AnsweredAssumed Answered

Reading back Tx FIR coefficients when not 1x interpolate

Question asked by matwillis on Aug 10, 2016
Latest reply on Aug 22, 2016 by mhennerich

I am setting up the Rx and Tx FIR's and want to use /4 decimate for Rx and 4x interpolate for Tx path.

The driver code (linux and No-Os, I am using No-Os) attempts to validate that the coefficients are written into the IC correctly. However, I am seeing some strange behaviour.


For the Rx path, with /1 or /2 or /4 decimation the coefficients read back and verify correctly.


When the Tx is set for x1 interpolate, the coeffs are read back OK.

However, when set to 2x or 4x interpolate, the coeffs are not read back 1-for-1; this causes the validation to give errors.

Interestingly, there is a pattern dependent on the interpolate rate selected, as below;


Coeffs written: -4, 7, 19, 27,20,-6,-47,83,-85,-27,...

for 2x interpolate setting, coeffs read back; -4, 19, 20, -47, -85, i.e. every 2nd coeff written

for 4x interpolate setting, coeff read back; -4, 20, -85, .., i.e. every 4th coeff written.


Would someone be able to confirm the validation of coeffs works correctly for TX x4 interpolate?

(or any pointers to what might be going wrong?)