Historical Notes: Ball and Roller Bearings

For some time we have been aware that many of our readers would appreciate engineering articles in the pages of Motor Sport, but there has been the difficulty of striking a balance between elementary and advanced presentation and of deciding which subjects to cover. After careful thought we have decided that what is wanted is an historical/technical introduction to various aspects of the motor car, for it can truthfully be said that to properly understand present-day practice it is necessary to possess some knowledge of the developments leading up to it.

Consequently, we propose to publish a series of articles, of which this is the first, outlining the evolution of different parts of the motor car, from which it will not only be possible to derive an idea of why things are now done in the way that they are, and to follow the design and manufacturing problems involved, but also to see in better perspective the merits and shortcomings of the early designers. These articles will be written by an engineer, who prefers to hide his identity under the synonym “A.B.C.”—Ed.

Ball-bearings and the motor car were not born together, but they grew up side by side, the one very much dependent upon the other, until the time when the new-found art of flying made prior claim upon the ball-bearing during the conflict of 1914-1918. The whole essence of the technical development that led up to the evolution of the motor car, and that which made its subsequent progress possible, may be summed up in the now somewhat overworked phrase “More power for less weight.” Both these conditions, to be satisfied, demanded operating principles which necessitated ball and roller-bearings, i.e., faster rotational speeds and higher unit loadings. The laws of mechanical losses made sure that the gas-engine, turning at about 300 r.p.m., would not long survive into the motoring era, although Benz himself would have had it so; but the angel of progress, using Daimler and de Dion Bouton as her evangels, soon increased motor car speeds and reduced motor car weights, so that both prime mover and “running gear” called aloud for ball-bearings. This tendency was, of course, greatly accentuated during the period of the first world war, when the cry for ” power-to-weight” became even more persistent in the air. Indeed, so great did the demand for these small components become that tales of special “missions” to other bearing-producing countries became legion, those associated with the Mosquito flights during the late war being too recent to need recapitulation!

By 1914, the ball-race had reached a state of development whereby the whole engineering effort of the country depended, then as now upon a ready supply, and it is very largely due to the motor car that such rapid progress was made by that year. When the motor industry first demanded bearings, that is from its earliest moments, three stages of development had to he experienced and passed through. Firstly, it was necessary to decide what was the right sort of bearing to produce, and it is surprising how difficult this proved. Secondly, having decided what to produce, it became necessary to decide how to make that particular sort, again no easy task in those far-off days when machining to .001 in. was not so easy or commonplace as is now the case, and in addition it was essential that the manufactured price was within reason. Finally, both the ball-baring industry and the motor industry had to learn how to apply the resulting product, so that it gave of its best in the hands of the average motorist of the Edwardian age.

It is difficult to set dates to the periods, during which these three separate stages took place, but in general it may be said that if we date the motor car industry from 1894, then the only proved ball-bearing in existence at that date was the “standard cycle” or “cup-and-cone” type.

This pattern had first forced its existence upon a flourishing cycle industry during the Paris-Rouen race of 1868, and very successful it proved during the intervening years, always provided that the speeds and loads did not greatly exceed those that prevailed under bicyling conditions. But, obviously, the weight of the average rider on the pedals, or the side thrust of the rear wheel of a pedal tricycle in racing trim, bore no comparison to the loads imposed upon the hub of a Panhard car put into a corner on a rutted road at 12 m.p.h. in 1895 or thereabouts.

A new approach was therefore needed. An approximate idea of the evolution of what we may now call the standard journal-type of ball-hearing may be gathered from Fig. 1. It will be seen that for a period, the idea of “double contact” and adjustability was pursued.,no doubt on account of the aforesaid bicycling experience, but the altogether heavier duties demanded by the new field soon showed that both these ideas spelt unreliability and a short working life. The double contact scheme failed because, in spite of the apparent extra load that might be carried per ball, its geometry was unsound. Again, although adjustability seemed an obvious advantage, it was soon discovered that by the time the adjustment was needed the tracks were worn so much that the new position failed to give a pure point-contact and a noisy and short-lived bearing resulted. Much of the early research work on the load-carrying capacity of balls was undertaken by Professor Stribeck at Neubabelsberg, backed by Deutsches Waffen and Munitionsfabriken of Berlin. An account of these experiments may be found in Engineering dated April 12th, 1901. Prof. P. L. Renouf read a very good paper dealing with the problem so far as cycles were concerned, entitled “Ball-bearings as Applied to Cycles,” before the Cycle Engineers’ Institute on May 3rd, 1900. Mr. J. D. Roots, who was responsible for some very early motor vehicles in this country, in 1895, records how he approached several ball-bearing manufacturers, asking them to make for him some bearings which corresponded in design with what we would now call a “standard-type” of ball-bearing. In each case he was strongly recommended to use the “cup-and-cone” type, and eventually he had to make his own, buying the balls alone from the specialists.

