neumoDVB

[satesco]

Deeptho,

Errors occur when reinstalling neumoDVB. How to eliminate this? Thanks.

 

[Llew]

Llew,
This crash was recently fixed, but the patch was not yet on github. Sorry for that.

I did a git pull and received your latest pushes, but I still have the crash problem. This is with my kernel 5.9.0-rc7-tbs+ in Ubuntu.

I'm using that kernel, as it has working frontends for my TBS cards (adapters shown in /dev/dvb).

The generic kernels installed with Ubuntu 20.04 (5.13.0.30 and 5.13.0.40) don't show my adapters, so they're obviously not of use. I tried 5.13.0.39-generic mentioned by @satesco earlier, but that's similarly bereft of adapters.

So my "working" kernel is useful only with neumodvb v.07. v.0.8 seems unstable/unusable in my PC.

 

[deeptho]

I did a git pull and received your latest pushes, but I still have the crash problem. This is with my kernel 5.9.0-rc7-tbs+ in Ubuntu.

I'm using that kernel, as it has working frontends for my TBS cards (adapters shown in /dev/dvb).

The generic kernels installed with Ubuntu 20.04 (5.13.0.30 and 5.13.0.40) don't show my adapters, so they're obviously not of use. I tried 5.13.0.39-generic mentioned by @satesco earlier, but that's similarly bereft of adapters.

So my "working" kernel is useful only with neumodvb v.07. v.0.8 seems unstable/unusable in my PC.

Llew,
I made a mistake in my answer yesterday: the code was up to date. I tried checking it out from start in a virtual machine and it works fine.
I think it may be a problem of something left from an older version compilation and not being reocmpiled. It should not happen, but remember there is some bug in cmake. Perhaps that could explain it.

You could try:
cd build
rm -fr *
cmake ..
make

 

[deeptho]

Deeptho,

Errors occur when reinstalling neumoDVB. How to eliminate this? Thanks.

This seems to be due to outdated compilers on ubuntu. I think I have fixed it now.
The best thing to try is
cd neumodvb/build
git pull
rm -fr *
sudo apt-get install lsb-core lsb-release #this is a new dependency
cmake ..
make -j8

This works in a virtual machine. Also blindscan works as it should. A quick test in 28,.2E shows that
scanning is also rather fast: 23 muxes in 3 minutes and 20 seconds for one polarisation. Other sats will scan
slower. The "fuller" the sat, the faster.

 

[Llew]

Llew,
I made a mistake in my answer yesterday: the code was up to date. I tried checking it out from start in a virtual machine and it works fine.
I think it may be a problem of something left from an older version compilation and not being reocmpiled. It should not happen, but remember there is some bug in cmake. Perhaps that could explain it.

You could try:
cd build
rm -fr *
cmake ..
make

Trying that, I now have the same make errors as @satesco.
I will attempt the possible fix in your reply to that post.

 

[Llew]

Fix working for me @deeptho, error-free, Thanks.

neumodvb now not dropping out during spectrum scan or blindscan, but does always after 5 seconds whilst viewing a channel. I shall check that problem later (logs etc)

Also, I'm having a problem blindscanning - It seems to be staying forever on a frequency (here, 10928H) despite my 'moving it forward' manually.

 

[Llew]

Tried another BS, passes 10928H, now staying on 11220H, 90minutes and counting...

Appears to be when waiting for the NIT, then somehow finding it Invalid (?)



Terminal result when viewing channel drops out after 5 seconds:

 

[satesco]

Now I can work with neumoDVB after fixing it. The scan time with 6903x is better than the previous version. It is also faster than Crazyscan under Windows.
I locked 281 Ks on 12556 H at 90.0E, the lowest symbol rate achieved with this card, the same as Crazyscan. But I didn't get 256 Ks, Radio Maria at 10.0E. From this point of view, I don't think you can ask more from 6903x/6909x whether it works on Linux or Windows.
I attach the scan with neumoDVB and Crazyscan at 90.0E with 6903x,6909x(V1), and 6903 cards. How many steps are set when scanning with neumoDVB? Can this be set manually? Maybe it helps.
I have also attached the neumoDVB log. As with Llew, sometimes the application hangs on one frequency and doesn't work on the next except manually.
Thanks @deeptho

 

[deeptho]

Tried another BS, passes 10928H, now staying on 11220H, 90minutes and counting...

Appears to be when waiting for the NIT, then somehow finding it Invalid (?)

View attachment 142791

Terminal result when viewing channel drops out after 5 seconds:

View attachment 142792

I think what happens is that the blindscan is still in progress and then tunes to another mux. This is not a normalk use case. It should not crash of course. The normal behaviour would be to stop live viewing if blindscan needs the lnb and show an error message.

