Re: choosing right adc for data acquisition system
Hi,
16 bit ADC vs 12bit ADC with oversampling:
you need to average (or sum) at least 16 pieces 12 bit values to get a 16 bit value.
But the 16 bit ADC has a DNL of typically +/-0.4LSB this equals to 0.025LSB at a 12 bit value.
You won´t achieve the same with the 12 bit ADC..
You need to know: The "averaging" method only works good when the bit values change. And they change if there is noise or a (fast) signal. You need noise with at least 1 LSB magnitude of a 12 bit value.
And for a good DNL value you need a linear distribution of the voltage values (almost impossible. You may generate a triangle style dither signal, or some artificial random noise with linear distribution, maybe with the use of DAC....but that´s overkill.. )
In your case: the more noise the better the averaging method. But who wants noise?
If you have clean signals, then you gain nothing: worst case you read 16 times the same ADC value. And this is not unusual with a 12 bit ADC and clean input signals.
This means the resulting DNL may be worse than +/-8LSB referred to the 16 bit value.
If your signal is noisy, then the single 16 bit value is not useful. Then the 12Bit oversampled, digitally averaged and low pass filtered signal may give an improvement.
My recommedation: Use the 16 bit ADC, generate clean signals and use a good anti-aliasing filter.
Klaus