OP25 stalling

[MCWKen]

I have installed OP25 on the laptop using virtualBox. When I call it up, it translates on the control channel, but after a few transmissions, it freezes up.

First off, sterr.2:
audio device: default
Allocating 15 zero-copy buffers
Allocating 15 zero-copy buffers
Allocating 15 zero-copy buffers
Allocating 15 zero-copy buffers
1604334873.231205 PCM underrun
1604334873.715816 PCM underrun
1604334874.708587 PCM underrun
1604334875.156909 PCM underrun
1604334876.109357 PCM underrun
1604334876.579225 PCM underrun
1604334877.566129 PCM underrun
1604334878.016765 PCM underrun
1604334878.509574 PCM underrun
1604334878.977330 PCM underrun
1604334879.451938 PCM underrun
1604334879.942300 PCM underrun
1604334880.441712 PCM underrun
r82xx_write: i2c wr failed=-7 reg=1b len=1
r82xx_set_freq: failed=-7
Flowgraph completed. Exiting
The command line:
./rx.py --args 'rtl' --gains 'LNA:49' -S 250000 -f 770.45625e6 -q 0 -o 15555 -T trunk.tsv -2 -w -V -U 2>stderr.2

Any suggestions would be appreciated.

Ken
Clinton, IA

 

[maus92]

Not sure if people have had a satisfactory experience using op25 in a virtual environment, but your constellation plot is not very tight.

 

[boatbod]

Not sure if people have had a satisfactory experience using op25 in a virtual environment, but your constellation plot is not very tight.

Yup, that's an awfully weak signal which I very much doubt will yield useful voice recovery. That said I don't think the weak signal is the cause of the problem given the following two errors just below the PCM underruns.

r82xx_write: i2c wr failed=-7 reg=1b len=1
r82xx_set_freq: failed=-7

It looks to me like op25 lost communication with the RTL device. Might be heat or power supply related, or just a plain bad dongle. If you run 'dmesg' in a terminal do you see any messages that relate to either USB activity?

 

[MCWKen]

Given that observation, I had an issue with that USB port on another application. This is on a laptop, so it is limited to 3 ports. Will re-assign a different port. The dongle is a Nooelec, and it has been spot on. But thanks, and will report later

 

[KA1RBI]

The eye diagram is pretty clean and it clearly indicates a C4FM channel. Therefore (assuming the rest of the channels in the system are the same), it suggests that attempting to develop a complex constellation is not the optimum way to receive this system. I'd definitely try receiving it as C4FM (fsk4).

And yes, the 'r82xx_' messages also indicate a separate, hardware or virtualization problem causing difficulty communicating with the SDR hardware...

Max

 

[maus92]

Hmm... I'm curious how the VirtualBox setup will work out. The gain value of 49 seems very high; not sure the distance the op is from the site he is trying to receive - seems to be the DeWitt multicast site. My Nooelecs use ~30, but then again I am within the footprint of a simulcast cell in a system designed for portable on street coverage. My constellation plot only has a few outlying points and 4 distinct concentrations (though the box is slightly off axis); my eye is crisp, and the symbol plot has 4 distinct lines.

(I understand that Clinton County lost their radio system during this past summer's Derecho. If they are going to permanently move to ISICS, they will need more sites. For example, our state 700 system in our county has a multicast site and a simulcast cell (~7 tower sites) to cover the county, and wants to add another.)

 

[wgbecks]

I have been using VirtualBox to run a Ubuntu VM with op25 successfully now for several years. However, it's been necessary to setup the USB virtualization of the Linux VM as a USB 3.0 (xHCI) Controller, else the USB throughput suffers.

My current VM is Ubuntu 18.04.4, but I have been running Ubuntu VM's with op25 Version 14 with absolutely no issues once the USB has been setup as outlined above. Additionally, be sure to allocate plenty of memory and at least two CPU's to your VM.

