Gary J FollettDukes HiFi wrote:
> I understand and appreciate this discussion of NOISE dynamic
> range. This does not explain how one can get SIGNAL dynamic
> range higher than ~96 dB with 16 bits. These are different quantities.
> If the A to D has say, 1 mV per bit response, then the largest
> input signal this 16 bit A to D can measure should be ~65,000
> mV. That’s about 96 dB. Please tell me what I am missing here…
As long as the quantization noise is not correlated with the signal,
ADC will successfully convert signals much, much smaller than the quantization
step.
The noise effectively interpolates ADC steps, enabling the information
about the present weak signals to enter the ADC digital outputs stream.
Those signals are then easily extracted by digital filtering.
Please note that for this to work, we must provide a "high quality noise"
in addition to the signal, with a level comparable to the ADC step.
The signal alone is not sufficient.
Also, please note that the ADC conversion itself is NOT improved by this
process.
In fact, it is made slightly worse because of the added noise.
There is a slight numerical error in the mentioned values
of the theoretical ADC dynamic range. The correct value is:
10 * log10 (3 * 2 ^ (2 * N - 1))
for N=16 we get 98.09 dB.
73,
Sinisa YT1NT, VE3EA
_______________________________________________
TenTec mailing list
TenTec@contesting.com
http://lists.contesting.com/mailman/listinfo/tentec
|