AnsweredAssumed Answered

Potential bug in dc_filter?

Question asked by andrew.dickson on Jul 21, 2014
Latest reply on Aug 12, 2014 by CsomI



Please could you explain how to use the dc_filter in the hdl reference for the FMCOMMS1 reference design?


I understand the concept of why it is there, I am just not sure what offset/co-efficient combination should be used (these are programmable from the Zynq processor).


I have simulated this (in Vivado) and so far I believe that port C (of the DSP48 "ad_dcfilter_1" macro) should be set to the full scale output (P) rather than the truncated/scaled input that is connected to port "A". This works when I set the coefficient to 0x0001 and when a small dc offset is present on a channel (+32). However, when the signal of interest is large (sine wave of 2^12 amplitude), the dc offset averaging is becomes quite noisy (e.g. when dc offset of of value +32).


Please confirm my fix, or whether this has been catered for if I have incorrect use of the co-efficient parameters without any change in the code.


Please provide timings (the filter settling time) for this to work and the min/max input (D) / dc offset amplitudes that it will work.


Kind Regards,