Evaluation is a key stage in all teaching-learning processes, but it usually demands significant efforts of preparation from students and teachers, not to mention that it is very time-consuming. The traditional model of evaluation prescribes that students must sit periodically to demonstrate that they can recite blocks of knowledge, and solve exercises and problems which usually resemble or refers to the same set of study cases presented in lectures, in the laboratory, or the textbooks. Thus conceived, evaluation is indeed lacking, particularly in physics teaching: Did the students just learn how to pass this exam? Pass without real learning!. In this work I present a set of new, heretical ideas concerning possible changes in physics teaching evaluation, namely: (i) exploiting the exams as opportunities for further learning, (ii) examinations as a way of acquisition of new knowledge, or learning new analytical techniques, and (iii) exams as an opportunity for the application of standard powerful tools which students learned in their previous mathematics and physics courses. I present evidence of the quality-of-learning discriminatory power of new model of evaluation. The changes proposed are partially supported by Herzberg model of psychological growth recently adapted and applied to physics education.