Fake coin problems using balance scales to identify one fake coin and its type among n coins (n > 2) were solved by Dyson in 1946. Dyson gave adaptive solutions with the minimum number of weighings where later weighings may be dependent on results of past weighings. In 2003 Born et al. gave non-adaptive solutions where all weighings are predetermined. Both solutions require the computation of a Dyson set, which is a list of placement of each coin for each weighing. The computation of a Dyson set requires substantial amount of time when n gets larger. We present a bounded partition approach to the fake coin problems without computing any Dyson set. Our approach uses bounded partition of coins recursively until the problem size may be small enough to use at most one weighing.