Project

General

Profile

SignalScopePage » History » Version 3

matt, 10/24/2017 05:49 AM

1 3 matt
h1. Signal Scope
2 1 zecke
3 3 matt
h2. Installation
4 1 zecke
5
Before running the scope app, check that the following standard pre-requisite packages are also installed:
6
 * GNU Radio
7
 * The op25 blocks
8
 * The op25 repeater block
9
10
The osmosdr http://sdr.osmocom.org/trac/wiki/GrOsmoSDR package is required when using USRP or HackRF (or other hardware supported by osmosdr)
11
12
Additionally, hamlib http://sourceforge.net/apps/mediawiki/hamlib/index.php?title=Main_Page is (optionally) needed if using a frequency-agile receiver that is to be tuned using hamlib (see below)
13
14
The Python Numeric package is also needed when using autocorrelation.  If you're running Debian or Ubuntu:
15
16 2 matt
>@apt-get install python-numeric@
17
18
19 1 zecke
Other than these pre-reqs, no special setup or installation is needed.
20
21 3 matt
h2. Targets
22 2 matt
23 1 zecke
 * Choice of osmosdr(uhd / hackrf, etc), discriminator-tap, or "audio-IF" hardware modes
24
 * Supports either GL or non-GL display mode
25
 * Both FSK4 and CQPSK demodulation modes supported
26
 * Lower and middle range PC hardware, not just the latest (optional features require more CPU)
27
28 2 matt
h2. Running the Program
29 1 zecke
30
There are four overall options or modes depending on your hardware; these are
31
 * Osmosdr
32
 * Hamlib
33
 * External receiver with discriminator tap connected to sound card
34
 * External receiver with IF in the audio sound card range (e.g., 24 KHz), referred to as "audio IF"
35
36 2 matt
h3. Running with the disc-tap option
37 1 zecke
38
The signal scope does not require the USRP.  If you have a discriminator-tapped receiver, use the "-a" option:
39
40 2 matt
>@./scope.py -a -v 10 -g 50@
41 1 zecke
42
43 2 matt
h3. UHD example
44 1 zecke
45 2 matt
46
>@./scope.py --args 'uhd,nchan=1,subdev=B:0' -g 65 -f 412.34e6  -o 50e3 -V -v 0   -T trunk.tsv -N 'ADC-pga:10,PGA0:29'@
47
48
49
h3. HackRF example
50
51
>@./scope.py --args 'hackrf'  -g 65 -f 412.34e6 -N  'RF:14,IF:32,BB:26' -o 50e3 -T trunk.tsv  -V -v 0@
52
53
54
h3. Hamlib example
55
56 1 zecke
Two receivers in cascade (USRP with LFRX connected to the 455 KHz IF output terminal of hamlib-controlled receiver)
57
58 2 matt
>@./scope.py --args 'uhd,nchan=1,subdev=A:A' -g 65 -f 455e3 -o 50e3 -V -v 0 -H 1234  -T trunk.tsv@
59 1 zecke
60 2 matt
h3. RTL-SDR Example
61
62
>@./scope.py --args 'rtl'  -g 65 -f 123.45e6 -N 'LNA:49' -V -v 0 -S 1000000 -q 39 -T trunk.tsv@
63
64
65
h3. File Playback Example
66
67 1 zecke
Selects input from a file instead of RF capture device.  The file must be in standard complex (I/Q) format at a sampling rate of 96,000 Hz.  Playback should automatically restart from the beginning whenever EOF is reached.
68
69 2 matt
>@./scope.py -F ~/gr/capture-96k.dat -g 10@
70 1 zecke
71 2 matt
72
h3. Running in the Audio IF mode
73
74 1 zecke
Receivers equipped with an IF output in the sound card range can be used.  This is known as "audio IF" mode.
