I come accross a problem recently,

I use a 16 bit AD
The Enob we test is 13.2bit with maxmum power input is 10dBm when the maxmum value of AD is 32767
Thus I can get the Equivalent hardware noise is 10 dBm- ((16-13.2)*6.02 + 1.76)dB = -71.22dBm;

The question is, when there is an LNA in front of the AD, such as the gain is 35dB, why the max AD value I got is around 21(peak - average)?

my calculation is: with this AD value(21), i got the power input is around -56dBm(10dBm - (6.02*log2(32768/21)+1.76)) which is 10dB larger than the thermal nosie power(-101dBm) + G(35dB) = -66 dBm.

 

Hardware noise ?

There is noise , and measurement error,

If its not DC on the input to the ADC, then the frequency in, as well as the jitter on the sample clock are also a limitation,

Add that to the data sheet spec,
i.e. with an oscillating signal in, a 16 bit ADC is not going to give you more than 14 bits ,

QED, we need a lot more information