Once the standard-type of bearing had been evolved, round about the turn of the century, practically all development went into this type, and from the purely design point of view, the only change was the gradual alteration of the groove radii from about twice the ball diameter to about nine-sixteenths of the ball radius. The cages, too, gradually evolved from the twisted-wire types to the present soft-metal rivetted affairs.

As regards accuracy of manufacture, it will readily be appreciated that this factor was all-important if ball-bearings were to succeed. It was quickly realised that failure to keep the size of the individual balls to within a limit of .0003 in. could quickly result in very severe loads falling upon the “big” ball, with consequent race failure. In the pre-motoring days such a high degree of accuracy was not needed, due to the altogether lighter work imposed. By 1909 most reputable ball-bearing manufacturers were producing halls that were accurate to .0001 in., but it would appear that there were still those who made the “cheap and nasty” sort. Elaborations of the basic idea gradually appeared, such as the Skefko in 1910, which we today should call the “double row self-aligning,” and each in its particular way contributed to the general march of progress.

Before discussing very briefly the actual materials and method of manufacture, it is well to consider the design history of the roller-bearing as many of the manufacturing difficulties are common to both types.

As a method of approach to the problem of higher load-carrying capacity, the roller-bearing naturally commended itself to the manufacturer’s thoughts—for why not a “line” contact to replace the “point” contact of the ball-bearing? From time immemorial man has known that the coefficient of friction could be reduced by placing rollers under that which he wished to move, and in this sense, perhaps, the roller-bearing was the more logical approach to the motor-car problem. Unhappily, however, the problem of machining rollers that were absolutely parallel quickly raised a seemingly insoluble problem, due to the rollers trying to roll in a helical path, which, being restrained by the cage, produced a severe end-pressure, raising the friction losses and, in extreme eases, pushing the end right out of the cage. The story of the evolution of the roller-bearing is really the story of a gradual reduction in the “aspect ratio” of the rollers from about 20:1 to 1:1 (at which figure it has remained fairly consistent ever since 1912 or thereabouts), coupled with improving manufacturing technique. Readers will probably be familiar with the Hyatt bearing, which was early introduced, and will have noted that each roller was given a right and left-handed special groove in order that, placed alternately, each “screwed” up against the opposite end of the cage, thus eliminating end-pressure. It was these rollers that made the back-axle of the model-T Ford so everlasting. Before the turn of the century the great house of Lanchester, like Roots, were forced to make their own rollers, being laughed to scorn by the manufacturers of ball-races when they made their approach. A whole book could and should be filled with the story of Lanchester’s genius–this being but one example of the pure technical thought that emanated from that happy mind. The house of Timken produced the taper roller in 1907, in time to meet the demand for a bearing that would deal in compact form with really heavy combined journal and thrust loads, and this rapidly became the standard fitment for front-wheel mountings.

Two processes were used in the manufacture of balls in the early days, that of turning them from the solid bar and that of forging them in “strings.” The first of these two alternatives was the one that best stood the test of time. Grinding was done between two pressure plates, this process sufficing until the Hoffman Company produced the “tumbling” process, which constituted a real advance in the technology of the ball-bearing, the accuracy of balls made by this process being of the order of .00001 in. It would take too much space to detail the actual manufacturing process, but those readers who desire to delve deeper are referred to the standard literature on the subject. Once the standard ball- and roller-bearing had appeared, the difficulties of application and protection rapidly became apparent, and it is interesting to see in which historical designs the bearings are correctly applied. Exclusion of dirt and proper lubrication are obviously pre-requisites of a long-lived bearing, and it is astonishing how few of the early manufacturers dealt adequately with these problems. In the early days the responsibility for providing adequate protection of these bearings fell squarely upon the individual designer and, by the care that each took over this particular problem, so are we able to sort out motor cars that established reputations for longevity and those which fell to pieces in a comparatively short space of time. Mercedes and Fiat, closely allied in design, are amongst those in which the importance of this feature was early appreciated. Grease has always been the standard method of lubricating such bearings so far as the “running gear” is concerned, and here the early motorist relied upon the individual honesty of his garage or supplier. Stories are on record of greases that were artificially thickened with various species of chalk, the average motorist and “garagist” of those far-off days usually selecting his grease on the basis of “the thicker the better.” The effect of such a mixture on the standard ball- and roller-bearings of 1900-1910 can doubtless be imagined!

