I had first examined Mike Stephenson's report and Michael Mayda's test results. In addition to Stephenson's report of the nonlinearity of the results, I observed that Michael Mayda's test graphs show that on several instances when the brake lever input force was reduced the corresponding reduction in braking force was delayed, sometimes for several seconds. That looked like a brake sticking or jamming in some way.
After the test ride in which the rear brake varied between one condition of powerful and approximately linear operation and another condition of locking the wheel under light application, I examined the brakes to see what might cause this condition. Naturally, I had not tested the front brake once the rear brake showed this tendency. That would be far too dangerous.
I first examined the unusual cable linkage, as shown in the attached drawings. The manufacturers claim that this linkage produces 40% more braking force than do conventional straddle-wire linkages. I analyzed this claim by comparing the theoretical operation of the linkage against straddle-wire linkages with different lengths of straddle wires. The shapes of the linkages are shown in the drawing named Cannondale Brake Linkage, with the Cannondale linkage shown on one side and several examples of straddle-wire linkages on the other side. I calculated the braking forces for all the variations, as shown on the calculation sheet. With straddle-wire linkages, the braking force is highest with a short wire and decreases as the straddle wire is lengthened. The Force-40 linkage produces only half the force of a very short straddle-wire linkage, but when the straddle wire is lengthened to what is typical, about where the brake mounting bolt hole is, the Force-40 linkage produces about 40% more force than the straddle-wire linkage. Since some straddle-wire linkages produce more force than the Force-40 and some produce less, the power of the Force-40 linkage, by itself and when operating as designed, does not appear to have caused the braking problem. I investigated the operation of the linkage, and all parts appear to be working as intended. Therefore, I do not think that the Force-40 linkage, by itself, was a cause of the braking problem.
However, the brake arms appeared to have too much clearance in their bearings, so that the axis of the brake arm could be made to assume an angle to the axis of the pivot on which the brake arm rotated. In short, the brake arms wobbled on their pivots. At one extreme of this movement, the leading end of the brake pad contacted the rim first, at the other extreme of this movement the trailing end contacted the rim first. The normal amount of movement appears to be about half a millimeter at each end of the brake block, but the brake-arm holddown bolt on the left front brake arm is not tight and the movement there appears to be about one millimeter at each end of the brake block. I disassembled the left rear brake arm and from that I prepared the attached drawing showing the brake block as it is when wobbled against the direction of rim motion.
The wobbling movement can have either of two effects, depending on the direction in which the brake arm has wobbled previous to the time the brake is applied, which appears to be a random matter. If the brake arm has wobbled in the direction of rim motion so that the motion of the rim will not drag it further, then the brake operates normally. However, if the brake arm has wobbled in the direction opposite to rim motion, then it has the potential to be pushed in the direction of the rim motion by the force of the brake block. The first motion to occur is to the position when the face of the brake block is parallel to the rim. This motion allows the brake arm to move closer to the rim, as it is, in any case, propelled by the tension in the brake wire produced by the rider's hand squeezing the brake lever. However, this is not a stable position, because the friction between the brake block and the moving rim still tries to push the brake block in the direction of rim motion. If this friction is sufficiently high, the brake arm will wobble in the direction of rim motion as far as the looseness of the brake arm's bearings allow. The looser the bearings, the more wobble. As the diagram shows, this causes the brake block to pivot on its leading edge, pressing this closely against the rim and, because every force has an equal but opposite reaction, pushing the brake arm outward. With a normal brake, this movement of the brake arm is resisted by the bearing itself, being transferred directly to the stud on which the brake arm pivots. However, with a loose bearing, until there has been sufficient motion to take up all the clearance in the bearing, this wobble produces an outward force on the brake arm. If the brake arm were free to move outward, this would cause no problem except uneven wear of the brake block.
However, the brake arm is not free to move outward. Brake cable systems have a large amount of frictional loss, sometimes reported as 50% for the older cables. Presumably the modern cables with slick plastic liners perform better. If the brake cable has a frictional loss of 40%, then for every 100 pounds of tension applied at the brake lever, only 60 pounds drives the brake arms inward. That is, with 100 pounds applied it takes 40 pounds just to move the brake wire inside its housing. If the brake arm is to move outward against the applied tension, it must work against the 100 pounds input at the brake lever plus the 40 pounds of friction. Therefore, when a brake arm is applied with 60 pounds of tension (100 pounds input - 40 pounds frictional loss), it requires 156 pounds of force at the brake arm (100 pounds at the input end + 56 pounds (140 x 40%) of frictional loss) to make the brake arm move outwards. If the loss in the brake cable is only 20%, then a brake that is applied with 80 pounds at the brake arm requires 124 pounds of force at the brake arm to make it move outward. In addition to the friction loss in the cable there is some in the bearings of the brake arm itself. Put simply, the force required to move a brake arm outwards against the force of application of the brake runs from over 1.5 times the initial application force to over 2.6 times, depending on the frictional losses in the brake cable.
