RadioAstron satellite admits in principle a testing the gravitational redshift effect with an accuracy of better than $10^{-5}$. It would surpass the result of Gravity Probe A mission at least an order of magnitude. However, RadioAstron's communications and frequency transfer systems are not adapted for a direct application of the non relativistic Doppler and troposphere compensation scheme used in the Gravity Probe A experiment. This leads to degradation of the redshift test accuracy approximately to the level 0.01. We discuss the way to overcome this difficulty and present preliminary results based on data obtained during special observing sessions scheduled for testing the new techniques.