Bill

 

[maus92]

I didn't have a spare Nooelec, so I used a FlightAware Orange. The command line:

./rx.py -l http:192.168.1.210:8080 --args 'rtl' -N 'LNA:30' -q 0 -S 1024000 -x 1.25 -d 810 -T trunk.tsv -V -2 -U 2> stderr.2

This is what a constellation plot should look like:

 

[wgbecks]

Looking pretty good! Dot clusters could be a little tighter, you could try a bit more LNA gain and or adjust antenna placement.
What does your frequency error show?

 

[maus92]

It's a simulcast site, and I'm pretty sure I am receiving signalling from 3 or 4 tower sites at my location (the closest being < 1km, the furthest about 15km) using a mag mount 4-ish" telescopic antenna stuck to a interior railing in my office (I used an attic mount 5/8 ground plane antenna, but the cheap mag mounts work just as well - or better.) The freq error is ~20Hz, variable.

 

[boatbod]

It's a simulcast site, and I'm pretty sure I am receiving signalling from 3 or 4 tower sites at my location (the closest being < 1km, the furthest about 15km) using a mag mount 4-ish" telescopic antenna stuck to a interior railing in my office (I used an attic mount 5/8 ground plane antenna, but the cheap mag mounts work just as well - or better.) The freq error is ~20Hz, variable.

Plot looks good to me.

 

[maus92]

Yup, I'm not having issues, other than I prefer the Nooelecs to the FlightAware radio. I posted the constellation plot to illustrate to the op what he should be seeing.

I'm hoping to get the mult_py version configured next since that is where development seems to be heading.

 

[KA1RBI]

Maus92, could you please post an eye diagram (datascope) of the same system and channel as post #8 of this thread (that is, to match that constellation diagram)

thx

Max

 

[maus92]

I'd also note that the op's sample rate seems small / low / slow. I usually run my Nooelecs at 960000, but I just happened to be messing around with around with the FlightAware's settings trying to dial in the mixer plot, and ended up with 1024K..

 

[maus92]

Eye pattern of the cch. Note it varies in quality on a per frame basis.

 

[maus92]

Constellation of the cch. Note that the grouping changes in quality on a per frame basis. I'll try to link a screen recording if I can get Linux / Kazam to cooperate.

 

[maus92]

Here's a link to the constellation plot in motion:

Meet Google Drive – One place for all your files

Google Drive is a free way to keep your files backed up and easy to reach from any phone, tablet, or computer. Start with 15GB of Google storage – free.

drive.google.com

 

[KA1RBI]

OK thanks, the eye plot was as expected for a simulcast system. There is a very noticeable difference between posts 1 and 15 - in post 1 the channel is c4fm whereas #15 clearly depicts a "cqpsk" (LSM style) simulcast channel. (Note the huge difference in the width/height ratio of the eye openings - with the extended on-symbol dwell time of cqpsk clearly in evidence).

I posted the constellation plot to illustrate to the op what he should be seeing.

No. These two channel modulation types are apples and oranges - there are two separate demod algos in OP25 for receiving the two separate cases. The cqpsk demod is not optimal for the type of system shown in #1 (and the fsk4 demod does not use constellations). There is no reason the op in #1 should feel deterred by the poor constellation...

Max

 

[maus92]

OK thanks, the eye plot was as expected for a simulcast system. There is a very noticeable difference between posts 1 and 15 - in post 1 the channel is c4fm whereas #15 clearly depicts a "cqpsk" (LSM style) simulcast channel. (Note the huge difference in the width/height ratio of the eye openings - with the extended on-symbol dwell time of cqpsk clearly in evidence).



No. These two channel modulation types are apples and oranges - there are two separate demod algos in OP25 for receiving the two separate cases. The cqpsk demod is not optimal for the type of system shown in #1 (and the fsk4 demod does not use constellations). There is no reason the op in #1 should feel deterred by the poor constellation...

Max

Hmmm..., I've never run op25 on a multicast site. Can you explain the diffuse nature of the symbol plot in post 1?