75
A soundcard sampling rate of 96K is used and the IF frequency (typically 24 KHz) is given using the {{{--calibration}}} parameter:
76
77 2 matt
>@./scope.py -A -c 24e3 -g 50 -v 10@
78 1 zecke
79
80 2 matt
h2. Feature overview
81
82 1 zecke
 * Spectrum plot
83
 * Baseband oscilloscope
84
 * Eye Pattern Diagram (Datascope) display supporting several standard symbol rates
85
 * Constellation Diagrams
86
 * Demodulated Symbol Output
87
 * Correlation (including Fast Auto-Correlation)
88
 * Direct-frequency entry, signal gain and fine-tuning controls
89
 * User-selectable demodulator (FSK4 or QPSK)
90 2 matt
 * Multi-system trunked receiver with IMBE voice support using the @-V@ command line option (requires frequency-agile SDR receiver)
91 1 zecke
92
In the baseband AF (disc. tap) mode, several program functions are disabled (spectrum FFT, constellation diagram, PSK demod, FAC and iDEN correlation)
93
because these functions require direct access to the signal with no demodulation.
94
95 2 matt
In all modes, the wireshark option is used to write received P25 packet data to Wireshark.
96 1 zecke
97 2 matt
h2. Program Options
98 1 zecke
99
Here is a full list of program options:
100
101 2 matt
@Usage: scope.py [options]
102
103 1 zecke
Options:
104
  -h, --help            show this help message and exit
105
  --args=ARGS           device args
106
  --antenna=ANTENNA     select antenna
107
  -a, --audio           use direct audio input
108
  -A, --audio-if        soundcard IF mode (use --calibration to set IF freq)
109
  -I AUDIO_INPUT, --audio-input=AUDIO_INPUT
110
                        pcm input device name.  E.g., hw:0,0 or /dev/dsp
111
  -i INPUT, --input=INPUT
112
                        input file name
113
  -b Hz, --excess-bw=Hz
114
                        for RRC filter
115
  -c Hz, --calibration=Hz
116
                        USRP offset or audio IF frequency
117
  -C Hz, --costas-alpha=Hz
118
                        value of alpha for Costas loop
119
  -f Hz, --frequency=Hz
120
                        USRP center frequency
121
  -F IFILE, --ifile=IFILE
122
                        read input from complex capture file
123
  -H HAMLIB_MODEL, --hamlib-model=HAMLIB_MODEL
124
                        specify model for hamlib
125
  -s SEEK, --seek=SEEK  ifile seek in K
126
  -S SAMPLE_RATE, --sample-rate=SAMPLE_RATE
127
                        source samp rate
128
  -t, --tone-detect     use experimental tone detect algorithm
129
  -T TRUNK_CONF_FILE, --trunk-conf-file=TRUNK_CONF_FILE
130
                        trunking config file name
131
  -v VERBOSITY, --verbosity=VERBOSITY
132
                        message debug level
133
  -V, --vocoder         voice codec
134
  -o Hz, --offset=Hz    tuning offset frequency [to circumvent DC offset]
135
  -p, --pause           block on startup
136
  -w, --wireshark       output data to Wireshark
137
  -W WIRESHARK_HOST, --wireshark-host=WIRESHARK_HOST
138
                        Wireshark host
139
  -r RAW_SYMBOLS, --raw-symbols=RAW_SYMBOLS
140
                        dump decoded symbols to file
141
  -R RX_SUBDEV_SPEC, --rx-subdev-spec=RX_SUBDEV_SPEC
142
                        select USRP Rx side A or B (default=A)
143
  -g GAIN, --gain=GAIN  set USRP gain in dB (default is midpoint) or set audio
144
                        gain
145
  -G GAIN_MU, --gain-mu=GAIN_MU
146
                        gardner gain
147
  -N GAINS, --gains=GAINS
148
                        gain settings
149
  -O AUDIO_OUTPUT, --audio-output=AUDIO_OUTPUT
150
                        audio output device name
151
  -q FREQ_CORR, --freq-corr=FREQ_CORR
152
                        frequency correction
153
  -2, --phase2-tdma     enable phase2 tdma decode
154
  -L LOGFILE_WORKERS, --logfile-workers=LOGFILE_WORKERS
155 2 matt
                        number of demodulators to instantiate@
156 1 zecke
157 3 matt
h2. Spectrum Display
158 1 zecke
159
[[Image(a.png)]]
160
161
The controls arranged along the bottom of the page are:
162
 * Frequency: to retune, type the new frequency here and press ENTER
163
 * Signal Gain: adjusts the baseband (demodulated) signal level
164
 * Fine Tune: adjusts tuning frequency over +/- 3000 Hz range
165
 * Demod: Selects demodulator (currently used in Demodulated Symbols only)
166
Except for the signal gain control, these controls are only available in USRP RX mode.
