DSD 1.3 and mbelib 1.2 released

[Jay911]

Jay911,

1st problem: The WAV file is recorded at 44100 Hz instead of the required 48000 Hz.
2nd problem: That is not a P25 transmission. That is a 2 level FSK signal.

Do you have Motorola Astro (VSLEP) digital systems up there ?

Nope, I don't believe there are any VSELP systems in use here. The file is actually a recording of what is commonly called "P25 audio" (incorrectly as it may be) on a 3600 baud Motorola Smartzone trunk system, which was what I understood this program to be able to demodulate (among other formats).

If there is some difference between this form of "P25 audio" and the actual audio format used on a native 9600 baud P25 system, please educate me on what the difference is and how a scanner can demodulate the audio of both with only one demodulation routine called a "P25 decoder".

 

[EricCottrell]

Thanks Eric. That implies ...

1. the IV is moved four bytes further down in the frame
2. the LID field is also moved down four bytes
3. the LID looks to be larger eg. 20 bits instead of 16

The "2175" value that separates the two IMBE cell pairs- is the same as the "classic" ProVoice frame sync of "5712" value - but with the digits in reverse. The engineers changed the header but retained the structure and format of the "payload" portion of the voice frames. Interesting stuff.

Back to the LID field in the header. In the published specs for ProVoice - the 16 bit LID field is used to send out 32 bit code words (you need two consecutive frames to get the whole code word). The code word implicitly contains an 8 bit color value. The 12 least significant bits are BCH or Golay protection (I forget at the moment). The total bits in the codeword are therefore 40 bits.

Assume the EA ProVoice headers expanded the LID field to 20 bits. Two consecutive LID fields could be combined into a 40 bit code word. Might they have done away with the implicit color code to continue using the same error protection mechanism?

Food for thought while you're staring at raw binary dumps.

Hello,

I was trying to use Unitrunker 1.0.0.11 to track the EDACS EA system but it keeps generating several new windows per second with a solid control channel. The legend has EDACS96 xx LIMBO with xx a hex number that counts up. I assume it is generating sequential site numbers because there was a whole bunch of S000000xx where xx is a hex number.

It is interesting to see how they implemented EA. On a non-EA system the two data words in a frame are two separate pieces of information. On an EA system the two data words are used together. Some of the messages are the same. I suspect they arranged the commands and bits so a regular EDACS radio will not try to affiliate with a EA site.

One task is to look at the data on the control channel and the voice channel together.

I want to go down to Florida or another location with a multiple site EA system to capture some logs and see if I can figure out more of the bits. Now that there is a chance to actually listen to digital voice it would be more fruitful.

The other task is to see I can use the DV Dongle to do decoding if I do the de-interleave and error correction on the computer and feed the chip a straight stream with no error correction. I could not find any chip configuration that did the P25 interleave.

73 Eric

 

[SCPD]

I was trying to use Unitrunker 1.0.0.11 to track the EDACS EA system but it keeps generating several new windows per second with a solid control channel. The legend has EDACS96 xx LIMBO with xx a hex number that counts up. I assume it is generating sequential site numbers because there was a whole bunch of S000000xx where xx is a hex number.

this is the first I've heard Unitrunker not tracking an EA system. This is offtopic to DSD so we can pick that up in another thread. Please send logs when you can.

 

[Jay911]

Hey, wait a minute.. why would the WAV file being recorded at 44100 Hz instead of 48000 Hz have any bearing on things, when I am playing that back over an analog cable from the line-out jack on the netbook to the line-in jack on the Linux machine? It should see it as any other analog audio source.

 

[SCPD]

Hey, wait a minute.. why would the WAV file being recorded at 44100 Hz instead of 48000 Hz have any bearing on things

It doesn't. I've received audio samples in WAV and even MP3 format that I was able to test against on the same machine by using a patch cable to loop the headphone jack back into the microphone jack (be careful about the volume).

 

[MattSR]

Jay911,

1st problem: The WAV file is recorded at 44100 Hz instead of the required 48000 Hz.
2nd problem: That is not a P25 transmission. That is a 2 level FSK signal.

Do you have Motorola Astro (VSLEP) digital systems up there ?

From the sound of it, it was is a P25 voice channel, but its massively distorted - it looks like 2FSK because whatever was recording it has been overdriven so far into clipping that it looks like a 2 level signal.

 

[SCPD]

Go to http://www.jay911.org/radio/753f-p25.zip.)[/i] Could someone assess them and see if I still have the volume set wrong, or not a clean enough signal, or a bad tap, or what?

Jay - these samples are clipped. Turn the microphone gain down. Run a 'scope program to verify the waveform never exceeds the +'ve and -'ve limits of the microphone (or line) input.

 

[SCPD]

whatever was recording it has been overdriven so far into clipping that it looks like a 2 level signal.

There you have it Jay - in stereo.

 

[Jay911]

Ok, so the volume is still too high. Or is there a chance that the tap is bad? - all my scanners are tapped with just a wire directly to the instructed tap points, without resistors/capacitors, mainly because I can't find a source of the recommended resistors/capacitors.

Sorry to everyone for semi-hijacking the thread. If you want to continue this in PM or another thread, please do.

 

[Jay911]

Jay - these samples are clipped. Turn the microphone gain down. Run a 'scope program to verify the waveform never exceeds the +'ve and -'ve limits of the microphone (or line) input.

