neumoDVB

[satesco]

I tried to set the limits where your arrow is, but it doesn't seem to save the setting. I'll try again tomorrow.
Today I was more interested if it finds and locks transponders with S2X extension and low symbol rate.
This scan limit is really needed, otherwise it's hard to see all the signals and parameters found.
I'll try again tomorrow

 

[Llew]

Something to be aware of, unlike for instance with Crazyscan, setting a scan range in neumodvb doesn't spread the chosen spectrum range over the whole graph area; it just shows the chosen limits as a section within the whole range in the graph.

Or maybe there is a way of 'spreading' it over the full graph?

 

[Llew]

Something to be aware of, unlike for instance with Crazyscan, setting a scan range in neumodvb doesn't spread the chosen spectrum range over the whole graph area; it just shows the chosen limits as a section within the whole range in the graph.

Or maybe there is a way of 'spreading' it over the full graph?

Ignore. I may be wrong, I'll check tomorrow.

 

[deeptho]

I think the best test you can do with DVB applications is on satellites with more spots, those without dth signal in the area. That's why I wanted to try the 6903x card with neumodvb at 90.0E, Yamal 401.
For this satellite, I had to edit the list of satellites of the application, which stops at 75.0E. I have added those in orbit up to 100.5E, the eastern limit where I can receive AsiaSat 5.

Yes, the list is made for Western Europe, to limit the "spam"

I'm still at the beginning of using neumodvb, maybe I'll find out later how I can set the limits between which I can scan the band. For now, the application setting is on the whole Ku band,i.e. from 10700-127500 MHz.
I'm a bit annoyed that the symbol rates on some transponders don't correspond to those, say, found by Crazyscan. I'll post the spectrum obtained by neumodvb and Crazyscan to compare some parameters. For example, at 11384 H, the symbol rate is 24500 on neumodvb, and on Crazyscan, 30000. There are other transponders with different parameters.

I am afraid this is a well-known problem of people not reading the documentation and then making incorrect assumptions.

Another example, 11435 H uses the S2X extension, but on neumodvb it shows a symbol rate of 11800, while on Crazyscan it is 13500.

I also tested this satellite if neumodvb lock low symbol rate. Unfortunately at 90.0E, it didn't find any such low symbol rate. While Crazyscan was found and locked at auto search symbol rate of 492 at 12606 H, and at manual search also the one of 281 at 12556 H.
Can you get more out of neumodvb?

I cannot receive that satellite so it is difficult to comment. Like I wrote before: it can help to set a low symbol rate even when
asking a blindscan. It does not have to be the correct one.

Also, the peak finding algorithm is different (not: better or worse, but different) than crazycat's and is not optimised for these rare cases. If a peak is not detected, you will have to enter a close enough frequency yourself. The frequency might be slightly different from crazycat, because my code learns about lnb frequency drift and corrects for it. The symbol rate is exactly the same (read the documentation).

The actual code for programming the chip is probably very close to that in crazycat's closed source driver. So there should be little difference, except due to luck (some things may work better with crazycat. Other things better with neumodvb).

Blindscan is always a compromise between speed and between preferring one type of problem versus another. It also
differs from card to card and it is possible to keep optimizing it for ever, and people will still find something that does not work. Ad fixing one problem then tends to introduce other problems.

 

[deeptho]

Something to be aware of, unlike for instance with Crazyscan, setting a scan range in neumodvb doesn't spread the chosen spectrum range over the whole graph area; it just shows the chosen limits as a section within the whole range in the graph.

Or maybe there is a way of 'spreading' it over the full graph?

That is correct: it only determines what to scan. You can use the zoom and pan buttons for a clearer view. They are a bit tricky to use, as they are part of "matplotlib", but as this provides zoom and pan for free, I prefer to not look a gift horse in the mouth. The graph data is saved to disk, so you can always use your own plotting program.

 

[satesco]

Tests with 6903X card at 33.1W, where we have a tv package(Swiss) using S2X extension. At 12597 V,10057,26/36 the spectrum shows another symbol rate than the real one, but after blindscan the app shows the real one but with an error in the PLS code. In fact, the code is the default one,1. Because of this error, although the tv programs have been saved, they can't be seen unless I change this code to 1.
Also, I could not lock the Bulgarian NURTS package at 12646 H,13910,3/4, stream 3, both by blindscan, and manually, even if I specified the stream used(3).
It often happens that when I get blindscan, the application stops at one of the frequencies without continuing, and if I manually move to the next one, after a while it returns to the previous frequency.

 

[Llew]

It often happens that when I get blindscan, the application stops at one of the frequencies without continuing, and if I manually move to the next one, after a while it returns to the previous frequency.

I get that occasionally too. For me, it's usually after about 75% of the blindscan completed.

Tests with 6903X card at 33.1W, where we have a tv package(Swiss) using S2X extension.

33.1E satesco. I wasn't aware of those S2X TPs. I'll see how I manage with those myself tomorrow.

