The conventional side-pull brake is light, durable, provides moderately good braking, and can ``follow'' rim irregularities, allowing a rider with a broken spoke to finish a ride with only minor rubbing when the brake is released.
The modern design was introduced by Campagnolo [[when]]. In incorporates many subtle design features. In order to achieve light weight, several parts are highly stressed. Thus, service and design changes which do not allow for those stresses can lead to brake failure.
This note describes several design points of the side-pull brake with special atention to areas around the brake bolt. It discusses the structure of the brake and some common situtations which increase the chance of failure.
The following discussion focuses mostly on six brake parts having to do with the pivot and its attachment to the bike. Except as noted, this discussion also focuses on the front brake. The six parts are: the ``front'' and ``rear'' arms the nut, locknut, and thrust washer; and the brake bolt. The diagram below shows these parts schematically.
Several bushings are missing from this diagram, as well as the various non-pivot parts such as the pads, return spring, and quick-release, and cable attachment.
The diagrams in this note show large clearances between various parts. In practice, the parts should fit as tightly as possible without binding so that forces and wear are distributed over the largest possible area. In particular, the nut and locknut should be adjusted so that the arms pivot freely but without fore-aft clearance.
Most pivots -- hinges, door knobs, and so on -- are built as a sleeve running on a shaft, with contact between the sleeve and the shaft. If side-pull brakes were built this way, they might look like the following diagram, where the black dots indicate a point of contact and high load.
Building a brake this way causes high brake bolt loads. Suppose each arm is subject to pulling force F at distance l from the center of the brake bolt, and the arm is of thickness t. In that case, the torque of each brake arm would be Fl. To resist that, the load on each part of the bearing would be Fl/t. Furthermore, the bending moment on the brake bolt would be Fl from each arm, thus 2Fl from both arms together.
For example, stopping a 100 kgf bike and rider at 1 g deceleration puts a load F of 50 kgf on each arm. A Campagnolo Record brake has dimensions of l = 50mm and t = 7.3mm. Given these dimensions and loads, stopping causes bearing forces of about 340 kgf and brake bolt bending torque of 5 m-kgf.
High bearing loads cause friction, flexing, and rapid wear. High torque on the brake bolt causes it to bend. If the load is very high, the bolt can bend permanently. If the torque is less, bending is elastic and the bolt springs back when the brake is released, but the brake bolt can gradually fatigue and then fail.
Overload problems can be reduced by using larger and heavier components, but at correspnding increase in weight and size. Another choice is to carry the loads as thrust loads on the shoulders of the arms, as shown in the following diagram.
Consider one arm of a brake of the same dimensions as above a shoulder separation of s, and a distance from the lower shoulder to the pivot bolt center of b. (In this case, 2b = s; in general the bolt need not be centered between the shoulders.) The torque about the lower shoulder of the front arm is F(l-b). To resist that torque, the opposing force on the upper shoulder is F(l-b)/s. The force on the lower shoulder is thus F + F(l-b)/s, which simplifies to F(l-b+s)/s. That is the load for each arm; since the rear arm presses on the front arm, the total force on the thrust washer is double that 2F(l-b+s)/s.
As described so far, the brake forces do not directly bend the brake bolt. However, the nuts on the end of the brake must resist the bearing force of 2F(l-b+s)/s. The force is applied at radius b, so the torque is 2Fb(l-b+s)/s.
The dimension s is about 17mm on a Campagnolo Record brake, and b is half that or about 8.5mm. Thus, for a stop as described for the sleeve bearing, so the bearing force is 2Fb(l-b+s)/s = 345 kgf, with a bending torque 2Fb(l-b+s)/s of 2.9 m-kgf on the brake bolt. Thus, the bearing force is nearly identical to a sleeve bearing, but the bolt bending torque is only about 60% of a sleeve-type bearing.
With the shoulder-bearing design, the bearing forces are about the same but the loads are applied at a much larger radius -- 8.5mm for the Campagnolo brake vs. a brake bolt radius of about 2.9 mm. Thus, all else equal, the shoulder-bearing design has about three times the bearing drag. [[Is there an easy way to approximate how much force is lost on the way to the pads?]].
When the nut and locknut are tightened, they push against each other so that the nut is being pushed off the front end of the brake bolt, while the locknut is being pushed towards the rear of the brake bolt. As a result, the thread clearances for the nut and locknut appear on opposite sides.
When the brake is applied, both nut and locknut are pushed in the same direction. The locknut supports load by reducing the preload on its threads and by transferring forces to the nut. The nut is held tightly in the direction of braking forces, and as the load on it increases, the nut is driven harder on to the threads.
The brake load pushes forward on the bottom of the nut/locknut pair. The threads resist the force, so the top of the nut/locknut pair tends to move to the rear. Thus, the top sees a weaker but opposite force from the bottoms; on the top, the nut preload is reduced, and the locknut load is increased.
Since the nut/locknut pair is loaded heavily (about 350 kgf in the above example), it is vital that the nuts are tightened securely against each other to prevent them from creeping loose under varying load.
The brake bolt holds the brake arms against a flange on the brake bolt, and the flange holds the brake on the bicycle. In principle, the flange can be a separate piece of material from the brake bolt. Some low-cost brakes are built that way.
However, the brake bolt is highly stressed and flexes under load. Thus, brake bolt can fail if it is used hard and has not been designed for the applied loads. One note is the brake bolt should have radiused curves where the bolt swells to make the flange, in order to avoid stress raisers. Raidused curves are not shown but would be used where the four dotted circles appear. Using a separate bolt and flange does not prevent bolt flexing but does make it hard to avoid stress raisers.
The brake arms press on the top front of the flange. For the 100kgf/1g/Campagnolo Record brake example used above, the brake arms push on the flange with a force of about 245kgf. That force is countered by tension in the brake bolt. There is also bending around the base of the brake bolt, about 2.9 m-kgf in the above example. Thus, the bolt is under both bending and tension, hence the need for a gradual radius.
A similar consideration applies to the back side of the flange. When the brake is applied, the preload is reduced on the lower part of the flange and the load is increased on the upper part. The brake bolt must be tight enough in the bicycle frame or fork that the reduced preload on the flange does not go to zero, otherwise the bolt tension must rise to support the load, which can overload the bolt.
The back of the flange is only about 7.2mm radius, while the front is about 8.5. The decreased flange radius allows the flange to clear headset bearing cups on most forks, but the reduction in size proportionately increases the brake bolt preload which is needed to keep the flange from lifting.
The discussion so far has focused on front brakes. The rear is similar but is not the same because the arms are pressed towards the flange, whereas on the front, the arms are pulled away from the flange.
The differences between front and rear become important if we try to, say, reduce the bolt torque by using an offset pivot. Note that this is not a classic ``drop bolt'': a ``drop bolt'' moves the whole brake down, while a changed pivot location leaves the shoulder locations unchanged.
An offset pivot puts the nuts more in line with the bearing forces and thus reduces the lever arm b and the bending moment on the bolt.
If the same brake is used on the rear, however, the bearing forces are reversed, and so the same modified brake on the rear increases the effective lever arm and thus increases the bending moment on the bolt.
A small loss in rear brake strength and durability to improve the front brake's strength and durability may be acceptable in many cases: rear brakes typically generate much lower forces. However, in tandem applications the rear brake may generate just as high braking forces and may be used more often than the front brake. However, building different front and rear brakes causes manufacturing and repair part stocking problems and also could endanger the rider if the front and rear brakes are mistakenly interchanged.