ADS9110EVM-PDK Operation
For 2 kHz SNR and ENOB evaluation at maximum throughput of 2 MSPS, the number of samples should
be 32768 or 65536. More samples than these will bring noise floor so low that the external source phase
noise might dominate SNR and ENOB calculations. On the contrary, for THD and SFDR evaluation, much
large number of samples should be used to reduce the noise floor below –140 dBc to analyze noise-free
harmonics and spurs in the order of –120 dBc Such analysis will require at least 262144 samples.
NOTE:
SNR for ADS9110 with 4.5-V reference is 1 dB lower than with 5-V reference and the ADC
typical SNR is expected to be 99 dB.
Figure 13. Spectral Analysis Tool
Finally, the FFT tool includes windowing options that are required to mitigate the effects of non-coherent
sampling (this discussion is beyond the scope of this document). The 7-Term Blackman Harris window is
the default option and has sufficient dynamic range to resolve the frequency components of up to a 24-bit
ADC. Note that the “None” option corresponds to not using a window (or using a rectangular window) and
is not recommended.
6.5
Histogram Tool
Noise degrades ADC resolution and the histogram tool can be used to estimate
effective resolution
, which
is an indicator of the number of bits of ADC resolution losses resulting from noise generated by the
various sources connected to the ADC when measuring a dc signal. The cumulative effect of noise
coupling to the ADC output from sources such as the input drive circuits, the reference drive circuit, the
ADC power supply, and the ADC itself is reflected in the standard deviation of the ADC output code
histogram that is obtained by performing multiple conversions of a dc input applied to a given channel.
19
SBAU249 – October 2015
ADS9110EVM-PDK
Copyright © 2015, Texas Instruments Incorporated