Week of 11/5

Here is the display of my oscilloscope when the input is a 1000 pulse per second output of a time-code generator. (The time-code generator is a device that locks to the 10 MHz output of an atomic clock and produces 1 Hz or 1 KHz pulse trains as well as human readable time synchronized to the atomic standard.)

I imported the data and plotted it along with its periodogram.

Notice that only the odd harmonics are present.

Here is a mathematica notebook that simulates this.

Download lots of fourier transform examples

---

Sampling theorem. See 10/31/07 lecture notes

We end up with $f(t) = \sum _ {n = - \infty} ^ \infty f(n/2f_s) \frac{\sin(\pi (2 f_s t -n))}{\pi (2 f_s t -n)}$

This amounts to taking samples of the data every 1 / 2f_s and multiplying them by a sinc function and adding up the results.

Download Sinc function interpolation via the samping theorem

Mathematica can find the fourier transform of a box function of width h centered on zero times sin(2πx):

$-\frac{i \sqrt{\frac{2}{\pi }} \left(2 \pi \cos (h \pi ) \sin \left(\frac{h k}{2}\right)-k \cos \left(\frac{h k}{2}\right) \sin (h \pi )\right)}{4 \pi ^2-k^2}$

Download 11-7-07 notes

Download timing the fft