 

[deeptho]

Tests with 6903X card at 33.1W, where we have a tv package(Swiss) using S2X extension. At 12597 V,10057,26/36 the spectrum shows another symbol rate than the real one, As I have

Well, consult the documentation about this (see the above posts)

but after blindscan the app shows the real one but with an error in the PLS code. In fact, the code is the default one,1. Because of

This was a bug in some older version of neumodvb. What version are you running?
I just scanned the complete vertical band of that satellite. The reported code is indeed 1.

this error, although the tv programs have been saved, they can't be seen unless I change this code to 1.
Also, I could not lock the Bulgarian NURTS package at 12646 H,13910,3/4, stream 3, both by blindscan, and manually, even if I specified the stream used(3).

You also have to specify the pls_code: 218997. This part of the code uses an internal table.

It often happens that when I get blindscan, the application stops at one of the frequencies without continuing, and if I manually move to the next one, after a while it returns to the previous frequency.

That too was an old bug, but at least today I cannot reprodcue it.

The current version of the code is 0.7

 

[deeptho]

And about returning to the previous frequency: as long as the blindscan is not finished (including because it
is stuck which would be a bug) then it will retry. You can turn that off by clicking on blindscan all until it becomes light grey.

 

[Llew]

33E. I'm not locking many TPs on this satellite. At least the TBS6903X finds the Swiss S2X channels on 12597V.

Also, the Bulgarian Stream 3 channels on 12646H.

I'll check lower down the band for some feeds later.

 

[deeptho]

More changes are underway for the next version. This includes better layout of the spectrum plot, making it
easier to see all the text without having to scroll up and down.

 

[satesco]

I also managed to scan and memorize the Bulgarian package.
I did tests with tbs cards installed on my PC. The tbs6903x card has the best support, both in spectrum and scanning, while the 6983 is the weakest.
I don't know any other neumodvb version than the one on Deeptho's website. Nothing is specified there.
I'll comment on the application after I do tests on several satellites.

 

[deeptho]

I have just released a new version of neumoDVB: neumodvb-0.8.0
There are several new features and many bug fixes.

​

Installation related​

• Improved installation instructions for fedora 36 and Ubuntu 20.04 LTS.

Spectrum acquisition and blindscan related​

• Added Export command, which can export service, mux, ... lists.
• Allow selecting multiple muxes for scanning, instead of selecting them one at a time and pressing `Ctrl-S` each time
• Allow blind scanning non detected peaks by selecting a region of spectrum graph.
• Improved positioning of frequency/symbol rate annotations on the spectrum graph. The labels are better readable and use less vertical space, preventing the need for vertical resizing in many cases.
• Added predefined pls code for 33.0E (BulgariaSat).
• Avoid python exception when spectrum scan results in zero detected peaks
• Avoid confusion between embedded and non-embedded streams on same frequency leading to missing or overwritten muxes in the database
• Bug when multiple subscriptions use the same mux, but one of them uses a t2mi substream.
• Prevent scan hanging for ever on some muxes
• Ensure that scan status is properly handled with embedded streams to prevent scans that never finish.
• Data race when starting embedded_stream_reader, causing tsp sub-process to be launched multiple times.
• Incorrect scan result reported on t2mi stream because of incorrect lock detection
• Autodiscover t2mi streams on 40.0E (workaround for detecting incompliant t2is)
• Handle empty NIT_ACTUAL in embedded stream on 40.0E 3992L T2MI. This causes the ts_id and original_network_id from the embedded t2mi SDT not to be taken into account.
• When tuning to t2mi mux from database record, preserve original_network_id and ts_id in the embedded si
• Allow resetting LNB LOF offset

EPG and recording related​

• Fixed layout bugs in grid epg.
• Improved handling of 'anonymous' recordings. These are recordings created by the user by entering start and end time instead of recording an epg record. In the new code, these anonymous recordings will be kept as instructed by the user, but also secondary recordings will be created for each epg record matching the time window. This is convenient for services with only now/next epg, which may arrive after scheduling the anonymous recording. Anonymous recordings will be shown on the EPG screen, but will be overlayed with epg records, should these exist. Canceling a secondary epg recording also cancels the anonymous recording.

Playback and viewing​

• During live viewing and recordings playback, replace the original pmt with a minimal pmt containing only the video stream, and current audio and subtitle stream. This should prevent playback problems of services on 30.0W which share the same pmt. mpv tends to get confused about the audio stream selection in this case. This should also fix the missing sound on Sport Italia HD (5.0W).
• Incorrect indication of live/timeshift/rec status in live overlay
• Hide snr and strength when playing back a recording.
• Reduce buffering in tsp sub process to avoid delay when live viewing t2mi services

Various bugs fixed​

• Anonymous recordings 60 times too long.
• Cannot activate edit mode in lnb network list

 

[Llew]

I have just released a new version of neumoDVB: neumodvb-0.8.0
There are several new features and many bug fixes.

​

Installation related​

• Improved installation instructions for fedora 36 and Ubuntu 20.04 LTS.