167
168 3 matt
h2. Eye Pattern Diagrams
169 1 zecke
170
The scope input source can be connected either before or after the symbol filter using the Viewpoint toggle.
171
172
Also the proper speed must be selected from the available options.
173
174
[[Image(b.png)]]
175
176 3 matt
h2. Constellation Diagram
177 1 zecke
178
[[Image(f.png)]]
179
180
The signal scope also features an angular population graph (shown above) in addition to the traditional constellation display.
181
In this mode the symbol magnitude (distance from center) is discarded.  Instead the circle is sliced into segments and a count
182
of symbols found in each segment is plotted.  This is similar to a histogram except that a straight line is drawn between each
183
result, and that the results are arranged in polar form instead of rectangular form.
184
185
With this display, the zone at the exact center of the plot can be used precisely to measure the degree of separation or margin
186
between the demodulated symbols.  The plots below illustrate the difference, with poor separation showing on the left image as 
187
revealed by the failure to converge at the center and by the clear territorial violations:
188
189
[[Image(mhp7a.png)]][[Image(c.png)]]
190
191
The two-color mode is used in these images, providing natural relief to highlight the distinctive feature of π/4 DQPSK in which 
192
successive symbols are chosen from two distinct constellations (each containing four possible symbol values) separated by 45°
193
194 3 matt
h2. Demodulated Symbols
195 1 zecke
196
[[Image(d.png)]]
197
198 3 matt
h2. Correlation
199 1 zecke
200
[[Image(e.png)]]
201
202
Cross correlation allows rapid identification of signals with known characteristics.  Frame Sync (FS) signatures of several commonly
203
used radio systems are included.
204
205
By convention correlation results are usually displayed using positive correlation peaks only.  In this system
206
however it is possible (and legal) for negative correlation products to be produced.  This can occur for two wholly separate reasons:
207
 * If the hardware polarity is inverted 
208
 * When the FS symbols are purposely inverted as an integral part of protocol processing (commonly used in certain protocols but not used in P25)
209
210
The first case commonly happens when using the disc-tap method of hardware connection, because the actual polarity of the signal seems to vary 
211
randomly among different sound cards and receivers.  In one actual case, two PC's of the same PC brand bought from the same store had opposite polarity.
212
213
The P25 software framer automatically detects the proper polarity and issues a message if negative polarity data is received:
214
215 2 matt
>@Reversed FS polarity detected - autocorrecting@
216
217
218 1 zecke
The automatic correction applies only to software framing and doesn't help with correlation.  For correct results for both software framing and
219
correlation, you should correct the polarity reversal problem at its source; this is done using negative values for the {{{--gain}}} parameter at 
220
program start time:
221
222
223 2 matt
>@./scope.py -a -v 10 -g -50@
224
225
226 1 zecke
The second cause of negative correlation peaks is that some protocols (although not P25) make use of both normal- and inverted-polarity FS 
227
sequences as a standard part of their processing.   Instead of clogging the GUI menu with several choices that are merely inverses of others,
228
just for the sake of always having positive-peaked correlations, we allow the correlation graph to reflect the natural polarity of the data.
229
Thus both + and - peaks are shown, allowing quick diagnosis of incorrect hardware polarity (see above), and allowing identification of the
230
particular family and sub-protocol in use.
