piratfm / eti-tools

ETI conversion software
Mozilla Public License 2.0
50 stars 18 forks source link

Satellite sources of DAB feeds - part 3 #35

Open andimik opened 4 years ago

andimik commented 4 years ago

Summary of closed issue #28 (= Satellite sources of DAB feeds - part 2)

jpuigs commented 4 years ago

I've read today that Media Broadcast has been awarded the contract to broadcast DAB+ services for NDR and RBB regional broadcasters for 4 years (+2 optional ones) So, as other Media Broadcast DAB+ emsembles are on Eutelsat 7ºE and Astra 23.5ºE, we should be watching those positions... maybe something appears....

https://www.radioszene.de/142405/ndr-rbb-media-broadcastdigitalradio-dab.html

andimik commented 4 years ago

Well, thanks for your message. But I personally don't expect them via satellite in future as the NDR will be split to regional multiplexes soon.

The only one which might be on satellite is Bundesmux II which starts later this year.

jpuigs commented 4 years ago

I read some days ago (thanks to online translators) that NDR was going to expand DAB+ this year on some regions, and regionalization of the NDR 1 national programmes would take place in the summer. But as some regional ensembles are found on 7ºEast (Bayern, BW and Hessen) , let's think it could be possible too.....

jpuigs commented 4 years ago

DAB Italia and EuroDAB Italia feeds have left 5º West, 12690 V.

andimik commented 4 years ago

Thanks, see 6096328

andimik commented 3 years ago

SDL and D1 are now also on 9 East, 12092H, but in DVB-S2. Readme has been updated.

Remark: BBC is still at 4.8 East. Screenshot (174)

jpuigs commented 3 years ago

So.......... we may supose that one of the two current feeds (4.8 E or 28 E) will be removed. Which one do you think will be removed ? I've never understood why the BBC has only one feed, and the two private ones (D1 and SND Dig) have two.

newspaperman commented 3 years ago

Anyone spotted BMux2 already?

jpuigs commented 3 years ago

I've tried to look for small carriers near the current known ones, at 7 East (12572 V) and 23.5 East (1264X V), but I don't find anything. 😢

andimik commented 3 years ago

A German DXer informed me that the Turkish transport stream only consists of NULL bytes at the moment.

$ cat ~/Downloads/42e_trt-dab.ts | tsanalyze --pid-list
8191
newspaperman commented 3 years ago

@jpuigs ok it seems the SFN s not yet active, there are terrwstrial tests already, but they only produce one single service "program 1" without audio.

jpuigs commented 3 years ago

But I've seen over last weeks on german forums that there are some programms on air. I supose there were sporadic "local" tests.

image

newspaperman commented 3 years ago

You are rights the test mux was active on one transmitter only, I think there the Dab muxer is located.

jpuigs commented 3 years ago

A German DXer informed me that the Turkish transport stream only consists of NULL bytes at the moment.

$ cat ~/Downloads/42e_trt-dab.ts | tsanalyze --pid-list
8191

Still NULL packets.....

image

andimik commented 3 years ago

TRT is back.

jpuigs commented 3 years ago

👍

andimik commented 3 years ago

Anyone spotted BMux2 already?

Yes, I found it, but it's in GSE.

Astra 23,5 East, 12668 V, DVB-S2, SR 1500, 8PSK Both are there, Bundesmux and Bundesmux 2

bundesmux bumux.ts.gse.zip

Depending on the TBS card it says encapsulated, generic or continuous stream. And I fear the MIS is 1 or even 2 (depending on the card).

So no chance for us (yet).

Have list it in the README.md, see https://github.com/piratfm/eti-tools/commit/c7ef54b72833b41cf59b23156f144b64b21ce5e1

jpuigs commented 3 years ago

Good and Bad news. But if we can't see what's inside, how do you know they are both Bmux streams ?

andimik commented 3 years ago

If you want to see what's inside, why not using a (Hex-)Editor ... ?

