Previously in the adc read function, a slow card would lead to 'max_loops' dropping to 1 and staying there for future cards.
Fast ADCs read after slow ADCs would not have the output data properly decimated.