In our previous work we determined the weak capacity region for the broadcast phase of two-phase bidirectional relay channel. It turned out that the set of achievable rates obtained by optimizing over the two communication phases exceeds that obtained by using the network coding principle, i.e. by applying XOR to the decoded messages. In this paper we supplement our result by a proof of the strong converse with respect to the maximum error probability to the coding theorem for the broadcast phase. This result implies that the capacity region of that phase remains constant for a certain range of values of average error parameters [epsilon(1), epsilon(2)].