When I was running it into Audacity, it provides a waveform visually as it records. I was under the impression that I had turned it down far enough, as I was seeing peaks and valleys in the waveform (instead of a solid cutoff).

 

[SCPD]

Sorry to everyone for semi-hijacking the thread. If you want to continue this in PM or another thread, please do.

This is highly relevant to the program. I take it you are recording from Windows. If so - run Uniscope.

To validate P25 recording, set Uniscope to 4800 baud and QPSK modulation (even if your system is C4FM). Adjust levels and post a screenshot. We'll go from there.

See the bottom two screenshots in this page. I just realized I don't have a C4FM example. I'll fix that.

If you can get a decent looking eye pattern in Windows, you can record the audio at that level - it should play back just fine. Using a patch cord does introduce additional distortion because of the capacitive coupling in the mic/line input but should be tolerable if you've got a good signal.

I should write a Linux version of Uniscope. Seeing an eye pattern (or lack of one) reveals much about the quality of signal available for decoding.

 

[Comint]

This is highly relevant to the program. I take it you are recording from Windows. If so - run Uniscope.

Is Uniscope available as a stand-alone program, or does one have to install UniTrunker to get access to it?

--
Comint

 

[Giulio]

I should write a Linux version of Uniscope. Seeing an eye pattern (or lack of one) reveals much about the quality of signal available for decoding.

A Linux version of Unitrunker would be really apreciated too

 

[KA1RBI]

I should write a Linux version of Uniscope. Seeing an eye pattern (or lack of one) reveals much about the quality of signal available for decoding.

Already done! See my "datascope" code (available in OP25's audio_p25_rx.py for users of the disc-tap).

We know that installing GNU Radio and the rest of the stuff is at least an order of magnitude more difficult than the DSD install is in Linux. I thought I understood Stevie to say that we're working on developing a "release" version of the software so at least that won't require an SVN pull (someday soon, hopefully)...

Best Regards

Max

 

[grosminet]

One problem with using a cheap scanner to decode Tetra is the bandwidth of the final IF. 15 kHz is typical for these radios but Tetra uses 25khz. In other words, you've lost some of the signal before it reached the software.

In my city I have may be 30 tetra frequencies . if you want records let know to european listeners

 

[offsite]

Quick access to UniScope in Linux

For Ubuntu, at least, we are able to run UniScope here by installing Wine and then Unitrunker. The whole setup (Wine/Unitrunker) takes about 5 minutes and was glitch free on a bootable, persistent 4 GB flash drive. And, of course, once you get everything on a flash drive, you can run it on most any pc!

While UniScope, and Tuner work fine, we have yet to get Unitrunker itself (either current, or retro) running.

-- Has anyone had success running UniTrunker under Wine?
This clearly is applicable to DSD, because we have not found a way to trunktrack PV on other than the PCRnn00 radios without it.

-- Is there a way to get the version of UniScope that appears on the UniTrunker signal comparison page?
It seems to be an upgrade from the classic one we are using here.

Another scope program that runs under Wine, is Christian Zeitnitz's SoundCard Scope:
Soundcard Scope

It is a free, very cool, full featured scope app, but it lacks UniScope's signal specific features, such as eye diagram presentations, etc.

73
-rb-

.......................
I should write a Linux version of Uniscope. Seeing an eye pattern (or lack of one) reveals much about the quality of signal available for decoding.

 

[SCPD]

Is Uniscope available as a stand-alone program, or does one have to install UniTrunker to get access to it?

At the moment, it is only available in the "Retro" MSI installer. I'll put up a stand-alone copy for folks to use directly.

For Ubuntu, at least, we are able to run UniScope here by installing Wine and then Unitrunker.

Good to know.

-- Is there a way to get the version of UniScope that appears on the UniTrunker signal comparison page? It seems to be an upgrade from the classic one we are using here.

It's an experimental (read "buggy") version with goals of (a) reduced CPU and (b) automatic selection of the appropriate eye pattern trigger mechanism. Pulse, C4FM, QPSK, and FFSK all require different triggers. The current public release of Uniscope should serve our purposes.

You can get an eye pattern from a scope program for some formats if it offers a dual trigger (rising edge and falling edge) set point (with time base adjusted to the expected symbol period). Most scope programs don't do this.

 

[Forts]

Just curious... are any of the available input types more resource intensive than another? The reason I ask.. I'm using an older IBM Thinkpad laptop and I can decode P25 reasonably well. I don't have ProVoice within earshot, but the samples provided on the 1st page play pretty well too. I am having absolute 0 success with MotoTRBO however (the one I want, naturally!). I'm wondering if I should perhaps try a different PC. There is a chance the system I'm monitoring is encrypted of course, but I've tried a few other TRBO systems in the area with zero success there too. Is anyone having good results with TRBO?

 

[Forts]

Also... is there any way DSD can show additional info regarding TRBO? (ie group codes etc) in addition to the time slot. Keep up the excellent work btw!

 

[ilgrant]

Just curious... are any of the available input types more resource intensive than another? The reason I ask.. I'm using an older IBM Thinkpad laptop and I can decode P25 reasonably well.

How old is older? I'm curious as to what the minimum CPU requirements are to have the program work correctly.

I found a couple of Mototrbo systems in my area but haven't been successful, either. Encryption is possible but no way of knowing that.