Push/Pull Method for Preloading 2-TS Spindles: Achieving Preload

Feb. 25, 2021
An accurate, mounted preload setting can be achieved for bearings in a 2-TS spindle – and the mounting method will determine repeatability for speed, stiffness, and accuracy requirements.

This is the second of two reports. See Push/Pull Method for Preloading 2-TS Spindles: Why Preload?

Preload force is very sensitive to the preload dimensional setting. The setting tolerance for a 2-TS Tapered Roller Bearing (TRB) arrangement is +/- 0.010mm.  The accuracy that a preload setting can achieve repeatably depends on the method used to set preload, and there are several established methods currently in use. To put the push/pull method into perspective it is necessary to understand these methods and their limitations. 

Setting TRB spindles: Lock-nut thread pitch. Because of errors on the thread pitch form and backlash between the male/female threads, and the reliance on the technician’s experience and feel to achieve the proper mounted preload, this method is not accurate enough and not recommended for machine tool applications.

As an example, a technician is attempting to set a 2-TS IM spindle with -0.050 mm of mounted preload, how much would he have to turn a 1.5-mm pitch thread locknut if he measured 0.150 mm of endplay? The cones need to be brought together 0.200 mm, so for every rotation of the locknut the cones are brought together 1.500 mm or 360°/1.5 = 240° per millimeter x 0.200 mm = 48° or and 1/8th of a turn.

Several methods exist for achieving the desired mounted setting for a 2-TS spindle depending on the level of accuracy and repeatability required by the application. The most common method used in general industrial applications to achieve a mounted setting (endplay or preload) for a 2-TS IM TRB arrangement is by using the thread pitch of the locknut to set the bearings. In this method the technician tightens the locknut, measures the endplay with a dial indicator, and rotates the locknut based on thread pitch and the difference between the measured and target settings.

Setting TRB spindles: Custom-ground spacer. A second method for achieving a desired preload setting is to calculate the length of a cup or cone spacer based on shaft and housing fits. The spacer length can be determined by calculating the lateral loss (telescoping) due to the cup and cone fits, and either adding or subtracting them from the bench end play (BEP) and desired mounted setting (depending on the mounting.) This method is common with TDOs and TDIs that come with a factory-ground spacer with a tightly controlled Bench End Play (BEP.)

This method requires the technician to accurately measure the cup OD, cone bore, and both bearing seats. In the case of the TDO and TDI, the BEP is provided but the spacer tolerance needs to be taken into consideration. If this method is being used for a 2-TS spindle, then the gap between the cones or cups needs to be calculated as well.

Small errors in the measurements described above result in large errors in the calculated lateral loss and the mounted setting. It is for this reason that this method is not accurate enough and not recommended for machine tool applications.

Setting TRB spindles: Push/pull method. The goal of the push/pull method of setting preload is to determine the length of either a cone backface or a cup backface spacer for a predetermined preload for TS-IM or TS-DM spindles, respectfully. The following discusses the procedure that consists of applying incremental axial loads to the spin in both directions, typically with a hydraulic piston, and plotting the force versus shaft/housing displacement with a gauge spacer that is known to yield endplay (EP).

The best way to explain the push/pull method is to walk through an example for a basic spindle used in machine tool applications.

Example. We will assume a 2TS-IM spindle arrangement mounted onto a rigid shaft and into a rigid housing.  We will also assume a line-to-line fit between the cones and the shaft as well as between the cups and the housing. It is worth noting that typical fit-ups for these types of spindles are between 25 and 50 microns of interference. This will cause the bearing stand (overall width from cone to cup backface) to grow and is referred to as lateral loss of the bearing. The push/pull axial displacement measurements are performed with the bearing mounted into the spindle, so any lateral loss is taken into account.

Other assumptions for this example include:

  Proper bearing has been selected for this application;

  Bearing seat size and geometric form meets Timken’s recommendations;

  Preload setting has been determined for the spindle through application analysis;

  The target preload setting is 0.050 mm (50 μ);

  A gauge spacer that yields an endplay of between +0.025 mm (25 μ) and +0.100 mm (100 μ);

  Spacer on the cone backface,accessible when shaft locknut or endcap is removed (See Figure 9);

  Shaft/cone temperatures are the same as the housing/cone temperatures, no thermal effects.

The spindle. As mentioned in the previous installment, this 2TS-IM spindle is very popular and widely used not only for machine tool applications but can be found in almost all applications, including industrial, automotive, off-highway, rail and many more. The big difference is the precision and the desired mounted bearing setting.

The housing provides mounting shoulders for both cup backfaces. They are typically captured or “pinched” in the housing with bolt-on end caps. The preload force when the bearings are set will theoretically hold the cups in the housing, however, endcaps will secure them in place and will prevent them from unseating under extremely high axial loads.

The spacer. The shaft has a single shoulder for the left bearing and right bearing, and a locknut and spacer on the left to set the bearing. Without a spacer, the bearings would be set in a preload that far exceeds the target value if the final nut torque alone was used, so a spacer is introduced to act as a stop for the locknut. If the spacer length exceeds beyond the cone backface the bearing will be set in end play. (See Figures 9 and 10.)

The bearing. The bearing that has been selected for this spindle is a precision Class 3 bearing 543085 cone and 543085 cup. The bearing has a 290.010-mm OD with a 215.900-mm bore and an overall length of 31.75 mm. (See Figure 11.) Because there are so many design parameters that contribute to how a bearing will perform in an application, it is important -- not just for machine tool applications, but for all bearings -- to consult with your bearing company representative to ensure the bearings selected have good high-speed characteristics. Just because the bore, OD, and width meet the application needs, there are other bearing characteristics that must also be considered. 

