See also also bike pic for other bike-related technical discussions and more failed parts.
These brakes are different than most sidepull brakes in several minor ways and at least one significant way:
The remainder of this note focuses on the bearings.
The brakes shown here were used for an estimated 15,000 miles over seven years by a light (65kgf) rider, including wet and sloppy riding. The brakes would not remain centered, so the brake pads rubbed on the rim even when the brake was released. A similar problem occurs with conventional sidepull brakes when the return spring ends are not lubricated. However, the use of hidden coil springs should have eliminated that problem. However, mis-centering was so common the owner carried a brake centering wrench on ordinary rides.
Disassembly shows the brakes use ball bearings. The bearings are exposed to environmental contamination, both because the seal is simply a close-fit plastic lip, and because the underside of the brake is bathed continuously in gritty water and mud during wet rides.
One potential problem is grit can get trapped between balls and races, forming a "hill" that interferes with free motion of the bearings. If one arm runs freely and the other does not, that would interfere with centering.
A definite problem is the rear arm bearing races (this is a front brake) have developed divots (pits, dents, ...) where the bearings roll. The divots can trap the arm position, similar to an "indexed headset".
The divots are due in part to softness and thinness of the bearing races compared to the loads placed on the bearings -- this is demonstrated by the rearmost race, which is dented through and bulges out on the back, even though the back side sits on a steel surface. All bearing races exhibt some grooving and slight pitting (visible under 10x magnification). which may be due to wear from grit intrusion and/or false brinelling, as occurs in headsets
(Note this brake was not disassembled in service, so bearing dents are not related to over-tightening unless it occured at the factory.)
The backs of some races are also scored and scuffed, indicating the races were moving relative to their backing surfaces, when they should have been stationary (if the ball bearings were doing their job).
The hidden spring and bearing design may exagerate bearing and other loads. In a conventional sidepull brake, braking loads are carried axially. See http://pardo.net/bike/pic/fail-007/000.html for bearing layout and loading on conventional sidepulls.
In these SunTour brakes, the loads between the two arms appear to be taken radially on the bolt, rather than axially on the shoulders of the arms. This appears to be a consequence of fitting the springs inside the arms, although the same design could carry loads on the "rings" outside the springs. The presumed bending load on the center bolt may be a safety issue: bearing dents are annoying, but a failed bolt might eliminate braking entirely.
It is plausible that self-centering on these brakes could be restored by replacing the ball and carrier with an ~2mm washer. That leaves the question of whether there are bending loads on the center bolt and thus safety issues.
See also bike pic for other technical discussions and more failed parts.