Therefore, the brake arm will not move outward until the outward force developed by the contact between the brake block and the rim is much greater than the applied force. Because the braking effect, the drag produced by the brake, is approximately proportional to the contact force between the brake block and the rim, the drag on the wheel increases by the same proportion as the increase in force required to move the brake arm outward. Therefore, if the brake is initially applied when the brake arm has previously wobbled in the direction against the rim motion, the brake arm will tend to wobble in the direction of rim motion, an action that, if the amount of wobble permitted by the bearing on which the brake arm pivots is sufficient, will cause the braking effect to suddenly become much greater than was intended by the rider. In some cases, that greatly increased braking effect will either skid a rear wheel or, if it occurs on the front brake, will throw the rider over the handlebars.
Another consideration is the degree to which the brake block has worn. Even when installed differently, brakeblocks soon wear parallel to the rim. With many brakes, this wear to parallel is the assumed normal operating condition. Of course, this is parallel when in the operating position. However, many other brakes, these included, are made with a mechanism that allows a new brake block to be adjusted to be parallel to the rim. If a brake with loose brake-arm pivot bearings has operated for some time, the brake block will have worn parallel to the rim when the brake arm has wobbled in the direction of rim motion. Then it will be impossible for the leading end of the brake block to make the major contact with the rim, but the force of contact will be distributed along the block instead. The combination of contact all along the brake block and the stiffness when the looseness in the bearings has been taken up largely prevent the jamming phenomenon from occuring. However, when adjusting or readjusting the position of the brake block, the mechanic is liable to adjust the position when the wobble of the brake block is in the center position, because that is the position in which the brake arm will rest whenever it is not pulled by the motion of the rim or jiggled by the vibration of riding. If the mechanic adjusts the brake block so that it is parallel to the rim with the brake arm anywhere except wobbled as much as it will move in the direction of rim motion, there exists the possibility that the jamming effect will occur until the brake block wears parallel to the rim when wobbled as far as it will move in the direction of rim motion.
There has been much advice given about adjusting brake blocks. The standard method is to adjust each brake block for height and angle so that it does not overlap the side of the rim. Many brakes, including those considered best in the world, do not allow adjustment for making the brake block parallel to the rim. These brakes are made sufficiently accurately that the brake block face is sufficiently parallel to the rim, but if this is not so it is accepted that the brake block will quickly wear parallel to the rim. Other brakes, including these, provide an adjustment by which the face of the brake block may be made parallel with the rim. Practically every cyclist, at some time or another, has experienced brake judder or brake squeal. When applied, the brake produces a vibration, varying from a few per second to many thousands. Efforts to reduce this tendency have caused many people, perfectionists perhaps, to advise adjusting the brake blocks with toe-in, so that the trailing ends contact the rim before the leading ends. I have rejected this advice because the toe-in quickly disappears, demonstrating that toe-in, by itself, is not required to avoid these effects. It may be that squeal, judder, and jamming are similar vibrations whose frequency depends on the amount of wobble allowed by the pivot bearings; the more wobble, the lower the frequency but the stronger the force. However, if the initial adjustment is made with more toe-in than the maximum amount of wobble in the brake arm bearings, then the leading end of the brake block cannot touch the rim before the trailing end. As the brake block wears in service, it will wear parallel to the rim in the operating position but it will not wear to the shape that allows the leading end to contact the rim before the trailing end. However, as stated above, if the brake block is adjusted at what seems to be parallel to the rim, the wobble in the pivot bearing may allow the brake block to assume the position in which the leading end touches the rim first, thus incurring the risk of squeal, judder, or jamming the wheel until the brake block wears into a safer orientation.