Here is a longer file, captured by TBS6903x Professional.

23.5E_12668,085_V1500(2020-09-26 12.54.15)_dump.zip

grafik

grafik

grafik

During the current tests German forums and Facebook reports show the same only service "Programm 1"

After a while the old DVB-S Stream will disappear for sure as GSE is much more modern and accurate.

If someone wants to try with a SDR (software defined radio) solution: Use a DVB-T (!!) Stick with E4000 Tuner (due to the very high frequency above 2 GHz) or HackRF or other and http://www.pabr.org/radio/leandvb/leandvb.en.html and then please report if you can see and use (!) the BB frames.

jpuigs commented 3 years ago

If you want to see what's inside, why not using a (Hex-)Editor ... ?

I did it, but I didn't notice anything about Bmux or Bmux2. Maybe I looked for too fast...... Which editor are you using ? (Speedview?) I use Hex editor neo, but I think yours is "lighter". I only have two RTL-SDR DVB-T sticks RT820, so I can't test.....

andimik commented 3 years ago

The build in editor in Transedit is fine, as well. This is a shareware called Speed Commander, but any other should be OK, too.

Let me try with my dongle, as I did it for neighboring DLF on the same transponder in the past.

newspaperman commented 3 years ago

@andimik great catch! Unfortunately I think old BMux feed will disappear soon. The new GSE stream has a better spectral efficiency. Fortunately yours TBS 6903x, Cine S2 V7 and MAX SX8 is able to receive the GSE correctly. It's done by reencapsulation into a pseudo TS generated by the ECP3 media bridge: https://rickcaylor.websitetoolbox.com/post/show_single_post?pid=1312613794&postcount=30&forum=106995

TBS and DigitalDevices cards however differ: TBS only reencapsulates the GSE packets DigitalDevices reencapsulates the whole BBFrames.

However there's still one piece of software missing that extracts the content of the IPv4 packets out of the GSE packets.

piratfm commented 3 years ago

Actually, there is libgse that can decapsulate network packets from GSE, but I found only testing tool as reference of how to do that, and it uses .pcap files, not a raw BBframes: https://forge.net4sat.org/opensand/libgse/blob/master/test/test_encap_deencap.c

jpuigs commented 3 years ago

My TBS5927 shows 2 Muti streams, but none of them has output stream ¿¿¿¿???? I think 5927 is not suitable......

andimik commented 3 years ago

Forget 5927, this card cannot handle that at all. I am not sure if the Multistream here is correct or not.

jpuigs commented 3 years ago

As my 5927 can't handle that........... which TBS device (USB) would work ?

andimik commented 3 years ago

Well, don't buy it, you will be very angry like many others according to the TBS forum. It's a massive step backwards for a feed hunter.

Believe me, I have it as well.

As long there is no better driver and no software to handle this kind of streams you only can use notepad.exe or your favorite hex editor to view the raw input.

And: There is no USB solution yet. I had to buy a desktop PC.

jpuigs commented 3 years ago

Thanks, I didn't plan to buy anything... just asked to know if any USB would see something. The only TBS that lets you "see" raw input are 6903-x and 6909-x pcie cards ?

andimik commented 3 years ago

Yes, newspaperman also mentioned two other cards.

With 5927 you might see some bytes. But not a constant stream.

k-otixxx commented 3 years ago

Do you have signal on 7.0ºE, 12572V -Bavaria mux ?

jpuigs commented 3 years ago

You have no signal because symbol rate has been increased. Frequency is now 12569 V., and symbol rate is 14746, Other parameters remain unchanged, like FEC 2/3 in QPSK/DVB-S2 using PID 101.