231
232 2 matt
h2. Auto Correlation
233 1 zecke
234
Also included is Frank's Fast Auto Correlation (fac):
235
236
[[Image(g.png)]]
237
238
For further details about Fast Auto Correlation refer to Frank's page at http://sites.google.com/site/radiorausch/
239
240 3 matt
h3. TRUNKING
241 1 zecke
242
New in late 2013, trunk following for multiple trunked P25 systems was added, supporting the following feature set:
243
  * Any number of separate trunked systems may be scanned
244
  * P25 Phase I (IMBE) voice channel decoding and audio output
245
  * Supports LSM/CQPSK systems that require CQPSK (not C4FM) demodulation in addition to C4FM systems
246
  * Since CQPSK demodulation is used, LSM simulcast distortion is suppressed (contrary to all current scanners)
247
  * In this release, systems and voice channels are scanned sequentially (like trunk tracking scanners)
248
  * Alpha tagging for talkgroup ID's
249
  * Per-system whitelist (closed group) support: only those talkgroups in the list are scanned
250
  * Per-system blacklist support: all talkgroups are scanned, except those listed
251
  * Configuration files are TSV (tab-separated); may be edited using spreadsheet software such as Libre office
252
  * Talkgroup ID hold: momentary delay after each voice transmission to allow following conversations
253
  * Manual talkgroup ID hold: click to pause, remains on current talkgroup until resumed
254
  * Manual lockout: click to lock out current talkgroup
255
  * All of the signal scope functions (see above) are live and may be selected in real time
256
  * Traffic history including list of active voice channels, key trunk control channel data, etc.
257
  * Hardware support provided by gr-osmosdr http://sdr.osmocom.org/trac/wiki/GrOsmoSDR including usrp/uhd, hackrf, and (in theory) RTL DVB-T based sticks etc.
258
The trunking code has not yet been merged into the master branch.  In the meantime after cloning the
259
repository you can issue this command to switch to the proper code branch:
260
261 2 matt
262
>@git checkout max-trunking-update2@
263
264 1 zecke
(you must first {{{cd}}} to the top level op25 project directory)
265
266 3 matt
h2. Trunking Configuration
267 2 matt
268 1 zecke
The primary configuration file {{{trunk.tsv}}} contains one line (or spreadsheet row) per trunked system
269
to be monitored. The file need not be named {{{trunk.tsv}}}; any name of your choice may be used, and is specified using the {{{-T}}} option.  The file contains the following information:
270
 * NAC: This is used as the "primary key" to tell different systems apart. 
271
 * Sysname: The name of the system (for display purposes)
272
 * Control Channel List: comma-separated list of trunk control channels.
273
 * Offset: if using a SDR with GPSDO, this value is not needed and should be set to zero.  Used for frequency drift correction.  The values shown are almost certainly incorrect for your system; the frequency error (drift) is different for every oscillator and is also almost always temperature dependent.  You can use the other functions of scope.py such as the eye diagram to determine the error (difference between nominal frequency and received frequency).
274
 * Modulation: C4FM or CQPSK
275
 * TGID Tags File: specifies the name of a TSV file for this system, containing alpha talkgroup ID tags
276
 * Whitelist: if set, the system is a "closed" system; only listed TGID's will be included
277
 * Blacklist: excludes listed TGID's
278
 * Center Frequency: set this to (Highest-Frequency-Channel - Lowest-Frequency-Channel) / 2 - and make sure it's at least (say) 50 KHz away from any active channel
279
280
Note: the first line of the file containing the field names must not be changed.
281
282 3 matt
h3. Software Tuning Mode
283 1 zecke
284 2 matt
A new feature which helps to workaround a current problem with the RTL-SDR, the software Local Oscillator allows software tuning within the passband whose width is equal to the current sampling rate (@--sample-rate@) parameter.  Currently this seems to be limited to 2.56 MHz.
