Hi, concerning the AD7176-2 Digital Filters I run into the spreadsheet **AD7176_2_filter_model.xls** provided by Analog Devices.

The spreadsheet results always varied from my understanding of a sinc5 + sinc1 filter.

The equation showed that the sinc5 + sinc1 filter is

3 x Moving average filter of the length 32

+ 1 x Moving average filter of the length of 16

+ 1 x Moving average filter of variable length depending on the output data rate.

Here is the Excel Code Analog Devices uses. 20*log(abs(

(1/f_dec *SIN(f_dec *$A44/Fm*PI())/SIN($A44/Fm*PI()))^(order)-2 |

* (1/f_dec2*SIN(f_dec2*$A44/Fm*PI())/SIN($A44/Fm*PI())) |

* (1/(f_dec*n_avg)*SIN((f_dec*n_avg)*$A44/Fm*PI())/SIN($A44/Fm*PI())) |

))

For example

Name | Symbol | Value | Unit | Range | ||

MCLK Frequency | Mclk | 16,00 | MHz | 1 ..16.384 | ||

ODR[4:0] | ODR | 1 | 0..20 | |||

Sinc Filter Order | f_ord | 5 | 3 or 5 | |||

Single Cycle / Multiple-Channels | 1 | 0 or 1 | ||||

SINC3_MAP | s3_map | 0 | 0 or 1 | |||

FILTER[14:0] | FilterReg | 5000 | 1 to 32767 |

results in order = 5 and n_avg = 2 and f_dec = 32 and f_dec2 = 16

**Questions:**

**So is the sinc5 part in reality a sinc3 + sinc1 filter? Or is there an error in the spreadsheet?**

Hi da_lundi,

The AD7176-2 filter model is slightly modified Sinc5 filter. It is a combination of 4 x Moving average of length 32 + 1 x moving average of length 16. The first part contains three of the Sinc5 moving average terms. The second part models the shorter moving average and the third part combines the remaining Sinc5 term plus the output data rate dependent average term.

Regards,

Jonathan