It has been suggested that the classical chemical analysis may be affected by systematic errors that would introduce abundance differences between dwarfs and giants. For some elements, the abundance difference could be real. We address the issue by observing 2 solar-type dwarfs in NGC 5822 and 3 in IC 4756, and comparing their composition with that of 3 giants in either of the aforementioned clusters. We determine iron abundance and stellar parameters for dwarf stars. Then, abundances of calcium, sodium, nickel, titanium, aluminium, chromium, and silicon were determined for both giants and dwarfs. The standard equivalent width analysis was performed differentially with respect to the Sun. We find an iron abundance for dwarf stars equal to solar to within the margins of error for IC 4756, and slightly above for NGC 5822 ([Fe/H]=0.01 and 0.05dex respectively). We show that, for sodium, silicon, and titanium, abundances of giants are significantly higher than those of the dwarfs of the same cluster (about 0.15, 0.15, and 0.35dex). Other elements may also undergo some enhanced, but all within 0.1dex.