Now, there are 4 more DAB multiplexes inside !!!! 239.229.96.38 port 50000 NDR NDS HAN (Niedersachsen, Hannover 7D) 239.229.96.42 port 50000 NDR NDS BS (Niedersachsen, Braunschweig 11B) 239.229.96.43 port 50000 NDR MV SN (Mecklenburg-Vorpommern, Schwerin 12B) 239.16.242.14 port 60014 Oberpfalz (6C)

image

k-otixxx commented 3 years ago

Thank you!

andimik commented 3 years ago

Many thanks. When I saw your question, I knew that the SR could have changed, but thanks for finding that out!

The funny thing is that for NDR the direction is different, see screenshot. And the traffic is twice than BR and SWR ...

grafik

I checked the content of the FF02 address, but it only contains internal data within NDR (I guess a fall back address in their intranet). No other DAB ensemble.

for example

Aruba JL319A 2930M-24G Switch dslforum.org

and

HHROTDAB001.hh.ads.ndr-net.de

jpuigs commented 3 years ago

There are 2,5 Mb/s of NULL packets, so we could see more additions in the near future....... ??

image

andimik commented 3 years ago

The transponder was never full in the past.

jpuigs commented 3 years ago

Yes, of course, but I don't remember such empty space inside in the past.

newspaperman commented 3 years ago

I'm getting no more lock on 12668v bundesmux. Did they change something?

andimik commented 3 years ago

Yes, they changed the frequency to 12641V 1500 S2 FEC Auto and switched off the old Bundesmux stream.

andimik commented 3 years ago

Assuming we get a valid stream from a TBS-6903X, would it be possible to modify fedi2eti? The (virtual) PID which is faked by the card is 280.

And I found https://rickcaylor.websitetoolbox.com/post?id=10621284&trail=45

I've updated the Readme, btw.

newspaperman commented 3 years ago

Thanks for updating. Unfortunately I have no blindscan with my Cine S2 V7. I wrote some software to extract the BBframes out of the Pseudo TS. Then I wrote a second tool that will extract the UDP payload out of the GSE frames. The good news is: I can really decode the nrk muxes and also the bundesmux 1+2 now. The bad news: It's really early alpha prototype and this is a hobbiest project only and my time is limited. There is still some work to do. I need to create 2 tools: One for TBS and another one for DigitalDevices. The common parts need to be put into common functions. However I don't have the complete bbframes on TBS cards, so I need to implement the gse processing a little bit different.

What's also missing: An analyzing function to see which addresses are active.

andimik commented 3 years ago

Wow, interesting!

BTW, as the symbol rate is not so high, DVB via Software (SDR) would be a solution.

https://github.com/pabr/leansdr/issues/16

andimik commented 3 years ago

So you can confirm it's EDI?

Can you tell us how you found out the right UDP addresses?

Do you have to use the Cine card under Linux or did you stream it from a Windows PC?

newspaperman commented 3 years ago

Ok right, then a 3rd version is needed as well.

Inside the gse frames there is a ipv4 header with udp payload. This payload can be forwarded to the EDI decoder. The src and dst ip addresses are in the ipv4 header.

I'm running linux only.

I also stumbled over a tool called gsextractor. After little modification it should be possible to analyze our stuff in wireshark.

andimik commented 3 years ago

Well, I already tested it and changed 0x4200 to 0x0192 for the modcode. But no luck with mine TBS and I also tested the NRK recording you made some months ago.

newspaperman commented 3 years ago

@andimik: After some investigation I noticed that your bundesmux capture is corrupt :( The problem is that it's gse with MIS. MIS 1 -> bundesmux1 and MIS2 -> bundesmux2. ATM both MIS are captured parallel and this confuses the GSE encapsulator. I get incomplete and mixed GSE frames. Is there any way to set MIS by hand before recording the stream?

You are doing your records on Windows, right? If there's no way we have to ask crazycat, if he can add this feature for us.

If you have time, can you please make a new recording of the nrk mux with your tbs6093x for me? There's only 1 MIS active and I want to test, if my decapsulator would work with it.

andimik commented 3 years ago

Oh, that was my fear, too, see above.

