# ADC measurement FFT analysis of sine functions with ramp Calibration

Generate Charts (Signal, FFT, INL, DNL)      Read hex data "x00A9C"      Read positive integer data

A default set of measurement data without calibration is displayed initially.
Without calibration paste sine data into the field input data.
Activate button Read hex data or Read data. Then push the button Generate charts.
With calibration first paste positive ramp data into the field input data.
Push the button calibrate which shows the data with best calibration.
Then you can paste sine data into the input field and proceed with read data and FFT.

 Number of points: 1024(10); 2048; 4096; 8192; 16384 (14); 32768; 65536 (16); 130172; 262344 (18); Number of Bits: (Limited to 20 bits, then out of memory error) Number of frequencies for SNR calculation: Windowing: None Hamming Kaiser Nutall

## Signal to Noise

A positive rising ramp measurement can be used for calibration.
Select the number of bits for the calibration.
Fill the data into the input field.
Fill in a last maximum possible value of sine signal.
Read hex oder integer data and press Calibration.
A lookup table for ADC calibration is generated.

Reset calibration

Calibration data can have offset and gain error.
Get global minimum and maximum.
Correct for minimum and eliminate unused codes.
All missing codes are eliminated by downsizing the output code.
Calibration data has noise.
Noise can be averaged, but then a step with missing output codes is also averaged.
Therefore a maximum curve and minimum curve is generated for each point.
The maximum curve goes from left to right taking always the maximum value.
The minimum curve goes from right to left taking always the minimum value.
Slope reversals in the transfer curve are also eliminated and results in a constant code.
A mid curve can be calculated minimizing the smearing of a step caused by missing codes.
The resolution B1 is given by the number of remaining codes (NRC), the noise difference between max and min curve (ND).
$B1 = ld \left( \frac{NRC}{ND} \right)$
It can also be limited by the number of points (NP), having no increase in code (NNI).
$B2 = ld \left( \frac{NP}{NNI} \right)$
The real resolution is the minimum of B1 and B2.
$B = Min(B1,B2)$

Show FFT output values         Show Calibration values         Show Calibration Histogram values