Initially adjusting the brake blocks to have an obvious toe-in, therefore, would avoid the danger of jamming the wheel, and possibly of having judder or squeal, caused by the wobble that is allowed by the pivot bearings. Thereafter, even though the brake blocks will wear to be apparently parallel to the rim, they won't wear to the shape that allows the leading end to contact the rim first and allows the jamming to occur. The bicycle is practically new and the brake blocks show no signs of wear. The front tire still shows molding flash along the center of its tread. With high-pressure tires this indicates only about 100 miles of use; with these lower-pressure tires it probably indicates no more than twice that. The accident, therefore, in one sense, was caused because this was a new bicycle that had not been adjusted in the non-standard way that is required, for safety, by the particular design of brake that it uses.
The brake construction revealed why the left front brake arm was so much looser than the others. I inspected the construction by disassembling the left rear brake arm, leaving the front brake untouched until others could be present during the inspection. See the attached drawing. All the brake arms appear to be similar, so I presume that the construction of the left rear arm is similar to the construction of the left front arm, except for being a mirror image. The arm is retained on the brake boss in the conventional manner by the brake arm mounting bolt, which screws into the end of the hollow stud. However, the arm is not mounted directly on the mounting stud; there is another sleeve between them. This intermediate sleeve serves to allow an easy adjustment of the position of the brake arm when the brakes are released. The spiral spring that pulls the brake arm away from the rim has a pin at each end. In the standard design, the pin on one end fits into a hole in the brake arm while the pin in the other end fits into a hole in the boss of the stud. The adjustment of such brake springs is crude, generally by having several holes for the end of the spring, any one of which may be used. In the CODA design, the hole in the boss is replaced with a hole in a disk. This disk has a hole so it can be slipped over the brake stud and its circumference is machined into a 13 mm hex. Therefore, by turning the disk with a cone wrench, the force of the spring can be adjusted to position the brake block at the correct distance from the rim when the brake is released. However, this disk would rotate, thus releasing the spring, unless it were locked. It is locked by the intermediate sleeve, which transmits the axial force developed by the head of the brake arm retaining bolt as it is tightened. Then there is a metal-to-metal grip between the base of the boss, the disk, the intermediate sleeve, and the mounting bolt. This grip locks the intermediate sleeve in place as well as locking the spring-adjusting disk against rotation. The intermediate sleeve is slightly longer than the arm is thick, so that the arm is free to rotate on the intermediate sleeve with only slight axial play. This essentially duplicates the normal brake. However, in this brake, the clearance between the outside of the intermediate sleeve and the inside of the brake arm is too great. This allows the brake arm to wobble on the sleeve, thus allowing the locking tendency to develop. This is either a manufacturing or a design error, depending on whether or not the parts are within the designer's tolerances.
However, there is another error in the design. If the brake arm mounting bolt is not fully tight two things happen. The spring adjusting bushing rotates so that the brake pad is not pulled away from the rim, but remains in light contact because it is driven by the opposite spring that is still working. This is merely an inconvenience, as it is with conventional brakes. The danger is that then the intermediate sleeve can wobble on the stud. Then the brake arm, and with it the brake block, can wobble about twice as much because the amount of wobble is the sum of the wobble between the stud and the intermediate sleeve plus the wobble between the intermediate sleeve and the brake arm. This extra amount of wobble, if in a front brake, could well cause the difference between a brake that throws the cyclist over the handlebars and one that does not.
I inspected the left front brake arm retaining bolt, and found that it was not tight. I left it in the same position as I had found it. This accounts for the extra wobble that I had detected in that brake arm. I presume that the front brake is even more subject to locking than the rear brake, making that front brake extremely dangerous.
I conclude that the brake is unreasonably dangerous for three reasons.
The first reason is that the brake arm is not fitted closely enough to the intermediate sleeve to reduce wobble between the brake arm and the intermediate sleeve sufficiently to prevent this locking tendency. This probably could be remedied by better manufacturing practices to closer tolerances.
The second reason is that if the brake mounting bolt becomes loose the added wobble between the stud and the intermediate sleeve is added to that between the intermediate sleeve and the brake arm. Unless the stud and the intermediate sleeve can be manufactured so that there is substantially no clearance between them, which would require substantially higher cost, this is a defect that is inherent in the design and cannot be remedied.
The third reason is that, while the brake may be reasonably safe after the brake blocks have worn parallel to the rim in the normal operating position, or have been adjusted to have a decided toe-in, it is unsafe when adjusted in the normal manner to have the brake blocks parallel to the rim at the time of adjustment.
John Forester, M.S., P.E.
Return to: John Forester's Home Page Up: Bicycle Engineering