----- Original Message -----
From: "Scott Dattalo" <scott@DATTALO.COM>
To: <PICLIST@MITVMA.MIT.EDU>
Sent: Tuesday, January 21, 2003 12:48 AM
Subject: Re: Dual sine wave generation.


> On Mon, 20 Jan 2003, Michael Rigby-Jones wrote:
>
> > > -----Original Message-----
> > > From: moonshadow [SMTP:moonshadow@NTLWORLD.COM]
> > > Sent: Monday, January 20, 2003 9:47 PM
> > > To:   PICLIST@MITVMA.MIT.EDU
> > > Subject:      Dual sine wave generation.
> > >
> > > I need to generate two sine waves. One fixed at 14 kHz and the other
> > > variable between 14 kHz and 14.250 kHz. I would prefer to use
Crownhill's
> > > BASIC Compiler to do this.
> > >          Can anyone suggest how I ought to go about this please ?. I
had
> > > thought to use the two internal  CCP modules and the PWM command.
However
> > > both modules have the same output frequency.  Same with FREQOUT.
> > >
> > > John Kent.
> > >
> > What resolution do you need for the adjustable frequency?  How much
jitter
> > can you tolerate?
> >
> >  You can (maybe) use the hardware PWM, but not in the manner you are
> > thinking. One idea is to build a table of SIN values and step through
the
> > table at a variable rate, outputing the table value to the CCP module.
The
> > PWM output would need to be low pass filtered to remove the PWM
frequency
> > components.  This implies that the PWM frequency has to be significantly
> > higher than 14KHz.  The higher the PWM frequency the worse the PWM
> > resolution, which will affect the the quality (e.g. distortion) of your
sine
> > wave.
>
> You may wish to consider investigating:
>
> http://www.dattalo.com/technical/software/pic/picsine.html
>
> This sin() routine takes as its input a 14-bit frequency variable. (Scale
> 2^14 to 2*pi). Assign two 16-bit frequency variables to your 14kHz and
> 14.25 kHz frequencies. Advance each of these by a fixed (but different)
> amount in a timer based interrupt routine and call the sine routine
> mentioned above.
>
> The sine routine takes 48 cycles. If you based your frequencies on a TMR0
> interrupt with the pre-scale = 1, you could easily run both of these
> routines (~100 cycles out of the 256).
>
> Choosing the update rate:
>
> Let's suppose you're running your PIC off of a 20MHz clock. That's 20/4 =
> 5 million cycles per second. A TMR0 interrupt can be programmed to occur
> every 256 / 5million = 51.2 uS. The period of a 14,000 Hz wave form is
> 71.42 uS. Oops. This ain't gonna work :(.
>
> ---
>
> Phase accumulators but no linear interpolation:
>
> The routine on my web page uses phase-accumulator techniques for the
> frequency variable (phase accumulator is just fancy term for rollover
> arithmetic) and linear interpolation for the sine computation. The linear
> interpolation takes quite a bit of time. So Eric Smith's sine computation
> is more suitable for your application:
>
> http://www.brouhaha.com/~eric/pic/sine.html
>
> His code will take about 13 cycles. Again, you'll want 16 bit phase
> accumulators for the frequency variables. Using Eric's code, it would take
> about 13*2 + 4*2 ~ 34 cycles to compute the two sine values. On a 20MHz
> pic will take 6.8 uS or roughly 10.5 samples per sine wave. The challenge
> now is designing the isochronous code to update at this rate. (You can
> make the TMR0 interrupt run fast than 256 cycles by manually writing to
> it.)
>
> ---
>
> A pure software technique sounds iffy. But as Michael asks, it really
> depends on your requirements. There is one more software technique using
> ADPCM but it would consume all of the PIC's resources.
>
> Scott
>
> --
> http://www.piclist.com#nomail Going offline? Don't AutoReply us!
> email listserv@mitvma.mit.edu with SET PICList DIGEST in the body
>

--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email listserv@mitvma.mit.edu with SET PICList DIGEST in the body