[chuck-users] pulse width modulation

[Hans Aberg]

On 8 Feb 2010, at 04:12, Andrew C. Smith wrote:

> You'll want to use PulseOsc instead of SqrOsc. When they say "square  
> wave tone" they probably mean the squared-off edges. Basically, a  
> pulse width of 0.5 would be a "square" wave, while 0.2 would be a  
> much shorter impulse. I'm not sure what the exact spectral content  
> of the different pulse widths are, though.

A Fourier coefficients tables say that for pulse width factor k, the  
series is the sum of b_i*sin(i*w*t), i = 1, 2, ..., the angular  
frequency w = 2*pi*f, f the frequency, and
   b_i = 4*A*sin(i*pi*k)/i*pi
where A is the amplitude (half distance between max and min). So for k  
= 1/2, one just gets odd partials of relative amplitude 1/i, but when  
distorted the whole spectrum starts to show up.

   Hans