 

[deeptho]

Now I can work with neumoDVB after fixing it. The scan time with 6903x is better than the previous version. It is also faster than Crazyscan under Windows.
I locked 281 Ks on 12556 H at 90.0E, the lowest symbol rate achieved with this card, the same as Crazyscan. But I didn't get 256 Ks, Radio Maria at 10.0E. From this point of view, I don't think you can ask more from 6903x/6909x whether it works on Linux or Windows.
I attach the scan with neumoDVB and Crazyscan at 90.0E with 6903x,6909x(V1), and 6903 cards. How many steps are set when scanning with neumoDVB? Can this be set manually? Maybe it helps.
I have also attached the neumoDVB log. As with Llew, sometimes the application hangs on one frequency and doesn't work on the next except manually.
Thanks @deeptho

If it hangs on one frequency, then report the frequency and satellite. Note that it also may APPEAR to hang. It needs to wait for a timeout, sometimes it needs to scan many streams on the same mux and so on.

About the low symbol rates: I think a lot depends on what happens on neighboring frequencies. It may be that you need a stringer signal because of that. E.g, on 5.0W I could reveive last year Alouette and Nostalgy with low symbol rates, but not anymore

 

[satesco]

I don't think you're getting 85.0E in your area,because I'd like you to check 11078 V which crashes every time,either at auto blindscan,or manually.I don't know if it helps,but I attach the file with neumo's log and spectra, which looks ok. But I hope from the log you can find out where the bug is.
I'm not convinced that because of the weak signal I can't lock low symbol rate with 6903X, but (if you accept the opinion of a noob, who has worked a lot with this card on Windows), rather it's a problem that is not only limited to the driver, but even in the demodulator.
With stv010X I have a 14dB signal, whereas with the stid135 it is weaker.But even if it were strong, such a card only rarely lock low symbol rate. So far the 6903x reaction to low symbol rate is similar to that under Windows, at least from tests done on different transponders at different positions.
So I can't expect miracles on Linux from this point of view, but I prefer to be wrong and to be disproven, even by your application. May the One above grant that you succeed

 

[deeptho]

I don't think you're getting 85.0E in your area,because I'd like you to check 11078 V which crashes every time,either at auto blindscan,or manually.I don't know if it helps,but I attach the file with neumo's log and spectra, which looks ok. But I hope from the log you can find out where the bug is.
I'm not convinced that because of the weak signal I can't lock low symbol rate with 6903X, but (if you accept the opinion of a noob, who has worked a lot with this card on Windows), rather it's a problem that is not only limited to the driver, but even in the demodulator.
With stv010X I have a 14dB signal, whereas with the stid135 it is weaker.But even if it were strong, such a card only rarely lock low symbol rate. So far the 6903x reaction to low symbol rate is similar to that under Windows, at least from tests done on different transponders at different positions.
So I can't expect miracles on Linux from this point of view, but I prefer to be wrong and to be disproven, even by your application. May the One above grant that you succeed

I cannot receive that mux, but there is also nothing in the log indicating a crash. Send me the log file with the crash in it (as a text file)

About the low symbolrates: locking can be improved b y changing some timeouts in the drivers, which can lead to longer scanning times,
which is why I reduced them. Yesterday I succeeded on 5.0W to tune several muxes with 500-600kS/s symbol rates after making some timeouts 4x larger.

 

[satesco]

Hi Deeptho,

I had to reinstall neumoDVB. I managed to install the app but at the end I find an error generated, it seems, by a python3 package.
Please take a look at the attached file and tell me how to fix it.
Thanks

 

[deeptho]

Hi Deeptho,

I had to reinstall neumoDVB. I managed to install the app but at the end I find an error generated, it seems, by a python3 package.
Please take a look at the attached file and tell me how to fix it.
Thanks

It seems you have a problem with you roperating system: wrong wxpython installed, or existing version
corrupted

 

[satesco]

I installed neumodvb without generating errors. For reasons that escape me, when I use the command line to start neumodvb, I get the following message:

"sat@ubu:~/neumodvb/build$ ~/neumodvb/gui/neumodvb.py
Traceback (most recent call last):
File "/home/sat/neumodvb/gui/neumodvb.py", line 22, in <module>
import wx.glcanvas #needed for the created mpvglcanvas
File "/usr/local/lib/python3.8/dist-packages/wx/__init__.py", line 17, in <module>
from wx.core import *
File "/usr/local/lib/python3.8/dist-packages/wx/core.py", line 12, in <module>
from ._core import *
ImportError: libpng12.so.0: cannot open shared object file: No such file or directory"

Where do you think I should intervene to fix this error? Thanks

 

[deeptho]

I installed neumodvb without generating errors. For reasons that escape me, when I use the command line to start neumodvb, I get the following message:

"sat@ubu:~/neumodvb/build$ ~/neumodvb/gui/neumodvb.py
Traceback (most recent call last):
File "/home/sat/neumodvb/gui/neumodvb.py", line 22, in <module>
import wx.glcanvas #needed for the created mpvglcanvas
File "/usr/local/lib/python3.8/dist-packages/wx/__init__.py", line 17, in <module>
from wx.core import *
File "/usr/local/lib/python3.8/dist-packages/wx/core.py", line 12, in <module>
from ._core import *
ImportError: libpng12.so.0: cannot open shared object file: No such file or directory"

Where do you think I should intervene to fix this error? Thanks

Satesco,

i think we discussed this problem before: you have some left overs of an old installation on your computer.
I sent you a private message regarding teamviewer.

