www.piclist.com/techref/logic/dsps.htm?key=fft

Hi Scott, thanks for your reply. Comments below.

On 26 Apr 2004 at 21:19, Scott Dattalo wrote:

> In some sense, you can view the Goertzel algorithm as a narrow band pass

> filter. I strongly suspect that if you use this concept as guide that it

> would be possible to design an extremely optimal filter bank for filtering

> the 8-DTMF frequencies. For example, the first stage can split the signal

> into highs and lows. Each of these streams are then split again. Finally,

> those four streams are split. You may wish to look up info on multirate

> filtering.

As I have mentioned, I did the DTMF decoding using the information

from the URL below. I can detect DTMF tones in less than 18mS on an

ATMega8 running at 3.58MHz with very good dinamic range and precision.

> > Do you know of some page describing the FFT process like that

> > one on the Goertzel algorithm?

>

> I guess that depends on what you mean by 'like', but you could look at:

> http://www.dattalo.com/technical/theory/dtmf.html

-------------------

The Goertzel algorithm can be implemented as follows:

At every sample:

Q0 = coeff*Q1-Q2+sample

Move Q1 to Q2

Move Q0 to Q1

After a block of N samples:

MAG = SQRT(Q1^2+Q2^2-Q1*Q2*coeff)

-------------------

I would like to see the same description on how to implement

the FFT process. I'm not interested in DTMF decoding. Thas was

just an example and an excellent exercise.

What I want to design is an 8 or 16 tone AFSK modem.

With the Goertzel algorithm I can do the decoding in less

than 20mS using the same ATMega8 running at 8MHz. My doubt is if

the FFT process would give me advantages such as decoding speed

or better error rate.

Thanks,

Mark Jordan

--

http://www.piclist.com hint: The list server can filter out subtopics

(like ads or off topics) for you. See http://www.piclist.com/#topics

In reply to: <Pine.LNX.4.44.0404262103450.19048-100000@ruckus.brouhaha.com>

See also: www.piclist.com/techref/logic/dsps.htm?key=fft