>>10681712/2
Many cantilever brakes have the pads mounted on long bars that exert sometimes considerable leverage on the brake, its mounting and the fork blade/chainstay, causing it to twist. This means both a reduction of braking force and the pad not fully contacting the rim. This is why poor cantilever designs often need much more pad toe in than caliper or V.
This can be mitigated with better choice of pads and, in the case of the mounts/frame flexing, a so called brake booster that is basically a stamped honeycomb pattern bridge that sits outside the brake at the end of the mounting pegs and prevents them from pointing in when the brake is applied.
Brake boosters cost next to nothing, as do pads. Magura make brake boosters still and KoolStop make good pads.
A second issue (in the front, mainly) is that the cantilever wire is often hung from the top of the headset instead of connected directly to the brake as with calipers or V. This means a pull on the lever wants to bend the fork, pegs and brake assembly upwards, and every little bit they do lessens brake force - whereas in turn it increases when they spring back. It is a positive feedback loop of these properties that cause the infamous cantilever brake judder. Even on assemblies that do not experience the runaway feedback can suffer poor brake performance and "twitchy" modulation.
One solution/mitigation strategy is to use a cantilever cable hanger that attaches to the fork crown instead of the headset stack, but that requires a drilled crown. The upside is that they cost peanuts.
If, after all this, you're still not happy I suggest getting better cantilevers and carry the upgraded pads, hanger and booster over. It will be a much better brake assembly than a brand new V brake assembly and levers without the upgrades, and still cost less over all. Even if you end up buying new cantilevers.
