In rotating machines, noise and vibration are indications of faulty operation of ball bearings. More often than not trouble is traced to the design and assembly of the machine rather than to the bearings themselves. The technology of their manufacture is such that ball bear­ings in general are precise and reliable components. Therefore, the satisfactory performance of bearings in the assembled product is clearly the responsibility of design and production.

The balls in a bearing require an absolutely parallel track to roll upon, entirely free from eccentricity, wobble or other variations. Study of the bearing discloses what can happen to destroy this accurate mechanism if a lack of squareness of mounting the bearing exists in any perceptible degree. Should the inner ring be cocked to a tilt position, the balls are forced to climb one side of the raceway during part of a revolution, with resulting drag on the bearing retainer. Then, the balls accelerate down the raceway to climb the opposite side in the other part of the revolution, reversing the strains on the bearing retainer. The balls instead of rotating about a true horizontal axis tend to spin from contact with the sides of the raceway and to reverse their direction of spin during the second half of the revolution.

Some degree of lack of parallelism or squareness of the rings or associated parts is permissible before bearing performance is considered unacceptable under commercial standards.

Misalignment is, of course, not the only cause of improper bearing operation. However, it is a fault that is clearly related to factors of design. Some constructions are more liable or sensitive to misalign­ment than others. Internal looseness, excessive end play and improper shaft fit also lead to bearing troubles, which usually are indicated by noise and vibration.

Noise is probably the most common cause of rejection, simply because it covers the most ground and its existence is so noticeable. By noise is meant those sounds objectionable to an inspector’s stand­ards of operation. Since the human factor is involved, such noise evaluation is not an absolute test. The level at which sounds become annoying is not the same for all persons. Except for borderline cases, however, inspectors agree consistently on whether or not the noise in a bearing under test is objectionable. A more positive evaluation of bearing noise can be obtained by use of sound meters, with or without benefit of “quiet room” test conditions.

Once it is decided that a bearing is noisy, the cause can often be determined immediately from the nature of the sound. Most bearing faults can be associated with definite types of noise.

Loose Clearance of Balls. Internal bearing looseness is usually identified by a high, even tone or rattle. It may be quickly checked by pressing axially against the end of the shaft with a pointed wood stick. If looseness is the only trouble, this action will immediately lessen the noise. Heavy-duty small motors and other rotating machines for everyday service at moderate speeds are often supplied with bear­ings of a medium fit to permit some latitude in the assembly of the inner ring on the shaft. When mounted, the axial play range is nor­mally between 0.0025 and 0.0045 in. Applications for high speeds or requiring above average quietness have bearings of a tight fit with an axial play range in small units of 0.0010 to 0.0025 in. This is strictly axial play and not radial play, which has a value of about one seventh to one eleventh of these figures.

Where only one of the two ball bearings is relatively loose in internal fit, any trouble in either bearing usually adds to the noise in the loose bearing, regardless of whether or not it originates there. Factors such as unbalance, bent shaft or wobble stand out in this re­spect by greatly increasing the magnitude of noise, even though the trouble may be at the opposite end of the unit. If noise control is essential, care should be taken to have both bearings in about the same internal lateral fit range.

High, singing noise in a bearing may signify one of three common things: dirt in a high speed bearing, rust at any speed, or radial preload from a too tight shaft mounting. More rarely, windage noises from slight pressure differential currents in a fan or blower unit are of this character. The trouble may be localized by removing the shaft from the housing. The outer ring of one of the bearings should be held so that the shaft is vertical. When the shaft is spun, the fingers will detect roughness if rust or dirt is present. Dirt tends to cause a drag in the slow-turning bearing so held, but rust does not. The bearing at the opposite end should be checked also, as noise sometimes is generated from an entirely different place than is initially supposed when the unit is running.

A high radial preload, produced by forcing a bearing on a shaft too large for normal practical fit, will expand the ring until no inter­nal play remains. This condition is easily checked by trying to move the bearing outer ring sideways. An excessively preloaded bearing has no sideways freedom and appears to produce a decided drag on the slowly turning shaft when the bearing outer ring is held in the hand. Any windage sing can be identified by covering ventilation openings in the housing. This should result in changing the noise tone or stop­ping it altogether. Noise from this condition will charge more rapidly with varying speed than will noise from the three other conditions and it will completely disappear at lower speeds. Although windage noises are not related to bearing troubles, bearings are usually suspect until the troubleshooter proves otherwise.

Shrill Noises and Chatter. An extremely shrill noise arises from raceways, end shields or ball retainers that were damaged or imprinted from light hammer blows. In these cases, the tone changes with the speed of the unit, but does not die out at low speeds. A feeling of roughness will be evident when the shaft is turned slowly by hand.

Chatter indicates too large a bore in the housing so that the bear­ing outer ring creeps and chatters from its own or other vibration in the unit. Chatter of this sort may be difficult to locate where a rough housing bore exists. The minute peaks and valleys in the bore surface provide a fit, which may seem on first assembly to be tight. However, with the vibration encountered in use, the bearing becomes loose as these peaks wear down. This fault is one of the common and recurring trou­bles found on production lines where tone quality control is rigidly required by the acceptance standards of the product. Cure for this trou­ble is to provide a smooth surface bore in the housing, held strictly within tolerances. Chatter may also be controlled by painting the hous­ing bores just before bearing assembly with a very heavy viscosity oil above 3,000 saybolt. The thin film on the broad surface of the circumference prevents chatter effectively, and appears to hold its ini­tial qualities upon examination after two years of continuous service. It is also somewhat effective in dampening noise from other causes.

