i'm as much a fan of pushing limits as the next person...but:
keep in mind the closer you run to the spec of the parts, the likelihood of failure increases dramatically (for a number of reasons). especially with cheap parts, many of which may be clones of other parts, and made with no or very little qc. or may even be the reject parts (that actually aren't even capable of the ratings in the spec sheet!) that the factory "threw away" but someone saved and sold as new good parts.... or the parts may not even be the parts they're marked as, and as such you have no idea of their limitations..but they're quite likely lower than the parts they're marked as.
(yes, these things happen)
so just because the parts are "rated" for a certain voltage, doesn't make that the limit of the unit they're in. the limit is much lower, to ensure reliability. the cheaper the unit was, the lower the limit should be (the greater the difference between the spec'd limit and the usage limit).
if you don't mind stuff unexpectedly failing, and riding is just for fun and you don't have to get somewhere on the bike/etc., or you don't need the motor to do what you do with it, then it's "safe enough" to try it out, with the caveat that something will eventually go wrong (how long that will take depends on too many things to give a number).
if reliability is important, then don't run it higher than it was originally "meant" for. (too often, even *that* is closer to the failure point than it ought to be.)
ok, all that said:
there are several limitng components.
capacitors that are wired across the battery voltage lines, and at the fets.
fets
lvps input stage (lvps is the low voltage power supply that converts battery voltage to the 12v and 5v that runs the non-fet parts of the controller).
this last part is often neglected when overvolting controllers, and you may find a controller that has nice high voltage caps, even the fets...but the lvps blows up a few volts above the "max" controller voltage.
