hi all

i design now decimation unit to downsample signal samplesd in 1M to 2K
i cascaded cic decimation then fir decimation and then downsample block ( in simulink).

when i simulate the unit it looks good - i mean i can see the filter gain ( when the input is pulse) and i see the down sampling when the input is sin wave sampled by 1M.

my problem is that i cannot reach good resolution because of the number of bits - i feed the cic by 12 bit and the output number of bits is 42 ( r =32 and n=6 bin =12 )
( bin + N *LOG2(R)) - but after the fir decimation i see the sine wave quantized to much...

does some one have an experience with it ? or some good books to recommend or PDF"S about the right way to choose the bits to get good resolution?

thank u all