Smooth tone quality or lack of objectionable noise is a common objective in design of high-speed motors or similar units. The con­ventional unit often has a flat spring washer to take up axial or lateral play. This tends to lessen variations of tone, which sometimes reaches the chatter stage from outer bearing ring creep. Noise from “looseness” is at its loudest in those units that, by chance, have been assembled of com­ponents, all or most of which are at the extremes of their dimensional tolerances. The cumulative play in these components is considerable.

The spring is intended to take up all of these variations and is required to keep a nearly constant pressure on the bearing outer ring. Any major increase in pressure will cause preload troubles, with re­sulting heating and loss of speed or performance efficiency. Another function of the spring is to grip the bearing outer ring to stop turn­ing, which otherwise might cause creep and chatter. Use of long springs, of double flat spring washers or other compressible elements, such as synthetic rubber washers, effectively provides uniform tone quality over the range of production tolerance variables.

One noise may be overlooked in trying for tone quality. This is grease slap, an intermittent click or slap noise coming from a double shielded bearing when the unit is first started on run-in test. This is simply a part of the channeling process, does no harm and disappears in a short time. The duration of the noise depends on the quantity of grease in the bearing and the speed of the unit.

Vibration. In this case, vibration is considered only for its rela­tion to noise; not for fatigue or other effects. To be objectionable from this standpoint, the vibration is above moderate amplitudes and can be measured by hand feel of the unit being tested.

Lack of dynamic balance is a frequent source of trouble. It is rec­ognized by an extremely strong and sharp vibration, which prevails on rundown test until the unit is well below the critical speed. Static balance will not provide dynamic balance.

Bent shafts, double bent shafts, or whipping shafts produce sharp vibrations that feel almost the same to the hand. However, vibrations from the first two causes retain their intensity on rundown until the unit is almost stopped, far longer than is the case for pure dynamic unbalance vibration. Shaft whip disappears at certain critical speeds and may be most easily located by use of a stroboscopic tachometer. Bent shafts may be of simple bend or compound bend, depending upon the handling. The bends vary as to location over the length of similar shafts. Long, thin shafts are, of course, the most susceptible to bends. Rejections commonly mount from one to twenty per cent and back again in a single day from this cause. The magnitude of the shaft bends that bring trouble run in the order of other off-square variables. While soft, thin shafts may be strong enough for a given application, expe­rience shows that more rigid shafts pay dividends in attaining uniform­ity in tone quality.

Wobble of mounted bearings represents a major cause of trouble where provision is not made for the proper proportion of the shaft shoulders. This trouble is recognized in a running unit from the mag­nitude of vibration, which is heavy like that from dynamic unbalance.

In units having long, thin shafts, vibration can be caused by parts mounted on the shaft extension if these parts are only statically balanced before assembly with the shaft. The possible result is heavy vi­bration, which is sharpest at the back end of the unit, away from the shaft extension. Where the back bearing has more internal lateral play than the front bearing, this condition will be accompanied by consid­erable noise in the rear bearing. In high-speed units of this type, shaft whip also enters the problem and increases the possibilities of rejec­tion. Short stocky shafts and extensions are the cure for rejection from these causes, along with a proper degree of dynamic balance for the parts mounted on the short extension.

Noise and vibration often develop in units where end caps inclose and clamp the bearing outer rings. The trouble arises when the caps are machined improperly so that the face against the bearing outer ring is not parallel to the cap flange face against the housing. Designs using die-castings are particularly liable to this trouble unless the surfaces of the casting are finished.

Vibration of a similar nature arises from clamp nuts and threads cut on the shaft. If the threads of both are not cut in the proper relationship with the axis of the shaft, the bearing and clamp surfaces may be in misalignment and wobble may result. This is especially true of tight fitting threads. The clamp nut should be free turning with a slight amount of play between the shaft and nut threads. Such play effectively absorbs out-of-squareness and prevents cocking of the ball bearing inner ring. In some cases, a cocked bearing may result in bending of thin shafts, causing unbalance in running tests.

Heating from axial preload may be caused by expansion of the rotating parts, with the result that all of the end play available at the floating bearing location is taken up. Alternatively, the axial preload may arise from the chance assembly of housing parts at the lower end of the allowable tolerances with rotating parts that are at the higher end of the tolerances. Such a combination would take up all endplay and cause bending. Heavy preloads may be placed on the bearing by either of these causes. Yet it is difficult to determine the magnitude of these loads during test because turning the shaft by hand slowly discloses only a slight drag or tight “feeling”.

Axial preload may also occur where housing fits are too tight to allow the floating position bearing to function. This can be checked by loosening of the housing bolts, which should immediately free the preload and improve the tone quality and speed of the unit.

Heating from radial preload is caused by insufficient internal play in the bearings and is rare. A false condition paralleling this is some­times created by mounting a close-fitting bearing on an oversize shaft. The inner ring expands from being forced on the shaft, takes up all internal play and establishes a preload. This condition, however, rarely reaches the assembled unit stage where a hand test of outer ring move­ment is made to check lateral play and free turning condition of the bearing.

The more common electrical troubles of noise-making proportions may come from factors such as an improperly designed brush or com­mutator, excessively small air gap, an insufficient number of armature slots for the application, an intense field with magnetic flux currents of disturbing values, and skewed armature or rotor cores. These may or may not produce vibration noises that are transmitted to the ball bearings. Sometimes, to prevent electrical noise from complicating the problem, it is of value to drive the unit from another motor.