neumoDVB

[Llew]

LLew,

I use diseqc here without any problem. Even the combination of diseqc1.0 and diseqc1.1 works.
So you may have misconfigured sometihng. Check your lnb configuration.

Ports are numbered starting from 0. So port 4 is diseqc1.0 value 0. Make sure that diseqc1.1 is turned off by setting it to -1
and that tune string has "C" in it.

Also make sure that you do not have more than 1 lnb cnfigured for 5.0W. It is fine to do that but then both configurations
must be correct.

The log files also contain the actual data sent to the swicth, so it is easy to verify.

Unrelated: Recently I also found an old bug in the drivers, which can cause problems on streams which use physical layer scrambling.
The symptom is that tuning may be slow or even fail in neumodvb. It is fixed in the latest drivers.

A delay in answering, apologies. I have had eye surgery recently (cataract), which renders my present reading glasses less than useful - mismatched lenses until the required time has passed between surgery and obtaining corrective lenses ( about four weeks time )

I'm not sure what the diseqc problem was, as I had the correct configurations (port numbering with diseqc 1.0). I did a full reinstallation, which cured the problem.
At present, I'm attempting to do a blindscan time comparison with neumodvb between my TBS6903 and TBS6903X. With the TBS6903X, time taken to do a full H scan is about 6 minutes.



With the TBS6903, although RF/SNR levels are good when tuning a frequency, the acquired spectrum signal level is way down and impossible to effect a blindscan . The tuner is obviously working fine.




On another point, the markers on found TP's on a spectrum scan are still making it hard to read the TP info at the top of the marker despite my attempts to pan and zoom down to a level without losing the bottom of the chart. (I know you have mentioned this, is it a work in hand?).

 

[deeptho]

A delay in answering, apologies. I have had eye surgery recently (cataract), which renders my present reading glasses less than useful - mismatched lenses until the required time has passed between surgery and obtaining corrective lenses ( about four weeks time )

I'm not sure what the diseqc problem was, as I had the correct configurations (port numbering with diseqc 1.0). I did a full reinstallation, which cured the problem.
At present, I'm attempting to do a blindscan time comparison with neumodvb between my TBS6903 and TBS6903X. With the TBS6903X, time taken to do a full H scan is about 6 minutes.

View attachment 140714
View attachment 140715
With the TBS6903, although RF/SNR levels are good when tuning a frequency, the acquired spectrum signal level is way down and impossible to effect a blindscan . The tuner is obviously working fine.

View attachment 140716
View attachment 140717

On another point, the markers on found TP's on a spectrum scan are still making it hard to read the TP info at the top of the marker despite my attempts to pan and zoom down to a level without losing the bottom of the chart. (I know you have mentioned this, is it a work in hand?).

tbs6903 is not supported. I guess the tuner driver would need to be adjusted (level of by a faotr of 10?), but even then the spectrum may be too poor quality

 

[Llew]

I should have remembered that, thanks for reminding me.

 

[statalite]

Any DVBT/T2 testings or results to show?

 

[Llew]

Any DVBT/T2 testings or results to show?

Unfortunately T/T2 isn't yet available in the program. Clicking on DVB-T/T2 in the Lists section brings up a page with a field and filter to put frequencies etc. in, but it's doesn't go anywhere.
Perhaps deeptho will find time to include it in due course.

 

[Llew]

@deeptho
A few years ago I used the 4T2 Content Analyser to open the C-band T2MI channels on the then new Express AM7 satellite on 40E - 40E Express AM7

As then, I'm using my Gibertini 1.2M dish, but with a DMX242 LNB instead of the Titanium.

My scan of the 40E spectrum (found wanting to lock some TPs, due to the restricted size of my dish) -



Of course, I have the usual problem of the markers stretching too far above the peaks to show the TP details (panning can only show the peaks or the highest marker details) -






What I want to do is use neumodvb instead of the 4T2 Analyser for the T2MI channels. I have tsduck installed, but neumodvb doesn't appear to recognize the T2MI pid. therefore no video.

You have mentioned recently that you have to fix the code in tsduck ?

 

[deeptho]

In order to see the t2mi muxes you first have to scan the mux (press control-M on them). Then it should appear as a second mux. In some cases, the t2mipid is not mentioned in the SI data. Then you have to create the mux yourself,
by entering the t2mi pid.

I found several bugs in the code (in neumoDVB and tsduck - both cause picture corruption). The tsduck fix should be already in tsduck git (was fixed quickly) but requires you to recompile tsuck.

I hope to commit the nuemoDVB fixes in the coming weeks. I have been side tracked by the recent storm (one antenna completely out of wack, one even with a broken ground stand) and by recent events... I am in the middle
of some changes and would like to finish them first before committing.

 

[Llew]

OK, I'll see what I can do with your suggestions. I had recompiled tsduck. Regarding scanning the mux, I just assumed that, as the channels appeared in View - Channels, it had successfully built the mux.