 

[satesco]

Thanks @deeptho for your help. A first test with 6903x

 

[deeptho]

NeumoDVB 0.9.1 was released today. Here is the list of chamges.
With the latest drivers (also uploaded today) m88rs6060cards are also supported: tbs6902SE (untested) and tbs6904SE
IQ scan does not work on these cards, but everything else does.

SI-processing related

• Incorrect activation of PMT parsers
• Avoid showing incorrect SI data in tuned mux panel
• Faster detection of stable pat, resulting in faster SI scanning
• Better handle invalid SI information on 7.0E
• Handle SI-sections with different version numbers
• Also parse PMTs during regular SI scanning
• Avoid needless frequent epg scanning on services without epg

Spectrum and blind scanning

• When adapter is not available, show error message and gracefully end blindscan
• Properly handle multi-stream with the streams having the same network and ts id. These muxes would overwrite each other while scanning. Happens, e.g., on 14W.
• Avoid python back-traces when no peaks found in spectrum
• Fixed bug: when LNB changes due to clicking on a mux in the spectrum panel, the LNB select combobox does not show the newly selected LNB, which will be used for tuning but uses it anyway. This causes blind tunes to happen on unexpected LNBs
• Improvement: when the user loads a spectrum from file in spectrum scan, then selects mux and blindscans, use the current LNB whenever possible, except if this one cannot tune the mux (wrong sat or wrong type of LN. Only in that case, switch to the LNB originally used to capture the spectrum.
• Include not only hours and minutes but also seconds in names of files storing spectrum data. This solves the problem that spectrum acquisitions started within one minute would overwrite each other.
• The code now asks for a 50kHz resolution spectrum, leading to slightly slower scans in stid135. This detects more low symbol rate muxes.
• Fixed bug: DVBS2-QPSK streams incorrectly entered as DVBS1 streams in database
• Fixed bug: incorrect detection of roll-off parameters.
• More useful packet or bit error rate estimation. The new code measures errors before FEC/Viterbi/LDPC correction. This requires blindscan drivers with version > release.0.9.

Installation

• Prevent some files from installing

GUI

• Show disabled menu items as grayed out
• No longer use the single letters 'E' and 'C' as shortcuts, except in live mode
• Fixed bug that causes channels to scroll offscreen
• Fix bug that prevented proper navigation with arrow keys on EPG screen
• Fixed bug: onscreen display of SNR not working in radio mode
• Properly handle cases with unavailable SNR
• Fixed some layout bugs
• Add timing lock indicator (replaces pretty useless signal indicator)
• Immediately activate changes entered by user on "frontend list" screen. This is important when user enables/disables a frontend
• Add channel screenshot command
• Show more stream numbers in the positioner dialog and in the spectrum dialog on muxes with large number of streams

Compilation and internals

• Switch to clang20 by default except on "ancient Ubuntu" aka "Ubuntu LTS"
• Fix double close of cursors and incorrect freeing, leading to assertion failures
• Add command for building fedora rpms. Main problem to start really using this is a dependency on mpl-scatter-plot, which cannot be installed as an rpm
• Fix incorrect recompiling of pre-compiled headers with cmake > 3.19
• Removed dependency on setproctitle and made it less error-prone
• Add script for testing peak finding algorithm

Diseqc

• Add more time between sending commands to cascaded swicthes, solving some erroneous switching

 

[satesco]

Thanks, Deeptho for the drivers for the m88rs6060cards chip and the improved version of the neumodvb app.
This time I managed for the first time to use the 6902se card and run it on Linux as under Windows, with all the lack of documentation for this chip.
I had the opportunity to check the new features you bring now, both in drivers and the Neumodvb app on MIS transponders with PLS code at 5.0W.
The PLS code on these MIS transponders was auto-detected by your app when scanning, without doing an extra operation, as Crazyscan does under Windows.
I also managed to lock the low symbol rate of 667 at 11565 H, and at one point also 542 at 11453 H, but at the limit.
Without exaggeration, I can say that you give life to DVB cards on Linux, just like CrazyCat does with these cards under Windows. Without you, our hobby would be poorer

 

[Llew]

NeumoDVB 0.9.1

@deeptho

A partial scan check of Hispasat 30W

Blindscan was proceeding in good time until stuck forever at the end H TP scanning in this example. It's not the sole TP that's stalling; there are others that do the same during my test spectrum runs - I haven't yet been able to complete a scan because of this.

What are the isi numbers that are being continually run through while it's stuck on the TP in the service results section? Is there a clue there as to why it stays halted checking the TP?

Edit > Just realised isi = input stream identifier