Calculating the length of the EP spacer. The last thing that is needed prior to performing the push/pull test is to determine the length of the gauge spacer that will yield endplay when the bearings are mounted into the spindle. Calculating the length of the gauge endplay spacer requires the technician to measure the distance between the housing shoulder, the bearing stand height of the front and rear bearings, and the length from the shaft shoulder to the shaft face where the spacer seats. If one or both bearings have an interference fit on the shaft or into the housing, the resulting lateral loss must be added to the bearing stand.
  LEP = BSR + LH + BSF + ΔEP – LS  (See Figure 12.)
  LEP = Endplay spacer length
  BSR = 31.745mm = Bearing Width, Rear – Measured under a 100N axial load, lateral loss due to any interference fit would need to be added to the bearing stand height
  LH = 160.000mm = Length between housing shoulders – Measured
  BSF = 31.747mm = Bearing Width, Front – Measured under a 100N axial load, lateral loss due to any interference fit would need to be added to the bearing stand height
  LS = 214.017 = Shaft length (shoulder to face) – Measured
  ΔEP = Desired endplay = .050mm
  LSpacer = 9.525mm

The push/pull test set-up. Performing the push/pull test requires means both for applying an axial load to the spindle shaft and measuring it.  A hydraulic piston and a load cell are both commercially available and cost-effective. The displacement between the shaft and housing must be measured to +/- 1μm. An LVDT is capable of providing a measurement to meet this requirement. (See Figure 13.)

It is critical that the shaft be rotated several times prior to each measurement to ensure the rollers are seated properly and the bearings have an adequate coating of grease or lube oil. Not doing so will result in an error in the displacement measurement. It is also important that the housing is affixed to a rigid base that does not deform under the applied axial loads, misalignment between the piston and the spindle shaft walso will result in measurement errors.

It is also important that the housing is affixed to a rigid base that does not deform under the applied axial loads. Misalignment between the piston and the spindle shaft also will result in measurement errors.

The push/pull test. Before starting the test, the maximum push/pull force needs to be established along with the incremental push/pull force. A good value for the maximum push/pull force can be 25% of the bearing Ca90 (axial dynamic capacity based on 90 million revolutions for L10 life.)

For the 543085-543085 Ca90 = 37,900N that results in a maximum push/pull force of 9,475N, 10,000N was selected for this example. Twenty measurements in each direction yields an incremental push/pull force of 500 N.

Start by applying a 500-N axial push force to the front of the spindle, measure, and record displacement between shaft and housing at each load increment. The spindle should be rotated adequately at each push increment before measurement is taken. It is important to rotate the spindle at each push/pull increment to seat the rollers. If the rollers are not seated, the axial load versus deflection points will not align well and skew the results. The spindle should be at rest when the axial displacement measurement is taken.

Once all twenty measurements are taken, flip the spindle and apply the same axial loads to the rear of the spindle shaft. Repeat the steps above and record axial loads as “+” values when recording the data. It is recommended to repeat the test several times to ensure that everything is seated properly, and the results are repeatable. This is especially true if there are inconsistencies in the push/pull plots which will be covered in the next section.

Creating a push/pull plot. With all the data recorded, a push/pull plot needs to be created to determine the spindle’s endplay. The “raw” or recorded displacements should be converted into absolute displacements using the following conversion method.

Take each recoded Push Displacement and subtract the Maximum Push Displacement (= 0.042 mm) to yield the Absolute Displacement. Next, the Recorded Pull Displacements need to be converted into Absolute Pull Displacement. To do this, add the Maximum Push Displacement to each of the recorded values. (See Table 1.) The next step is to plot the absolute axial displacement versus the axial push/pull load. (See Figure 14.)

Determine the end play. The end play of the spindle is half the plateau between the Y-intercept of the push and pull portions of the plotted points.  The plateau width in Figure 14 is 0.99 mm; half of this value is 0.050 mm, which is the current mounted endplay. Visually it is very symmetrical about the Y-axis, 0.048 mm versus 0.051 mm, a difference of only 3 μm.

The push/pull plateau should be very symmetrical about the Y-axis. Elastic deformation of the shaft and housing along with gauge error can account for a few microns of asymmetry. If the asymmetry of the plateau exceeds 5 μm there is a good possibility that the bearings are not seated correctly, and that looseness is impacting the push/pull test.  It is suggested to repeat the test in this case. 

Determine the preload spacer length. The last step is to calculate the length of the spacer to yield the desired preload of 0.050 mm (50 μ) when seated by the locknut; this is the preload spacer length. It is the endplay spacer length minus the preload setting, minus the measured endplay, and it is determined by the following equation:
  LPL = LEP – ΔPL – EP = Preload spacer length
  LEP = Endplay spacer length = 9.525 mm
  ΔPL = Desired preload = 0.50 mm
  EP = Measured end play = 0.050 mm
  LPL = 9.525 mm – 0.050 mm – 0.050 mm = 9.425 mm

Because machine tool spindles push the boundaries of speed, precision, and repeatability, it only makes sense to use a mounting method capable of consistently preloading the spindle bearings. The tooling and gauging necessary to perform a spindle push/pull test are all readily available and relatively inexpensive. They can be used on many different types of spindles to help set and verify preload.

The axial displacement versus the axial push/pull load plot used to determine the mounted end play of the spindle also can be a valuable tool to visualize discontinuities and asymmetrical irregularities that would indicate that the bearings are not mounted or seated properly.

This is the second of two reports. See Push/Pull Method for Preloading 2-TS Spindles: Why Preload?

Eric Faust is an Application Engineering Specialist with The Timken Company. Contact him at [email protected]