I am using the CORDIC core to implement an arctan function in my design.
I am noticing however, that at near-zero values of X inputs, the result from CORDIC varies considerably from the output of calculating arctan through Matlab.
Why is there a difference in the output and how can I resolve the error?
The difference in output is due to an output quantization error introduced by the digital nature of the core.
Details on this subject can be found in the CORDIC datasheet, under the section Output Quantization Error.
The datasheet suggests that the output of the core might be more accurate if the bit magnitude of the input vectors is increased.
Was this Answer Record helpful?