Here's another interesting signal. Don't know if it's a well known signal type, and I've heard it a lot before on my scanner (FM demodulation only) but never really looked at its actual raw signal before. I got out my Icom PCR1000 today, and using my own custom made software set it to upper-sideband mode and 50kHz IF filter width (a combination not allowed using Icom's own software). This basically allowed me to downconvert a signal to the audio part of the spectrum, without FM demodulation. I then tuned to an interesting signal. This allowed me to record the actual raw signal's RF content (just downconverted to the AF baseband) on my computer, using the audio software GoldWave. Unfortunately, even with such a wide bandwidth and the samplerate set to 48kHz, the audio signal path (all analog AF electronics in the radio and in the sound card) is not completely flat across the entire AF spectrum, and tends to weaken signals at higher frequency. So then I used GoldWave's spectrum filter to do 2 things, filter out everything outside the bandwidth of the desired signal (approximately 16.5kHz wide signal), and also to boost the higher frequencies (frequencies near the upper part of the desired signal were about 17dB weaker than frequencies near the lower part of the desired signal). The last thing I did was to use GoldWave to FM demodulate the signal. This produces a much better FM demodulated signal than using the radio's FM mode (even when using a wide IF filter), as the radio's audio output is highpass filtered to block frequencies below common speech frequencies, which can block important low frequencies on some types of signals. This FM demodulated signal produces a sound that is very similar to what you would hear if you were listening to it with most scanners, and is what you should use to compare to other recordings of radio signals on the internet for the purposes of identifying the signal (the other two I included just so that someone could more thoroughly analyze the mode itself and maybe make a MultiPSK style soundcard-based decoder software for the mode). This FM demodulated copy has the advantage that only the DC component has been removed (calculate the average value for the entire recorded signal, and then subtract it from each sample in the signal), which makes sure that all low frequency audio remains (including frequencies that are well below the lowest hearable frequency). By doing FM demodulation in software, and not in the radio hardware, it makes sure that the entire content of the demodulated signal exists, so that it can be decoded into something meaningful using additional software (which I don't have now, as I don't know what digital mode it is that I am working with here). Also by using wideband USB mode, and demodulating it later in software I don't need direct access to the radio's discriminator (which usually is what's used for this purpose).
As these files are much too large to attach to this post (over 60MB), I have instead put the 3 WAV files in a ZIP file and uploaded that ZIP file to Mediafire. Below you will see a download link for that ZIP file on Mediafire.
Download link:
https://www.mediafire.com/?6kshok2156nmnbl
File Descriptions:
Unknown Signal (866.027MHz, USB, IF-Width 50kHz, 6-23PM PST, 2-3-2015).wav
This is the raw signal as recorded on my computer. The sample rate is 48kHz.
Unknown Signal (866.027MHz, USB, IF-Width 50kHz, AF-Width 16.5kHz, Spectrum Equalized, 6-23PM PST, 2-3-2015).wav
This is the same as the above file, but with the spectrum equalized, and filtered to remove all frequencies above 16.5kHz. The sample rate is 48kHz.
Unknown Signal (866.027MHz, USB, IF-Width 50kHz, FM Demod, 6-23PM PST, 2-3-2015).wav
This file has the output of the final step I performed, FM demodulation. It sounds like what you would hear on most scanners when listening to the signal. The sample rate is 8kHz (samplerate is lower because demodulated signal only occupies 1.6kHz, so 4kHz nyquist limit is more than enough to hold this).
My own analysis of the signal:
It appears to consist of a multiple separate packets of data, with a fixed duration between packets. The duration appears to be about 18 seconds long (+/- 150ms).
The carrier is not maintained between transmissions, but rather is switched off.
The data packets are usually about 8 seconds long, though the exact duration will vary slightly based on content (the longest one I have seen is actually twice this long, indicating possibly that it's a special packet of some kind).
Each data packet starts with unmodulated carrier for a a period of exactly 870ms.
After this, each data packet is is modulated by square-wave, which switching the carrier between the highest frequency and lowest frequency. This part of the packet lasts for a duration of exactly 20ms. The square-wave appears to have a frequency of 800Hz.
After this comes data.
After this comes a final segment of unmodulated carrier. This last segment of the packet lasts 20ms.
Then there's also a rare packet type that I'm not sure what it is, but it's not a normal data packet, as it is much shorter than data packets. It only comes after a long string of data packets. It's possibly an "end of file" indicator, assuming that what's being transferred is a file. The duration of off-time between a data packet and this special packet is 5.86 seconds on the one that I recorded (rather than the normal 8 seconds, as between any 2 data packets), but this might vary (I don't have any others to compare it to that I recorded).
The duration of unmodulated carrier at the beginning of this special packet is 494ms.
Then there is a segment of squarewave modulated carrier, switching the carrier between lowest and highest frequencies, for a duration of 480ms. The frequency of the square wave is again 600Hz.
Then there is a very brief data segment, the vast majority of it appears to be repeating a specific pattern.
At the end of this packet there is a segment of unmodulated carrier lasting for a duration of 19ms.
Now in the data packets' data segments, there are a number of different types sub-modes in use. The data always is sent at 1600baud in this recording. However, I've seen on very rare occasions when 3200baud is used instead, but I don't have an example of this recorded, and usually it switches to 3200baud mid-packet and then back again, so 1600baud appears to be default. My description provide here is for the 1600baud signal. There's a standard 2-tone FSK, and also a 4-tone FSK (probably more properly called MFSK, as it uses more than two tones). There are also two different shifts in 2-tone FSK, a wider shift and a narrower shift. The two tones in wide 2-tone FSK are at center frequency +/- 5kHz. In 4-tone FSK it uses the same two tones as in 2-tone FSK, and also a tone that is 1/3 the distance between the lower and upper tone, and also a tone that is 2/3 between the lower and upper tone. From tone with lowest frequency to highest frequency, I will call them tone-1, tone-2, tone-3, and tone-4. Wide shift 2-tone FSK uses tone-1 and tone-4. Narrow shift 2-tone FSK uses tone-2 and tone-4 (though on some occasions think I've seen it use tone-1 and tone-3, but I don't have an example of this recorded). When in 4-tone FSK submode, it uses all 4 tones. It seems that at some times in the data portion of a packet there are square-wave alternations between tone-1 and tone-4. This may be how it represents long strings of binary 0s in the data it's sending or as some kind of field separator if the data is formatted into fields. I'm not sure about this.
Further analysis of this signal by somebody who's experienced in signal analysis is needed, if it is not already a well known signal type that is completely documented somewhere.
