>From:         Kalle Pihlajasaari <kalle@DEVICE.DATA.CO.ZA>
>I'm told that the classic FFT is of little use unless you are looking at
>data that has very small discontinuities at the ends of the sample set).

Good point.  I remember from my DSP courses that you can accomplish this
by 'tapering' the ends of your data by suitable envelopes.  The drawback
is that you introduce artifacts in your final FFT answer from the envelope.

All in all not too simple to do with limited RAM.

+---------------------------------------+-----------------------------------+
|          Edward Cheung, Ph.D.         |  The opinions expressed herein    |
|Satellite Servicing Robotics Laboratory|    do not necessarily reflect     |
|    NASA Goddard Space Flight Center   |       those of my employers'      |
|         Code 714.1, Bldg T11B         |                                   |
|          Greenbelt, MD 20771          |                                   |
|   301-286-1269(office) 286-1717(fax)  |  My next book:                    |
| Internet: oadebc@robots.gsfc.nasa.gov | Statistics, Demos and Other Lies  |
+---------------------------------------+-----------------------------------+