The Petrova (2000) model to calculate pulse profiles is extended to a variable emission height model to make it physically self-consistent. In this context variable means that the emission height is no longer considered to be the same for different magnetic field lines. The pulse profiles calculated using this new model seem to be less realistic due to a focusing effect and cannot be used to fit (typical) multifrequency pulsar observations. Apart from the focusing effect the general morphology of pulse profiles is not greatly affected by introducing a variable emission height. Additional extensions of the model will be needed to be able to fit observations, and several suggestions are made.