AVR-Chat

enkitec@... wrote:

> I have used the low-pass filter described
> below, but I'm not sure on how to make it
> a high-pass: ...
> FILT <--- FILT + FF(NEW - FILT)
> ... Does it mean that (NEW - FILT) is the
> high-pass filter output?

Mark,

So, your question actually appears to be "how do I design digital filters?".  You'll generally get your answer quicker if you ask your question clearly, rather than "I'm trying ..., how do I do that?"

I recommend these on-line filter design tools:
http://www-users.cs.york.ac.uk/~fisher/mkfilter/

For your application, first select Butterworth / Bessel / Chebyshev.

In section 1, leave the selection at Butterworth if you don't know what the options mean.  Select Lowpass or Highpass as appropriate.

Ignore section 2.

In section 3, enter the filter order.  You are presently messing about with first order, so enter 1, but you should also find second order approachable and I would suggest you use that so enter 2.

In section 4, enter the A-to-D sampling rate.

In section 5, enter the desired cut-off frequency as Corner frequency 1.  Ignore Corner frequency 2.

Ignore sections 6, 7 and 8.

Click the Submit Query button.  You get a filter design complete with frequency plot and C code.  But, the code uses floating point, so just scale everything for integer arithmetic.  Use 16 bit unless you have a high sample rate as when scaling for just 8 bits it can be tricky to balance numerical accuracy with freedom from overflow.

I find that the easiest way to test a filter and also to make sure that I understand the algorithm, is with a spreadsheet.

Graham.