285
286 1 zecke
To use this mode
287
 * find the lowest (F1) and highest frequencies (F2) in the system
288
 * You will have to adjust F1 and F2 such that (F2-F1) is less than the selected sample rate
289
 * RTL SDR apparently supports rates higher than 2.56 MHz but there are reported problems
290
 * Find the center frequency = F1 plus one-half of (F2-F1)
291
 * Adjust the center frequency by a few tens of KHz if necessary to ensure it does not coincide with any of the system channel frequencies, to avoid DC offset
292
 * The higher the sample rate, the more CPU is used
293
 * If the system contains frequencies outside of the band from F1 to F2, tuning error messages will be issued but the remaining channels should still work
294
 * The "Center Frequency" is defined as a new column in the trunking TSV file (added to the right of the Blacklist column after the example shown below was created).
295
296
[[Image(t2.png)]]
297
298 3 matt
h2. TSV Files
299 1 zecke
300
Use the following command to open a TSV file for editing:
301
302 2 matt
@oocalc trunk.tsv@
303 1 zecke
304
Here are the options used when opening the file
305
306
[[Image(t1.png)]]
307
308 3 matt
h2. Talkgroup ID Alpha Tags
309 1 zecke
310
There is usually a separate tags file for each system, although sometimes multiple systems can share a common set of tags.  The talkgroup ID (first column in each row) is in decimal (base 10).
311
312
[[Image(t3.png)]]
313
314 3 matt
h2. Trunked Traffic
315 1 zecke
316
[[Image(t4.png)]]
317
318 3 matt
h2. P25 PHASE 2 TDMA
319 1 zecke
320
Use the {{{-2}}} command line option to enable Phase 2 TDMA.  Unlike in P25 Phase 1 FDMA, TDMA voice channels cannot be tuned manually.  TDMA voice channels are always operated under control of a trunked system having a 9,600 FDMA trunking control channel.  Accordingly you must establish a trunking definition for your system as detailed above; no other special setup is required.
321
322 3 matt
h2. TALKGROUP LOGGING
323 1 zecke
324
This records in parallel all talkgroup activity to .WAV files (except talkgroups excluded via the white list and black lists system). The band of frequencies that can be spanned is equal to the SDR sampling rate, e.g., 2.4 MHz or so if using an RTL, wider in other devices (8 MHz in HackRF, for example). Only one SDR is needed though, no matter how many talkgroups are to be logged (as long as these frequency limits are observed). The option is enabled via the "-L n" command line parameter (where n is an int specifying the number of logfile workers - one plus the number of concurrent talkgroups to be logged).   Note that in TDMA one worker is required per active frequency (not per talkgroup). The wider the spectrum, and the more workers defined, the higher the CPU usage ...
325
326
The logging option must be used with the -T option, and (in this version of scope.py) only the first data row of the trunking TSV file is utilized - subsequent rows are not used.
327
328 3 matt
h2. BUGS
329 1 zecke
330
Possibly bugs exist, here are a few of the known ones as of this writing
331
 * Symbol filters totally brain damaged (need separate filters for each speed)
332
 * When switching modes using the notebook tabs, leftover data from before may appear momentarily
333
 * Highest and lowest speeds are not well tuned resulting either in sluggish updates or CPU exhaustion
334
 * Currently ignores all except first frequency in list of trunk control frequencies
335
 * The selected tab may override the mode (C4FM vs. CQPSK); be sure to select "constellation" to enable CQPSK mode
336 2 matt
337 3 matt
h2. CREDITS
338 1 zecke
339
I ripped off the "tabbed notebook" theme (and code) from Stevie; it already had the spectrum, baseband, and decoded symbol displays. I added the data scope, constellation scope, cross correlation, and trunk tracking features.
340
341
The Fast Auto Correlation (fac) code came from Radiorausch.
342
343
Special thanks to Mossmann for leaving one commented-out "print" statement in c4fm-decode.py. Of course it was tempting to wonder "what would happen" if that statement were un-commented. The final result? The cross-correlation feature.
344
345
Special thanks also to GPL(v3), for encouraging these mashups.
Add picture from clipboard (Maximum size: 48.8 MB)