Post Go back to editing

Category: Hardware

I am currently trying to evaluate the performance of two ADCs.

These are the 1- LTC 2500 (32 bit) and the other is the 2- AD4030 (24bit).

Based on their datasheets the first one has a SINAD of 104dB  while the AD4030 achieves a SINAD of 108.4dB.   .

If we utilize the equation ENOB = (SINAD -1.76) / 6.02 (assuming operation under the FSR) we come up with the ENOB1= 16.9  and ENOB 2= 17.71.

That practically means a significant impact in performance of the ADC just due to quantization error.

If we add additionaly noise sources starting from the PCB layout up to signal conditioning electronics and other factors (EMI etc) this performance will get even worse.

Does this means that the AD4030 performs better? and why then LTC2500 even if it is specified with a 32 bit resolution the only you can get out of it is maximum 17 bits?

Thank you!

[edited by: emassa at 8:26 PM (GMT -5) on 6 Mar 2023]
• Hi,

Effective and Peak-to-Peak resolution should not be confused with ENOB, as ENOB uses an FFT analysis of a sine-wave input to the ADC while Effective resolution and Peak-to-Peak resolution measure the ADC's noise performance at essentially narrowband/DC measurement where spectral distortion is not factored. The peak-to-peak resolution represents the resolution for which there is no code flicker or noise-free code and so it gives a better indication of the performance vs effective resolution. Just like how the dynamic range is calculated, the effective resolution is calculated using the rms noise, whereas the peak-to-peak resolution is calculated based on peak-to-peak noise. The peak noise is statistically 6.6 x rms noise; therefore, the effective resolution is greater than the peak-to-peak resolution by 2.7 bits but does not highlight the number of bits that flicker. May we know your specific application so we can assess which ADC is more suitable for it?

Thanks,
Andrei

• It is not clear to me what you are trying to point out from your text above. Do you have any available references?

(I found this one https://www.analog.com/media/en/technical-documentation/application-notes/AN-615.pdf  but this does not really means that performance is going to be better)

Regarding this, "FFT analysis of a sine-wave input to the ADC while Effective resolution and Peak-to-Peak resolution measure the ADC's noise performance at essentially narrowband/DC measurement where spectral distortion is not factored", yes my signal is going to be of an oscillatory nature....that's why I am examining the ENOB and I see that performance is much lower than this advertised.

My application requires a precision ADC for a single ended signal (0...5V), for a bandwidth of 50kHz and being able to distinguish a voltage level of 100-400 nV minimum.

The ADC has to be a SAR and we are planning on oversampling && decimating to achieve a sampling frequency of 1.6MSample/sec to get 2 additional bits, since the ENOB is relatively very low for both the LCT2500 and the AD4030.

• The AD4030 has lower noise and better linearity than the LTC2500-32. The number of bits is not an indicator of accuracy. Better indicators of accuracy are INL for DC applications and THD for AC applications. Transition noise is a good indicator of DC noise performance and SNR is a good indicator of AC noise performance. The LTC2500-32 provides 32 bits of resolution to minimize roundoff error when averaging large numbers of samples. If you look at the AD4030-24 data sheet you will see that when doing block averages, the data format is 30 bits for this same reason.