Re: [SpectrumLabUsers] Something perhaps novel
- HI Wolf,
That's very helpful. I looked about for the plugin information, but didn't find it. With regard to detecting periodicity, won't a comparison of bin number contents be sufficient. If not, never mind, this will take a bit of noodling to sort out.
Best - Ben
On 7/10/2013 1:11 PM, wolf_dl4yhf wrote:
As Jurgen correctly pointed out, you could do this with your own plugin (compiled into a windows DLL).
Those plugins originally operated on short-term fourier transforms of the FFT-based audio filter, but since a couple of years there are also callback functions (invoked by SL in the plugin DLL) to perform any kind of processing in the time domain (i.e. on the waveform data). The tricky task may be that you'll have to detect the periocity of a signal somehow; so a combination of processing in the frequency- *and* the time-domain may be required.
The plugins are explained here:
There is also a link to a sample plugin (written in 'C') in that document.
All the best,
Am 10.07.2013 01:07, schrieb M. Ben Schuetz:
A few days ago I raised a question about "signals" with and without harmonics. Since I have only been using SL in one way for the past couple of years, the other signal types weren't that familiar to me. Graham and Wolf responded with thought provoking observations. So, I made a few simple tests with a signal generator to explore the issue.
Here's the situation. A square wave or any periodic pulsed signals is coupled into the computer audio board through a 0.1 uF cap. Given the internal resistance of the audio board, this results in roughly 100 us pulses. SL responds to these showing approximately equal amplitude signals of all the harmonics. If a pulse or two are missing here and there, the appropriate harmonic is enhanced. The consequence of this is that any signal not having 2nd, 3rd, . . . harmonic bins are just noise and can be disregarded. Hopefully, and even better, in software they can be discarded from any further processing. That would bring the noise baseline down to about nothing while leaving a clean detected signal.
It would obviate the need to look at waterfall clutter and improve the minimum detectable limits.
This may have applications anywhere a periodic signal can be converted to pulses before going to SL.
I suspect the software changes would not be huge. Or, is there a way to do this external of the SL source code. Any thoughts?