Sliding Channel VLF Spectrogram

how fine a resolution do you need? takes a long time to resolve small bandwidths, like over 1 sec to have 1Hz bandwidth. and 10 seconds to get

If speed is critical, you could do what the sonar monitoring people did/do, they multiply the frequency up, taking it times 1000, then the speed of the resolution increased from 1 sec to 1mS. From memory they did it by launching the signal into a crystal letting it bounce around, several times, monitoring it as it went by. Sure they have digital ways to do that now,

but back to your question, doubt if there's an exact program to do what you want. they are easier to write then you might think, though.

THe sound card can do your job. You need a mixer to up the freq scale. have u looked SDR software?

jamie

Check out this link:

They have a software package that might do what you want.

Software radio is interesting stuff, but doesn't appear to suit the ELF frequency range I am looking for.

To phrase my OP another way, I would like a software definable, sliding bandpass filter. Maybe something like this, but as an existing GUI-based "product" because coding it is outside my area of expertise.

The ability to apply several filters simultaneously to the same signal would be ideal.

For example, if the acquired bandwidth is 0-200Hz, a separate bar graph would show the varying composite amplitude for 0-10Hz, 30-35Hz,

Ken Rockwell

If you really want that 1:100 frequency range, you would most likely need to use subchannel bandwidths depending of the center frequency. RTAs (Real Time Analyzers) have done this for decades by displaying bar graphs for each octave and later for each third. For DSP systems, the scaling is basically depending of the sampling rate. If you could drop the sampling rate to say 200-300 Hz, you might get what you want.

If not real time monitoring is needed, record the signal for say 100 s, then playback it at 100x, thus your frequency range (1-100 Hz) will be translated to 100 Hz - 10 kHz, which any RTA can handle.

You may have to repeat your recording, which may cause some aliases effects, but by using some sliding windowing, by advancing the window by 1 s every second of real time recording, these could be minimized.

Have you looked at the various waterfall displays intended for amateur radio applications. These are typically intended for processing the audio range (300-3000 Hz) from an SSB receiver. From such displays, the sidebands from narrow band modulation methods such as PSK31 and manual Morse can be clearly seen.

Again speeding up the signals by 10:1 or 100:1 might also help. Using different speedup rates for 1-10 Hz and 10-100 Hz with two separate waterfall displays would do the job.

The waterfall displays are generated using FFT, which was already turned down by the OP.

It might be easier to FFT it and throw away the parts you are not interested in. DAQARTA (sp?) will do that in realtime quite nicely.

You are up against time and frequency being cunhugate variables so you can't demand fine frequency resolution of a short time series unless you have some additional heuristics available to help the processing.

That just leaves the chirp-z, which is way more complicated. In linux, there are FFT routines in the library.