But it is even more fascinating at this date to look through the early motoring books and to see the number of bearings that were wrongly applied from the very drawing board, due to ignorance on the part of the individual designer. Mr. G. F. Barrett, in his very illuminating paper called “Causes of Failure in Ball-bearings,” read before the I.A.E. in 1911, related how more bearings failed in the early days due to this ignorance of application than to faulty manufacture. He deals at some length with specified cases, such as front and rear hubs, back axles and gearboxes, space forbidding that other than the latter be dealt with in the present article.

Appended are two drawings, based upon those of Mr. Barratt, which show the subtle difference between gearboxes typical of the period 1910, in which the bearings are correctly and incorrectly applied. In Fig. 2 it will be seen that the input shaft has the first journal bearing held on by means of a thrust-washer only, while the second journal bearing is not clamped upon the shaft at all. The main shaft has its journal bearings mounted in the same way. The adjustment of the two thrust hearings is obviously through the outer races of the two outward bearings, through two special washers on to the two single thrust bearings, and then through the inner races of the two journal bearings on to the shoulders of the driving and main shafts, the thrust between these two being taken on two hard steel buttons inside the driving shaft. The faults of such an arrangement are clearly apparent. The adjustment of the thrust-washers depends entirely upon the fitting of the inner dust caps, situated at each end of the gearbox, and it is almost certain that the adjustment will be wrong. The adjustment also depends upon the sliding of the outers of the journal races, which means that if no end-thrust is to be placed upon these races, the inner must also be a sliding fit upon the shafts, in spite of the fact that “creep” invariably intrudes if such a fit be permitted.

Looking at the layshaft we see that although the inners of the journal bearings are correctly clamped on the shaft, and the outers are held firmly by the machined internal flanges, a very small error in the machining of these flanges in the length of the sleeve on which the gears are mounted, can easily result in a permanent side-thrust being placed upon the bearings. Fig. 3 shows the corrected layout. Double thrust-washers have been placed upon the input and output shaft, both of which may be properly adjusted and locked before being mounted. The journal bearings have their inners firmly clamped to the shaft, with their outers as sliding fits. On the layshaft the outer races are a sliding fit in their housings, while the end-thrust, if any, is now dealt with by the hardened steel pads or buttons mounted between the ends of the shafts and the caps themselves. The dirt shield will be noticed, expressly designed to prevent the ingress of the abrasive components then quite commonly used to lap in the gears themselves, prior to the final assembly of the gearbox, obviously another prevalent source of bearing trouble if the maker were careless.

Such examples of wrongful application, together with the aforesaid lack of design provision against dust and dirt and the possibility of poor lubricant generally, undoubtedly mitigated against the popularity of ball- and roller-bearings, so that it was no uncommon thing for makers in 1903 or thereabouts to say that they preferred plain bearings, but the example of Mercedes in 1902 showed clearly what could be done. That chassis well repays close study so far as its bearing layout is concerned, a feature in line with its other advanced specification. But progress can never be long delayed and by 1906 the greater proportion of manufacturers had adopted such bearings for most parts of the transmission gear, although even as late as 1910 there are many examples of gearboxes still operating with plain bearings. It is also remarkable how long the practice of fitting two purely journal ball-bearings for front hubs lasted, especially as the taper roller-bearing was available as early as 1907.

That, very briefly, is the story of the adoption of a specialist component, itself a triumph of engineering science, without which our present-day motor cars would indeed be poor affairs. By 1939 ball- or roller-bearing failure was almost unheard of, and, when that state of affairs prevails, one wonders whether the last word has not been said.—A. B. C.