what is the minimum noise floor on electronic circuit board?

[david753]

noise floor

I am trying to use AD6640 now.
But, I wonder what the minimum sensitivity is.
According to the theory, 12bits output (1 Vp-p Dynamic range)
has -68dbm(4-12*6).
If consider to the noise issue, is it still possible?
what is the minimum noise floor on electronic circuit board?

 

[Jeekoky]

noise floor

If only thermal noise, the noise floor at 290K temperature is -174dBm

 

[flatulent]

Re: noise floor

You calculated the dynamic range. The 4 dB is added to six times the number of bits. The easiest way to do the calculations of system performance is to put in a signal that runs the ADC to full scale and then calculate the quantization noise from the dynamic range and then add the Johnson noise at the ADC input caused by the stages preceding the ADC. Remember that quantization noise is a nonlinearity and also produces distortion (harmonic and IMD) and the quantization noise is correlated with the signal.

The usual system design has the preceding stages have enough gain that their Johnson noise toggles the ADC by a few counts.

 

[david753]

Re: noise floor

According to the quantization noise theory of DSP.
SNR(db)=6.02N+1.763, N=AD bit numbers
So, ideally, 12 bits AD converter, SNR has about 74 db.
Since white noise is -174dbm, it almost can be ignore it.
The conclusion is,
AD converter output has -68dbm resolution.
Am I right?
Should I consider any others interference?

 

[flatulent]

Re: noise floor

White noise power of a passive resistance is -174 dBm/Hz. So it all depends upon the bandwidth of the system prior to the anti-aliasing filter and the gain.

In most systems that operate at high signal levels you do not have to worry about the noise prior to the ADC. One advantage of having the noise toggle the ADC count is that it dithers the quantization noise so it is uniformly spread out and not noticeable spectral lines.

 

[GDF]

Re: noise floor

White noise power of a passive resistance is -174 dBm/Hz. So it all depends upon the bandwidth of the system prior to the anti-aliasing filter and the gain.

In most systems that operate at high signal levels you do not have to worry about the noise prior to the ADC. One advantage of having the noise toggle the ADC count is that it dithers the quantization noise so it is uniformly spread out and not noticeable spectral lines.

The noise in your descripton is contributed by former stages or ADC itself?
Another question for consultation is how important the noise contributed by the devices in ADC, compared with quantization noise?
Thanks,