Anyway, I've recorded NRK on 0.8 West with TBS 6903-X.

https://we.tl/t-qV3BMyDf8U (file online for 7 days)

The beam is Nordic and I'm located close to Italy and Slovenia, EBSPro said 20% quality only. Could try with a larger dish on the weekend, if necessary.

So I don't know if this recording is error-free because - for example - the ensemble labels are mixed with others in a string:

$ strings -6 /tmp/0.8W_10719\,874_V_5400_\(2020-11-12\ 18.30.14\)_dump.ts | sort | uniq | grep "NRK Reg"
NRK Reg1
NRK Reg1 OsloVGA
NRK Reg1 OsloVik
NRK Reg2 VeTeVGA
NRK Reg2 VeTeVik
NRK Reg3 S
NRK Reg4 Vest   
NRK Reg4 Vest GA
NRK Reg5 InnlaGA
NRK Reg5 Innland
NRK Reg6 Tr
NRK Reg7 NoTrFGA
NRK Reg7 NoTrFi 

Now let's come to the bad news.

There is no chance to select the MIS for Bundesmux transponder in TBS6903x (see screenshot). Have tested it in EBSPro and in Crazyscan. Both tools say it's a single input stream (see screenshot). The card is called 'professional', but it isn't in my opinion (at least under Windows). It's very slow and randomly shows fake frequencies. And you need 1 or 2 dB more compared to TBS5927 ...

Only with a TBS5927 I've seen that there is a MIS 1 and 2 (but this card cannot handle the GS). I have no idea if this a driver issue in 6903x or a hardware restriction or a bug in the latest streamreader (by Crazycat).

Depending on the TBS card it says encapsulated, generic or continuous stream. And I fear the MIS is 1 or even 2 (depending on the card).

Screenshot (1)

newspaperman commented 3 years ago

@andimik: thank you very much for the capture from nrk, however there are still dropouts in it. If you have time it would be really good to make the capture with a bigger dish again

newspaperman commented 3 years ago

Are there any parameters like these you could set in the ini file you could set before starting crazyscan? FORCE_CONTINUOUS=1 TSOUT_NOSYNC=1

Remove the leading ;

andimik commented 3 years ago

The 6903X does not lock correctly on Bundesmux or NRK mux. I've opened an issue at Crazyscan's page. https://sourceforge.net/p/crazyscan/discussion/general/thread/b7b081b676/

I've tried with another, larger dish (I had to adjust it to 0.8W by hand), but I fear this is not a reception issue. I get 11dB even with 85cm and quality=100 and BER=0 with TBS5927, but - on the other hand - just a quality=20 with 6903X and a much higher BER.

For recording the files I've used TBS Recorder (2 different menu settings: Output as GS or TS, see screenshot) and Crazyscan.

Does anyone of the files help? https://we.tl/t-tlNRpv9nSy (files only online for 7 days)

nrk_tbsrecorder

The parameters FORCE_CONTINUOUS and TSOUT_NOSYNC seem to be linux-only (found it in https://github.com/mirror-opensource/linux/blob/8837c70d531a1788f975c366c254a5cb973a5291/drivers/media/dvb-frontends/stv0910.c for instance), it makes no difference.

jpuigs commented 3 years ago

I've tried with another, larger dish (I had to adjust it to 0.8W by hand), but I fear this is not a reception issue. I get 11dB even with 85cm and quality=100 and BER=0 with TBS5927, but - on the other hand - just a quality=20 with 6903X and a much higher BER.

If I "try" TBS5927 with EBSPro, signal and MIS streams 1 and 2 appear and dissapear continously, sometimes show only 1 stream, then 0, then 2...... and it's no signal problem.

I could try an old TBS5925 , which was "faulty", it worked fine after being plugged for more than 20 min. Is is better 5925 than 5927 ? Or better not to lose time with those.. ?