An investigation has been made to find out the impact of channel variance mismatch to the performance serial turbo codes (SCCC) equipping with talwar penalty function over impulsive channels. The impulsive noise channel is blamed to be the culprit of signal impairment and causes severe performance degradation to modern SCCC which traditionally has been designed to perform optimally in additive white Gaussian noise (AWGN) channel. Undoubtedly, it is a major limiting factors for systems such as Powerline Communication Channel (PLC), digital subscriber line (DSL) and digital TV. SCCC have been proven to approach Shannon Capacity bound in Gaussian channel. The constituent decoders are implemented using the log- MAP and MAX-LOG-MAP algorithms. It is shown that serial turbo decoders with talwar penalty function that are implemented with the LOG-MAP algorithm are sensitive to varying channel variance, where underestimation and overestimation are almost equally detrimental over impulsive noise channel. On the other hand, the performance of the MAX-LOG-MAP based serial turbo with decoders with talwar penalty function exhibit complete independence with respect to errors in the variance estimate in impulsive channel. The study also found that using talwar penalty function can significantly improve decoding capabilities of SCCC in impulse noise channel under varying channel variance environment. Therefore Shannon predicted excellent performance of turbo codes can be perfectly obtained from such impulsive channel.