From: Dan Kennedy
Date: 4/2/00
Time: 10:59:23 AM
I have a command line program that is having some success at producing dechirped files. The resulting file is showing the Doppler test signal, from Doppler.wav, between 6 and 7 minutes using CoolEdit. I have not seen this signal before.
This is our original Doppler Test Signal from Dr. Lodder. It shows a visible signal changing amplitude at 1 minute intervals and linearly decreasing in frequency from about 1620 to 1550 Hz over 5 minutes.
We have studied this signal in pretty good detail. Use the archive search function above and search on Doppler if you want to see the posts on it. http://setiweb.org/asrgargus/DiscussionArchive/00000051.htm had an example post which calculates the drift rate at -0.2272 Hz/sec or -13.632 Hz/min.
Using DeChirpTD1.exe to shift the samples in the time domain, gives the dechirpTD1.wav file. In CoolEdit, it looks like this (using the same settings as above).
The signal has been de-chirped. Some artifacts have also been produced, but I think that may be impossible to stop these considering the digital sampling and the further digital shifting.
Using the CoolEdit Analysis Scan function on the invisible 6 to 7 minute interval shows this small peak. It appears to be about 2 db above the noise floor. But the integration time was less than a minute. I suspect that a drift interval duration signal (13 minutes?) would yield several more decibels of sensitivity. And I cannot be certain that the scan absolutely excluded the stronger earlier samples.
Using MathCAD to integrate over the 6 to 7 minute interval gives the following graph. Here I am sure these are only the samples from the quiestest range.
I am experimenting with other shift techniques in the time domain. I think I am losing sensitivity because some of the power spectrum is feeding the artifacts.
The program dechirping program is at dechirpTD1.exe
The current source code is at dechirpTD1.c
The exact command line that produced the de-chirped file is
dechirpTD1.exe doppler.wav dechirpTD1.wav -0.041
The last parameter controls the shifting. It is negative, as is the frequency shift of the Doppler signal. Changing this changes the slope of the signal. I experimented to find the -0.041
The resulting WAV files are not perfectly formatted. I see that GoldWave complains about the Chunksize not being correct. It is not correct because the resulting file has a different number of samples than the original. This is the result of my shifting them around. If we get to using some utility in batch files, I will square away the WAV header so that it is correct.