267 replies to this topic

[Dimitrij]

The new project.
Let's make a new version of the scanning(blindscan) services in the image.

For all models.
Speak ideas on how it should look.

Edited by Dima73, 13 January 2015 - 20:06.

[Erik Slagter]

You can't make it for all models.

 

Tuner needs to support blindscanning (hardware)

Manufacturer needs to supply closed source binary, there is no v4l api

 

If we could do this, I much prefer to do it within enigma2, because enigma2 manages the dvb resources.

[littlesat]

Not completely true... For blindscan tuners you only need to configure polarisation, band, frequency and symbol rate.... Fec, modulation etc is automatically found... You can build a general blindscan with these facts....

[el bandido]

What about getting USB tuners to blindscan? Seems that this would be an easy way to add DVB-S2 blindscan feature to different receivers like the Vu+ Zero.

[Rob van der Does]

Not completely true... For blindscan tuners you only need to configure polarisation, band, frequency and symbol rate.... Fec, modulation etc is automatically found... You can build a general blindscan with these facts....

That will be a 'software blindscan' than if I understand correctly. Much slower and less reliable then a hardware blindscan.
for the latter you need the support of the drivers and hence the blindscan binarie we get from manufacturers.

[Dimitrij]

Not completely true... For blindscan tuners you only need to configure polarisation, band, frequency and symbol rate.... Fec, modulation etc is automatically found... You can build a general blindscan with these facts....

For sure.
That is what I propose to do.

[littlesat]

That will indeed be some kind of soft blindscan... But you cannot say ithe result will be less reliable... I suggest it might be even more reliable... I suggest it will be slower but because it is never made we do not have anything to compare...

Beside blindscan I better prefer to have a way to get some kind of spectrumanalyzer in e2, where you can choose a peek and then perform a scan on that peek... I personally still waiting for an implementation like that in some manufacurer's drivers...(I do not like that binary that is required for blindscans). As far i know the dr. HD box has such kind of functionality (non E2)

Edited by littlesat, 14 January 2015 - 08:01.

[jeanclaude]

you know, a long time ago, when I was still working with my DM7000 in enigma1, I had this satellites.xml file (from some forum, think satellites.co.uk) which contains all possible frequencies using all possible format specifications currently in use, whatever satellite. This scan would take a very long time (as 90% of the transponder info is probably wrong) but when finished you really had a complete scan of anything receivable on any satellite, so that's what a blind scan should do, yes ?

obviously, now with dvb-s2 where the FEC cannot be automatic and you have qpsk and 8psk, this would make one hell of a transponder list, if we are to select all the possible transmission protocols for all of the transponder frequencies which are being used anywhere on any satellite. But if it takes the whole night to have the box scan all possible variations, why not ? And you needn't do anything special or program anything, you just have to create the necessary satellites.xml and it will work on any enigma receiver.

[malakudi]

Software "blind" scan will take forever. Even with auto FEC (which not all tuners support), with 5MHz increment, you have 480 combinations for each polarity for each symbol rate. And which symbol rates you will test? If incrementing symbol rate from 2000 to 45000 every 50 symbols (which still migh fail to tune some intermediate values), there are 860 possible values. So 480*2*860, without testing DVB-S QPSK/DVB-S2 QPSK or 8PSK. If you need 3 tests for that, the total number of combinations is 2,5 million. If you take 2 seconds to test lock (which might be not enough for weak signals), you need 5 million seconds, or about 57 days.

 

Bottom line: Software "blind" scan is impossible. You can only do it if you narrow very much the frequency and SR range - or search for a few standard SR values.

[Rob van der Does]

I thought something like that (although I didn't expect it to take that long).
I also don't like the fact that we need (closed source) binaries from manufacturers to get hardware blindscan to work. But that's fact of life, simple as that, just as we don't like to be dependant on drivers on which we have no influence.
so let's be happy when we do get those binaries, and make the general part of the plugin, the GUI, as good as we can. And report blindscan issues to manufacturers, in the hope that they will address them (as VU did in the past).

[littlesat]

That was why actually you need a spectrum analyzer approach.... not supported by any E2 box tender yet...

And you do not need to scan all these symbol rates......

 

As far my experience reaches I never found a real optimal blind scan. Having a good complete satellite.xml is the best approach...

Edited by littlesat, 14 January 2015 - 12:28.

[jeanclaude]

you don't have to test every possible combination for a software scan, only all the combinations which are currently being used anywhere. True, this _will_ be a big list.....

[Huevos]

 

Not completely true... For blindscan tuners you only need to configure polarisation, band, frequency and symbol rate.... Fec, modulation etc is automatically found... You can build a general blindscan with these facts....

For sure.
That is what I propose to do.

 

You can't be serious. This is not blind scan. It is "brute force search". This is the most inefficient code breaking strategy known to man. You would need to compile a list of every single possible combination of frequency, SR, FEC, and modulation. That list would contain literally millions of entries that E2 would have to scan.Compare this to a scan of a busy satellite like Hotbird that contains 100 active transponders. Takes a good 15 minutes. And to make things worse, when E2 is give a frequency that it can't lock it hangs around for a few minutes on that frequency.

[Huevos]

That was why actually you need a spectrum analyzer approach.... not supported by any E2 box tender yet...

And you do not need to scan all these symbol rates......

 

As far my experience reaches I never found a real optimal blind scan. Having a good complete satellite.xml is the best approach...

Well you need to try the plugin in the feeds then. And by the way this plugin relies on you NOT "Having a good complete satellite.xml".

 

Having a satellites.xml full up with rubbish slows E2 scans, and as feeds, etc a usually not repeated, you are not going to find much by scanning old frequencies.

Edited by Huevos, 14 January 2015 - 12:45.

[Dimitrij]

Huevos

I thought that was you support my idea.

[Huevos]

Huevos

I thought that was you support my idea.

 

I don't understand. If you mean did I support porting the plugin for PLi then the answer is yes.

 

What I don't support is the suggestion of software blindscan.

 

Lots of STBs work with the OE-A blindscan plugin but we've never attempted to try and get blindscan functionality where a receiver doesn't have the hardware, as this is just not feasible.

[Huevos]

What I don't understand is why OpenPLi is soldiering on with their stupid blindscan plugin with all the zeros, when this lovely plugin is available.

[littlesat]

Then offer us a good plugin....

[Erik Slagter]

What I don't understand is why OpenPLi is soldiering on with their stupid blindscan plugin with all the zeros, when this lovely plugin is available.

Please explain? (this probably the plugin I made once...)

[Erik Slagter]

BTW I agree, blindscan in software without any assistence of hardware in the tuner, is futile.