I am trying to implement a chebyshev filter using the values from the Chebyshev tables here: (Figure 8.27)

I am using four cascaded double precision general filters.

My controller generates the coefficients and sends them to the ADAU1452 in 8.24 format.

The filter response is OK for Butterworth, but is way off for Chebyshev. I plot the response using cascaded sections and the result looks perfect. I am thinking that the 8.24 precision is not enough to accurately place the poles on the S-plane. Is it possible to write 56 bit values to these filters?

Thanks,

-Robert

Thanks Dave,

I made a silly error. The tables in the PDF were for low pass and I was applying them to a high pass filter.

I am plotting in a C# application. I made a 1/x error there too which made the plots look correct. It's all working perfectly now.

Thanks.

-Robert