Low Pass Filters in FPGA's

Question

I am trying to implement a Low Pass Filter in an FPGA using VHDL. Now, I worked out all the filter coefficients, but I am having trouble trying to understand how to work with the coefficients, given the fact that they are non-whole numbers. I would appreciate any help with an example of a simple filter implementation in VHDL.

Thanks a lot, Chris

Answer 1

You should convert the coefficients to fix-point numbers.

Very short:

• Chose a precision and multiply all coefficients by the nearest higher reciprocal power of two.
• Do the normal multiply-add operation in the FPGA.
• The end result should be divided by that same power of two.

Example:

• Coefficients are C0 = 0.707, C1= 0.123
• Lets take a precision of 0.5% which is 1/200
• Nearest power of 2 above 200 is 256
• C0 becomes 0.707*256 = 181, C1 becomes 0.123*256 = 31.
• Use those in your filter.
• At the end throw away the bottom 8 bits (As extra you can round using bit 7)