Spectrum acquisition and blindscan related​

• Added Export command, which can export service, mux, ... lists.
• Allow selecting multiple muxes for scanning, instead of selecting them one at a time and pressing `Ctrl-S` each time
• Allow blind scanning non detected peaks by selecting a region of spectrum graph.
• Improved positioning of frequency/symbol rate annotations on the spectrum graph. The labels are better readable and use less vertical space, preventing the need for vertical resizing in many cases.
• Added predefined pls code for 33.0E (BulgariaSat).
• Avoid python exception when spectrum scan results in zero detected peaks
• Avoid confusion between embedded and non-embedded streams on same frequency leading to missing or overwritten muxes in the database
• Bug when multiple subscriptions use the same mux, but one of them uses a t2mi substream.
• Prevent scan hanging for ever on some muxes
• Ensure that scan status is properly handled with embedded streams to prevent scans that never finish.
• Data race when starting embedded_stream_reader, causing tsp sub-process to be launched multiple times.
• Incorrect scan result reported on t2mi stream because of incorrect lock detection
• Autodiscover t2mi streams on 40.0E (workaround for detecting incompliant t2is)
• Handle empty NIT_ACTUAL in embedded stream on 40.0E 3992L T2MI. This causes the ts_id and original_network_id from the embedded t2mi SDT not to be taken into account.
• When tuning to t2mi mux from database record, preserve original_network_id and ts_id in the embedded si
• Allow resetting LNB LOF offset

EPG and recording related​

• Fixed layout bugs in grid epg.
• Improved handling of 'anonymous' recordings. These are recordings created by the user by entering start and end time instead of recording an epg record. In the new code, these anonymous recordings will be kept as instructed by the user, but also secondary recordings will be created for each epg record matching the time window. This is convenient for services with only now/next epg, which may arrive after scheduling the anonymous recording. Anonymous recordings will be shown on the EPG screen, but will be overlayed with epg records, should these exist. Canceling a secondary epg recording also cancels the anonymous recording.

Playback and viewing​

• During live viewing and recordings playback, replace the original pmt with a minimal pmt containing only the video stream, and current audio and subtitle stream. This should prevent playback problems of services on 30.0W which share the same pmt. mpv tends to get confused about the audio stream selection in this case. This should also fix the missing sound on Sport Italia HD (5.0W).
• Incorrect indication of live/timeshift/rec status in live overlay
• Hide snr and strength when playing back a recording.
• Reduce buffering in tsp sub process to avoid delay when live viewing t2mi services

Various bugs fixed​

• Anonymous recordings 60 times too long.
• Cannot activate edit mode in lnb network list

I downloaded neumodvb 0.8.0, but neumodvb.py crashes upon execution. Until I try to figure out what's amiss, I've reverted to the my working version.

 

[Llew]

I had tried a git pull on this latest version, but if I recall, it aborted on files in neumowxsvg (missing and depcomp I think).

 

[satesco]

Sorry, you couldn't work with version 0.8.0. I installed it with no problems. I tested it at 33.1E again with the 6903x card.
Scanning works fine with this card, but I don't understand why the pls code appears on some transponders that don't use this code? That's why you have to edit to scan and save manually the found channels.
The difference between Crazyscan on Windows and Deeptho's blindscan is that Crazyscan displays only the transponders it locks, while on Linux there are many transponders but few of them can be locked.
Deeptho explained this by the different algorithms he uses.

 

[deeptho]

I downloaded neumodvb 0.8.0, but neumodvb.py crashes upon execution. Until I try to figure out what's amiss, I've reverted to the my working version.

You could report what the crash message said. I justed tested on another computer and no problems occur for me.

 

[deeptho]

I had tried a git pull on this latest version, but if I recall, it aborted on files in neumowxsvg (missing and depcomp I think).

If it could not compile, then how can it crash?

 

[deeptho]

Sorry, you couldn't work with version 0.8.0. I installed it with no problems. I tested it at 33.1E again with the 6903x card.
Scanning works fine with this card, but I don't understand why the pls code appears on some transponders that don't use this code? That's why you have to edit to scan and save manually the found channels.

It happened in (much) older versions on tbs6903x due to a driver bug. I suspect it comes from using some of your other cards, for which the driver may have a similar problem. Specifically the question marks in some of the results point to such a problem. In any case, I do not see this problem myself and I scanned 33.0E very recently.

Try deleting all the muxes and services on one sat and rescanning only with tbs6903x. If this works, then you could do a similar experiment with another card. If I know the cause I can perhaps fix it.

Note that if you perform a manual scan (click on tp and then press tune), then whatever values for PLS was last used
could still be present a the default value. Then you have to change it.

The difference between Crazyscan on Windows and Deeptho's blindscan is that Crazyscan displays only the transponders it locks, while on Linux there are many transponders but few of them can be locked.
Deeptho explained this by the different algorithms he uses.

There is a difference between "spectral peaks" and muxes. Spectral peak detection is the first step of blindscanning. It also happens with all other scanning methods and drivers, except that they don't tell you.

 

[satesco]

Why does the app crash when I tune to pip?