 

[deeptho]

That could be one of the bugs I fixed recently (mux disappears sometimes).

 

[satesco]

Hi deeptho,

I started installing your Blindscan for my dvb cards on Ubuntu 20.04. The kernel used is 5.13.0-35-generic. Installing it didn't pose any problems, so it recognized all my cards, as you can see from the picture below.
Unfortunately, there is a problem with Neumodvb. Each time, after using different versions of Cmake, the result is the same, i.e., it gives errors(2). I tried to solve this problem, but I don't see what is causing the errors. Maybe you can determine the cause of these errors in my screenshot.
Thanks in advance

 

[deeptho]

Satesco,

you probably link against the wrong library. Perhaps the python library is the problem. That is what the message says.
You link with statically with python (should be a dyamic link: .so instead of .a) and the python version is 3.8 (perhaps not a problem, but mine is 3.9).

Did you change any of the cmake settings? Did you install all prerequisites? Specifically python3-libs?

Be also careful with mixing /us/rlib and /usr/local/lib. It could very well happen that you have some left over of an older python version
installed in /usr/local//lib. I do not know ubuntu well enough to know about its conventions, but on fedora this is exactly what
would go wrong if ever you compile your own python (will end up in /usr/local/lib) and then later install the official one (will end up in /usr/lib).

In general it is better to post the log files of the build (on github. Just rename them to .txt) than screenshots which disallow cut and paste.

 

[Llew]

@satesco

If it's any help, all I have in /usr/local/lib/python3.8 (not libpython3.8) is some dist-packages (histogram, scatter-intensity). For me, python3.8 resides in /usr/lib.
My full dist-packages folder is in /usr/lib/python3 (no version number)

Linux Mint Cinnamon 20.2 kernel 5.9.0-rc7-tbs+

 

[Llew]

In order to see the t2mi muxes you first have to scan the mux (press control-M on them). Then it should appear as a second mux. In some cases, the t2mipid is not mentioned in the SI data. Then you have to create the mux yourself,
by entering the t2mi pid.

Nope. Doesn't work for me.

Entering the t2mi pid (4096) in the new mux and rescanning simply reverts the t2mipid to zero and results in a NODATA scan result.

As far as I can make out in the logfile, Debug just shows my entering 4096 in the t2mi column and subsequently rescanning the mux, with the zero t2mpid result mentioned above.

 

[satesco]

It seems that the respective folders are installed correctly.
But you better see where the error comes from, reading the whole Neumodvb installation process in Cmake.log folder (in Attach..).
I also checked the driver for 6903x with the Kaffeine player. As you can see, everything is working fine. I'm waiting for a blindscan with Neumodvb. But before that, I must of course install it properly.
I am attaching some log files, which I hope will help @deeptho indicate the source of the errors that occur when installing neumodvb.
Thanks

 

[Llew]

It seems that the respective folders are installed correctly.
But you better see where the error comes from, reading the whole Neumodvb installation process in Cmake.log folder (in Attach..).
I also checked the driver for 6903x with the Kaffeine player. As you can see, everything is working fine. I'm waiting for a blindscan with Neumodvb. But before that, I must of course install it properly.
I am attaching some log files, which I hope will help @deeptho indicate the source of the errors that occur when installing neumodvb.
Thanks

Might be your version of cmake (3.20?) - deeptho warned against using that version in the Install doc. Ubuntu usually offers 3.20 in their downloads.

I use 3.16.3. You could try a pre-3.20 from cmake's list -

Index of /files

 

[deeptho]

@satesco

If it's any help, all I have in /usr/local/lib/python3.8 (not libpython3.8) is some dist-packages (histogram, scatter-intensity). For me, python3.8 resides in /usr/lib.
My full dist-packages folder is in /usr/lib/python3 (no version number)

Linux Mint Cinnamon 20.2 kernel 5.9.0-rc7-tbs+

Well, the error message shows that the build loads something from there. So this is a local configuration error (probably).

 

[deeptho]

Nope. Doesn't work for me.

Entering the t2mi pid (4096) in the new mux and rescanning simply reverts the t2mipid to zero and results in a NODATA scan result.

As far as I can make out in the logfile, Debug just shows my entering 4096 in the t2mi column and subsequently rescanning the mux, with the zero t2mpid result mentioned above.

Llew,
it is then best to wait for the next version. It will come soon. Like I said: I fixed some problems related to this.

 

[Llew]

OK, not a pressing problem. I'm happy enough with what I already have.

 

[deeptho]

I have just released neumodvb-0.7.0. A list of changes can be found below.
Note that correct t2mi handling also requires downloading, compiling and
then installing tsduck from git, as it contains a bug.

You may also have to upgrade to the latest drivers to avoid some bugs.

neumoDVB works best on tbs 6909x or tbs6903x. It can also work with
other cards, but then functions like blindscan and spectrum scan will not work.

neumoDVB should also work with regular linux DVB drivers, but this is not well tested either.
If you experience problems with non supported cards, you can try both the regular tbs drivers and
my blindscan driver because the bugs can differ....

Such errors will be fixed when reported but non-supported cards and regular drivers are not well
tested. Although not intended - the GUI might also crash if you use unsupported functions.

Last but not least: neumoDVB also supports DVB-C (has been tested) and is intended to
work on DVB-T2. DVB-T2 has never been tested and probably does not work. The needed fixes
(if any) are probably minor, but I cannot receive any DVB-T(2) here...

-------------------
Positioner related

-More consistent layout of positioner dialog. Disable some functions when unusable

-Fixes for diseqc12, e.g., new diseqc12 value not stored when user updates it

-Positioner bug fixes: send diseqc switch commands before sending any positioner command;
report better error messages in GUI; properly handle continuous motion.

-Send diseqc switch commands before spectrum scan to avoid scanning the wrong lnb

-Prevent dish movement during mux scan

-Satellite is now only shown as "confirmed" (no question mark) if the position was actually
found in the NIT table.

-Positioner dialog: show SNR and constellation even when tuner is not locked

-Avoid interference between positioner dialog and mux scan

-Disallow starting positioner dialog from some screens, when context does not allow guessing which
lnb the user wants to use

-In positioner dialog allow negative SNRs. Yes, there are muxes which lock with a negative SNR!

-Documentation of positioner commands

-Allow rotors which can rotate by more than 65 degrees. The old code used to cause a goto south
in this case.

-------------------
Scanning related:

-Mux tables now show the source of the mux information. For instance if the frequency
was found in the NIT table or rather from the tuner and if sat position was guessed or
found in NIT

-Improved handling of bad data while scanning muxes:
0 frequency (7.3W 11411H) or other nonsensical values;
contradictory information in muxes from 0.8W and 1.0W;
invalid satellite positions;
on 42.0E, muxes are being reported as from 42.0W;
some muxes report AUTO for modulation
incorrect polarisation reported in C-band reuters mux on 22W;
incorrect polarisation reported on 7.3W 11411H
22.0E 4181V and other muxes which claim QAM_AUTO in NIT

-Fixed incorrectly reported roll-offs (may require driver update)

-Correctly handle some cases where mux or blindscan never finish

-Remove "scan-in-progress" data from database at startup to avoid an old scan from
restarting when neumoDVB is restarted

-Improved handling of muxes reported on nearby satellites (e.g., 0.8W and 1.0W). Avoid creating
duplicates in this case. The downside is that when scanning e.g., 0.8W the found mux may
appear in the list for 1.0W and thus confuse the user

-Fixed incorrect display of scan status on mux list when mux scan is in progress

-Fixed bug where incorrect pls_code and pls_mode was retrieved from driver

-Correctly save DVBS2 matype in the database and do not show this value for DVBS

-Fix incorrect lock detection

-When scanning multistreams: retain tuning info when scanning next stream

-Spectrum blindscan now moves faster to the next candidate peak when mux does not contain
a DVB stream


-------------------
t2mi related:


-Fixed corruption in t2mi streams. This requires a patched tsduck as well.

-Improved handling of non-t2mi DVB-T service information on DVB-S muxes. Avoid entering
such muxes as DVB-T

-correctly handle duplicate packets in transport streams; tsduck needs to be patched
to also handle t2mi streams correctly

------------
GUI related

-Handle some corruption in service and mux list

-Various GUI improvements.

-Changed key-binding for activating edit-mode.

-New "status list" screen shows which adapters are active and what they are doing

-Fix some crashes when ending program

-Fix some errors when service list is empty

-Allow user to select a stream_id even when dvb streams reports incorrectly that stream is not multistream

-Correctly update recordings screen when database changes

-Minimum RF level shown on live screen is now lower than before

-Fixed some issues with list filtering


------------
Other

-Fixed some bugs in the code used to look up muxes. For instance, in rare cases a mux with
the wrong polarization was returned

-Fixed bug: lnb offset correction was applied in wrong direction on C band

-Detection of whether dish needs to be moved was sometimes incorrect; database now contains
the ultimate truth on this.

-Fixes to support ubuntu 20.04

-Fixed race between closing and then reopening the same frontend, which caused sporadic assertions.

-correctly release lnb in more cases then before: when user closwes window; when spectrum blindscan
finishes

-various other bugs


=========================

 

[satesco]

I gave up Clang-10 for Clang - 11.Unfortunately with the same error 2. What exactly does this error generate? Who can tell me? Thanks.

Added:meanwhile deeptho releases a new version of Neumodvb. I'll try this one. Who knows, maybe I'll make it.
Added 1: Reading the news that brings the new version, I realize that the installation procedure of Neumodvb has not changed. So elsewhere I have to look for the solution ?!