Noise density of ADC

1. Please teach us about ADC internal noise.

What kinds of noise are there, how does it relate to it, how is it measured, etc.?

• SNR

Signal to noise ratio is calculated using the full scale RMS input signal and comparing it to the RMS noise.  It includes all noise sources.  The equation is given below and it is covered in our precision lab series.  Normally SNR is given for wide bandwidth type ADC.

snr = 20*log(Vfsr_rms/Vn_rms)

https://training.ti.com/ti-precision-labs-adcs-calculating-total-noise-adc-systems

• Quantization noise

Quantization noise is the “rounding” error introduced when converting a continuous signal to a quantized signal.  Higher resolution devices have less quantization noise then lower resolution devices.  If you apply a DC input to an ADC and see a single column in the histogram of the device than it is “quantization noise dominated”.  On the other hand if you see a Gaussian distribution it is thermal noise dominated.  We have a lot of content on this in our precision lab series.

https://training.ti.com/types-noise-adcs

• Thermal noise – noise that is proportional to the square root of absolute temperature
• Transition noise – the ADC noise translated into LSB. It can be thought of as an uncertainty in where the transition from bit-to-bit happens.
• Noise (in datasheet of ADS8329, what kind of noise is it?). This is the total noise of the ADC.  Typically this does not include reference noise as that will depend on the reference connected.  You can think of this as the same as transition noise but given in uV instead of codes.

2. What noise is reduced by oversampling? All types of noise are improved by oversampling

Is quantization noise the only improvement? No.  Oversampling introduces a digital filter and noise shaping.  See our precision lab series.

https://training.ti.com/ti-precision-labs-sar-delta-sigma-basic-operation

Please also teach us about improvement and non-improvement mechanisms.

3. Whether noise is discussed by total amount or by noise density, I think it will change.

The video below introduces how noise density is translated to total noise.  Normally total noise is used to calculate the system error as total noise is in units that compare to the signal (e.g. Volts or codes)

https://training.ti.com/ti-precision-labs-adcs-calculating-total-noise-adc-systems

4. I would like to thank you for introducing any good application notes regarding these.

Both ADC and amplifier precision labs cover noise in detail.  Below is amplifier noise content.  I would watch this first, and then go through all the ADC content.

https://training.ti.com/ti-precision-labs-op-amps