the time being. Some time later, while turning a wheel for ‘Bantam Cock’ on the self-same lathe he started getting chatter marks; inspection revealed a little play in the bronze bush, it had worn just slightly oval. The replacement ball bearings were found to be a perfect fit. The bush was put into ‘honourable retirement’, with due reverence. This story was told in 1946 – the temporary fix had done its duty for over twenty-three years.

Milnes Type R

A Milnes Type-R 3 3/4” lathe, fitted with lubricators for the mandrel bearings, as used by LBSC for around a quarter of a century (see www.lathes.co.uk for more details of this and other veteran lathes).

So Curly wouldn’t touch a roller bearing with a barge pole, but a year later in 1947 they were stoutly defended in the Editor’s correspondence. E. H. Doughty, British Timken Ltd.’s Chief Technical Engineer, challenged some of the magzine’s responses to queries concerning the rigidity and finish achievable with roller bearings earlier in 1947. The detailed letter included arrangements for four-, three- and two-bearing spindles. The two-bearing arrangement is exactly the type that applies to mini-lathes, and is stated to be ideal for smaller lathes where temperature variations are small. He claimed high precision, robustness, long life, low friction and the potential for high speeds as the benefits of their bearings. Mr Doughty did not agree that plain bearings were preferred by most users, claiming that the use of roller bearings was increasing rapidly both before and after the Second World War. He added that during the war tens of thousands of units gave “every satisfaction” in machine tools of all kinds.

Timken two bearings

This design for a two-bearing work spindle by British Timken is comparable with that applying to mini lathes.

Another advantage of ball or roller bearings is ease of lubrication. A plain bearing mandrel demands a slow but uninterrupted flow of oil, otherwise rapid wear will soon be apparent. A sealed ball bearing may not expect any lubrication during its normal service life, and even ‘open’ grease-packed bearings need only occasional attention (how often do you repack the axle bearings on your family car?)

Now Curly’s problem was not just chatter, it was also fine surface patterns on the work. Unsatisfactory finish was clearly a concern among likely users of roller bearings. Mr Doughty claimed very fine finishes, even at high production rates. Perhaps a clue to where poor finishes arose is given by his explanation of the need for no or minimal preload. Apparently the popular impression was that a fair degree of preload was necessary, generating excessive friction. I imagine the ‘wedging’ action of taper roller bearings makes them easier to overtighten than angular contact ball bearings, and that one symptom of this might be a poor finish. Another one-time Timken employee - ME Technical Editor, Neil Read, pointed out that in 1923 LBSC’s roller bearings would have been made to standards of fit and finish far from what we might expect today.

Whilst there is no doubt that plain bearings can give excellent results in small lathes, most modern small lathes don’t use them because of cost - ironically the relatively complex ball bearing is a mass-produced article. It is considerably cheaper than a tapered phosphor bronze bush – particularly a skilfully hand-fitted one. Our economically priced mini lathes use this cheapest option – ball-races.

Mini Lathe Bearings

Ball races offer only point contact, and in theory even if perfectly adjusted they are inferior to taper bearings (whether plain or roller), both of which have a far greater bearing area, promising greater rigidity and accuracy. Because of the arrangement of the headstock a plain bearing is out of the question for us, but roller bearings are only about 50% longer than an equivalent ball raced bearing, and we can squeeze them into the space available. The standard (ball raced) bearings fitted to mini-lathes are pretty hefty units, and able to take a considerable load. I understand, however, that depending on the source of the bearings on any particular unit, you may get ones made to different specifications. Certainly a proportion of mini-lathe owners report persistent problems with surface finish, chatter and particularly with parting off. If we accept Mr Doughty at face value, those with these problems should be able to solve them by fitting taper roller bearings and possibly improve the rigidity and accuracy of their machine to boot. Certainly there are an increasing number of mini lathe owners who have either carried out the ball to roller conversion, or who are contemplating it.

Rear bearing

The original rear ball raced bearing removed from a mini lathe.

For a mini lathe where loads are much lighter and the motor is less powerful, another option is to fit angular contact ball bearings. These share the benefits of better control of end loading while being better suited to high-speed use; they also can by exchyanged for standard bearings with no dimensional changes making coinverion rather simpler.. They have become increasingly popular as a choice of mini lathes.

Front rollers

Replacement front inner race with its caged rollers fitted to the mandrel of a mini lathe (together with the front bearing shield and a short spacer).

All of which begs the obvious question – if it ain’t broke, why fix it? I am another traveller on the endless quest for the ‘ideal lathe’. Adding roller bearings would not harm my lathe, and unlike roller bearings, I gain some real control over the bearing adjustment. If they gradually wear, I can take it up properly.

In a second article I will describe my experience in fitting roller bearings to a Mini Lathe, and give my assessment of the effect they have had on the lathe.

See also: