Hallo Lee,

sounds very interesting. May I ask, what kind of FIR filter you use? How many taps? What wordlength?

My lab board is going to demonstrate only two filters: bandpass and lowpass. It has 16 taps with 16-bit coefficients. The analog input is from an 8-bit A/D and the output is an 8-bit D/A.

Is FIR filter self designed or is it a ready solution?

The coefficients are from a filter program that one of my professors has. I noticed that MatLab on the school computers has a digital filter toolbox, so I will try to cook up my own sometime. Recommendations on free filter-coefficient-generating programs are welcomed.

Some years ago, I was involved in a project where someone wanted to implement FIR filter. Result was a little bit strange and it was found out that rounding errors poduced some inconvenience. May be that this is the reason for what you called 'math never came out quite right' ? Only an idea...

The filter program that made the coefficients calculates roundoff error for me. I chose 16-bit coefficients because 8-bit will have obvious error, even on an 8-bit output. The assembly code has mistakes, thats all. I spent two weeks on it and it doesn't work. I spent ONE NIGHT on the C code and it works just fine!

- Lee