Radio Detection Basics – Software – JAnalyzer




This software was also written by Esko Lyytinen and his son Öllie. It too is freeware.This written in Java so it can be run on any system that has Java installed on it. I have run this on Windows XP, Linux, Unix, and on an Apple iMac (Pentium).

Janalyzer is a lot more flexible than it’s ancestor, mAnalyzer. The user can control fft bin sizes, timing, rates of image scans, and data output.

Like mAnlayzer, JAnalyzer has a small footprint, the executable .jar file is only 118 kb in size.

I usually run JAnalyzer so it produces 3 FFT spectrograms. The first is a time compressed image that contains a full 24 hours in a single line. The image can store a week’s with of data in one image. I do this so I can see aurora, Es, lightening, and other sources of interference. Having a weekly time compressed image also aids in seeing weekly trends. The image below original size is 1043 x 680.

24 hours a line

All echoes seen in the time compression are over dense echoes. Normal duration echos are not seen at this compression rate. A careful look shoes the long duration echoes in the first row which is August 9, 2010, start off rather spotty. As we approach the peak of the Perseids on the 13th we see more and more over dense echoes.  The light gray blocks are created by man made noise, washer and dryers, microwave ovens, and arching AC lines during high winds.

The second FFT image is an image of the past 24 hours as seen below. This helps me easily spot overdense echoes that I might be able to correlate with the all-sky video camera. Original image size is 964×2750.

Daily August 13th Perseid peak


The  image above shows many overdense, long duration echoes. It also shows the noisy periods as well. if you look carefully you’ll see an occasional carrier streaking across the pass band. Hint the last line of the image at 1200 UT. Hash marks are at 1 minute intervals. There are two lines per hour – the shorter line is the end of each hour.

The third FFT image is a fast scan. It usually scans at a rate of an image every three minutes. It can run faster so head echoes can be seen.

(Reminder add fast scan image here).

A cursor over the start of an echo will show the frequency of the echo as well as it’s start time. Moving the cursor to the end of the echo shows the end time.

I have JAnalyzer set to generate 3 text files. One is a 10 minute file which is similar but not the same as the mdata file mentioned in the mAnalyzer section. It records four sets of power bins and the counts and duration of each bin. It also produces an hourly report much like it’s mAnalyzer counter part, mhdata. If my research problem requires it, I have the program write a line for every echo heard.

The individual raw echo file format is:

2010-03-03 14:20:37.487, 00003, 02.29745, 03.03210
2010-03-03 14:21:13.626, 00013, 02.61173, 03.79104
2010-03-03 14:22:17.924, 00042, 03.09038, 04.30555
2010-03-03 14:22:42.468, 00002, 01.22930, 01.31838

YYYY-MM-DD HH:MM:SS.sss Time stamp beginning of echo

Duration of echo (timing depends on fft buffer size) To determine duration count you divide buffer size by sample rate. I run several buffer sizes at 8000 samples per second. Buffers can be 256, 512, 1024, and so on. For example a one unit of duration recorded with a 1024 at 8000 samples per second = 0.128 of a second or 128 ms. And unit for a bin of 256 at 8000 samples/sec =  0.032 sec or 32 ms

Mean Audio power in log(2)  (That is, the mean of the logarithmic values)

Maximum power in log(2)